Content uploaded by Anders Wiik
Author content
All content in this area was uploaded by Anders Wiik on Mar 27, 2025
Content may be subject to copyright.
Data visualizations in journalistic
media studied from the perspectives
of visual-numeric literacy, everyday
mathematics and mathematization as
a social process
Anders Eickstedt Wiik
Doctoral Dissertations at
the University of Agder 525
Data visualizations in journalistic media
studied from the perspectives of
visual-numeric literacy,
everyday mathematics and
mathematization as a social process
Anders Eickstedt Wiik
Data visualizations in journalistic media
studied from the perspectives of
visual-numeric literacy,
everyday mathematics and
mathematization as a social process
Dissertation for the degree philosophiae doctor
University of Agder
Faculty of Engineering and Science
2025
Doctoral dissertations at the University of Agder 525
ISSN: 1504-9272
ISBN: 978-82-8427-243-6
© Anders Eickstedt Wiik, 2025
Print: MAKE!Graphics
Kristiansand
v
Foreword
This dissertation concludes a team effort. It would not have been realized without
the support and help of many people. I am grateful to Pauline Vos and Martin
Engebretsen for your endless support, patience and curiosity. You have found
order in my mess and added motivation to my frustration. More colleagues than I
can mention here have contributed with discussions, enquiries, encouragements,
inspiration, insights and ideas. Peers at conferences and seminars have discussed
the research and offered new perspectives. Not to mention the endless archives of
researchers’ works that I have tapped into! Last, but not least, my family has
supported me throughout this endeavor. Thank you!
vi
vii
Abstract
Data visualizations (DVs) are visual representations of quantitative data, which
are used to convey information. The aim of my PhD research was to better
understand the implications for readers of the use of DVs in journalistic media.
Journalistic DVs were explored from the perspectives of
(1) Visual-numeric literacy (VNL), which describes the capabilities that DVs
demand from readers,
(2) Everyday mathematics, which is the mathematics that people engage with
in various life situations (school, work, domestic life, etc.), and
(3) Mathematization as a social process, the tendency of some human
practices to become increasingly quantitative and mathematical.
In this PhD, the main theoretical perspective was social semiotics, but it was also
informed by a sociological perspective of late modernity. The empirical base for
the studies in this PhD was textual analysis of newspaper weather forecasts
(NWFs) in the period 1945-2020 and journalistic COVID-19 DVs, and an
analysis of interviews with young adults on their sense making of COVID-19
DVs.
The analysis revealed that NWFs shifted over time from verbally ‘telling’
readers about the weather, to offering abundant information in tables and maps
that the readers must organize and interpret themselves. The senders’ voice
changed from being a conversationalist or scientist to a blend of an advertiser and
a scientist. These changes relate to processes of mathematization in meteorology
and journalism.
The analysis of journalistic COVID-19 DVs showed that the DVs convey
much information (how many, where, how it changes, etc.) through numerous
formats (maps, line graphs, etc.), complex sign systems (coordinates, relative
numbers, color codes, etc.) and flexible use of conventions (e.g., missing vertical
axis). Readers were expected to make sense of these DVs and interpret their
significance and implications. Cues about data sources, data handling methods
and errors invited readers to reflect on the trustworthiness of the data and their
visualization. The interviews showed that adults have unequal opportunities for
making sense of DVs. It was observed that the three aspects of VNL, decoding,
acting (e.g. toggling in a DV, using a DV for making decisions) and reflection
were mutually supporting one another, and a readers’ background knowledge
viii
about the situation (i.e., COVID-19) supported the understanding of the sign
system.
Regarding VNL, everyday mathematics, mathematization as a social process
and the connection between these perspectives my research offers evidence that
the use of DVs in journalistic media has increased over time, that they mediate
information from experts (meteorologists, epidemiologists) to lay people, that the
VNL required of readers is quite sophisticated, and that journalistic DVs have
changed everyday mathematics. The changes do not consist of more or less
mathematics, but of an increased variety of quantitative information presented in
visual, flexible and informal systems. A sociological synthesis relates the
complexities of reading DVs to mathematization as a social process. For
example, globalization and reembedding (of journalism, DVs, meteorology,
epidemiology, data collections, mathematical models, VNL, etc.) enable readers
of DVs to access more and more diverse information yet creates obstacles for
intimacy and trust through the increased opacity of underlying data collections
and mathematical models. Insight into these mathematical processes is necessary
for reflecting critically on DVs. Mathematics education can play a key role in
helping students to develop their VNL and pave the way for participating in
society, and lifelong learning.
ix
Sammendrag
Datavisualiseringer er visuelle representasjoner av kvantitative data og brukes til
å formidle informasjon. Målet med denne PhD-studien var å bedre forstå hva
bruken av datavisualiseringer i journalistiske medier innebærer for lesere.
Journalistiske datavisualiseringer ble utforsket fra følgende perspektiver:
(1) Visuell-numerisk literacy, som beskriver den kapasiteten som
datavisualiseringer krever av leseren,
(2) Hverdagsmatematikk, som er den matematikken mennesker omgås med i
ulike livssituasjoner som skole, arbeid og i hjemmet, og
(3) Matematisering som en sosial prosess, tendensen i noen menneskelige
praksiser til å bli stadig mer kvantitative og matematiske.
Hovedteorien i denne avhandlingen er sosialsemiotikk. I tillegg har jeg brukt et
sosiologisk rammeverk på senmodernitet. Det empiriske grunnlaget for studiene
som presenteres i denne avhandlingen er tekstanalyser av værmeldinger fra
papiraviser mellom 1945 og 2020, tekstanalyse av journalistiske
datavisualiseringer relatert til COVID-19, og analyse av intervjuer med unge
voksne om hvordan de skaper mening når de leser journalistiske
datavisualiseringer relatert til COVID-19.
Analysene har vist at værmeldingene har endret seg over tid, fra å verbalt
‘fortelle’ leserne om været, til å tilby store mengder informasjon gjennom
tabeller og kart som leserne selv må organisere og tolke. Stilen i
avsenderstemmen i værmeldingene endret seg fra å være en uformell
samtalepartner som forteller om været, eller en vitenskapsperson som konsist og
vitenskapelig beskriver værvarselet, til å bli en blanding av en vitenskapelig og
kommersiell fortellerstil. Disse endringene kan relateres til
matematiseringsprosesser i meteorologi og journalistikk.
Analysen av journalistiske datavisualiseringer relatert til COVID-19 viste at
disse datavisualiseringene formidler mye informasjon (om antall, lokasjon,
endringer, osv.) gjennom mange formater (kart, linjegrafer, osv.) og fleksibel
bruk av konvensjoner (f.eks. manglende vertikal akse). Det var forventet at
leserne kunne gi mening til disse, forstå hvilken betydning de har og hva de
innebærer. Leserne fikk kortfattet informasjon om datakilder, metoder for
datahåndtering og feilretting, og dette inviterte leserne til å reflektere over
troverdigheten til dataene og måten dataene ble visualisert på. Intervjuene viste at
voksne har ulike muligheter for å forstå datavisualiseringer. Det ble observert at
x
de tre aspektene i visuell-numerisk literacy, avkoding, handling (f.eks. å veksle
mellom ulike formater og å bruke datavisualiseringer til å ta avgjørelser) og
refleksjon, var gjensidig støttende. Leserens bakgrunnskunnskap om situasjonen
(altså COVID-19) støttet forståelsen av tegnsystemet.
Angående sammenhengen mellom visuell-numerisk literacy,
hverdagsmatematikk og matematisering som en sosial prosess har forskningen
min gitt nyt funn. Forskningen har vist at bruken av datavisualiseringer i
journalistikk har økt over tid, at de medierer informasjon fra eksperter
(meteorologer, epidemiologer) til ikke-eksperter. Videre har forskningen vist at
den visuell-numeriske literacyen som forventes av lesere er ganske avansert, og
at journalistiske datavisualiseringer har endret hverdagsmatematikken.
Endringene består ikke av mer eller mindre matematikk, men en økt mengde
kvantitativ informasjon som presenteres i visuelle, fleksible og uformelle
systemer. En sosiologisk syntese relaterer funnene til matematisering som en
sosial prosess. For eksempel, globalisering og gjeninsetting (av journalistikk,
datavisualiseringer, meteorologi, epidemiologi, datasamlinger, matematiske
modeller, visuell-numerisk literacy, osv.) gir lesere av datavisualiseringer tilgang
på mer og mer variert informasjon, men skaper på samme tid nye utfordringer for
nærhet og tillit. Dette skjer blant annet fordi underliggende datasamlinger og
matematiske modeller ofte er utilgjengelige. Forståelse for slike matematiske
strukturer er nødvendig for å reflektere kritisk over datavisualiseringer og deres
rolle i samfunnet. Matematikkutdanning kan spille en nøkkelrolle for å hjelpe
elever og studenter til å utvikle visuell-numerisk literacy, og legge til rette for
samfunnsdeltakelse og livslang læring.
xi
Contents
Foreword ................................................................................................................. v
Abstract ................................................................................................................. vii
Sammendrag .......................................................................................................... ix
List of abbreviations ............................................................................................ xiv
1 Introduction ......................................................................................................... 1
1.1 Background ................................................................................................... 1
1.2 Rationale and aims ........................................................................................ 3
1.2.1 Empirical aims ....................................................................................... 3
1.2.2 Theoretical aims ..................................................................................... 5
1.3 Overview of the dissertation ......................................................................... 6
2 Literature review.................................................................................................. 7
2.1 Data and DVs in journalistic settings ........................................................... 7
2.2 Literacy regarding DVs ................................................................................ 9
2.3 Everyday mathematics ................................................................................ 12
2.3.1 Conceptual clarification ....................................................................... 12
2.3.2 Research findings regarding everyday mathematics ............................ 15
2.4 Mathematization as a social process and its implications .......................... 19
2.4.1 Overview of the literature on mathematization ................................... 19
2.4.2 The mathematization of meteorology and weather forecasting ........... 24
2.4.3 The mathematization of epidemiology ................................................ 25
2.5 Research questions ...................................................................................... 26
3 Theoretical perspectives .................................................................................... 31
3.1 Social semiotics .......................................................................................... 31
3.2 Supporting theories ..................................................................................... 34
3.2.1 Hasan’s theory of literacy and Tønnessen’s VNL ............................... 34
3.2.2 Giddens’ sociology............................................................................... 38
3.3 A reflection on combining theories ............................................................ 42
4 Methodology ...................................................................................................... 45
4.1 Research paradigm ...................................................................................... 45
4.2 Study A: Newspaper weather forecasts ...................................................... 47
4.2.1 Data collection for Study A ................................................................. 47
4.2.2 Data analysis reported in Paper I ......................................................... 49
4.2.3 Data analysis reported in Paper II ........................................................ 53
4.2.4 Ethical and legal issues ........................................................................ 56
4.3 Study B: COVID-19 DVs ........................................................................... 57
xii
4.3.1 Data collection used for Paper III ........................................................ 58
4.3.2 Data analysis reported in Paper III ....................................................... 58
4.3.3 Data collection used for Paper IV ........................................................ 62
4.3.4 Data analysis reported in Paper IV ...................................................... 64
4.3.5 Ethical and legal issues ........................................................................ 67
5 Extended abstracts of papers ............................................................................. 69
5.1 Paper I: Trends in everyday mathematics: The case of newspaper weather
forecasts ............................................................................................................ 69
5.2 Paper II: From scientist to friend to advertiser: Norwegian newspaper
weathercasters’ identities, roles, and reader relations 1945–2020 ................... 70
5.3 Paper III: Visual-numeric literacy: The case of COVID-19 DVs for news
media and their expectations of readers ............................................................ 73
5.4 Paper IV: Making sense of journalistic COVID-19 DVs: An in-depth study
of two adults’ VNL ........................................................................................... 75
6 Discussion .......................................................................................................... 79
6.1 Discussion of results in light of the concept of mathematical literacy ....... 79
6.2 Discussion of results in light of the concept of everyday mathematics ...... 84
6.3 Main findings discussed in light of Giddens’ framework of late modernity
........................................................................................................................... 87
6.3.1 Displacement and reembedding ........................................................... 88
6.3.2 Intimacy and impersonality .................................................................. 89
6.3.3 Expertise and reappropriation .............................................................. 90
6.3.4 Privatism and engagement ................................................................... 94
6.3.5 Reflections on using Giddens’ framework to research mathematization
as a social process ......................................................................................... 95
6.4 Discussion of the results in light of the concept of mathematization as a
social process .................................................................................................... 96
6.4.1 Mathematization in the context of displacement and reembedding..... 96
6.4.2 Mathematization in the context of intimacy and impersonality........... 98
6.4.3 Mathematization in the context of expertise and reappropriation ....... 99
6.4.4 Mathematization in the context of privatism and engagement .......... 101
6.4.5 Mathematization and demathematization .......................................... 102
6.5 The connection between mathematization as a social process, everyday
mathematics and mathematical literacy .......................................................... 105
7 Conclusions and recommendations ................................................................. 109
7.1 Empirical conclusions and implications thereof ....................................... 112
xiii
7.2 Theoretical conclusions, implications and reflections .............................. 114
7.2.1 Theoretical conclusions, implications and reflections regarding the
VNL frameworks ........................................................................................ 114
7.2.2 Theoretical conclusions, implications and reflections regarding social
semiotics ...................................................................................................... 116
7.2.3 Theoretical conclusions, implications and reflections regarding
mathematization as a social process ............................................................ 116
7.3 Strengths, limitations and reflections on methodology ............................ 118
7.4 Recommendations ..................................................................................... 120
List of references ................................................................................................ 123
Appendix ............................................................................................................ 133
xiv
List of abbreviations
Abbreviation
Definition
DV
Data Visualization. A visual
representation of data. Examples
include line graphs, bar charts and
choropleth maps.
NWF
Newspaper Weather Forecast. The
weather forecasts that appear in
newspapers.
VG
Verdens Gang (the way of the world),
the most read Norwegian newspaper.
VNL
Visual-Numeric Literacy. The special
form of mathematical literacy needed
to make sense of DVs.
1
1 Introduction
In this introduction, I will first outline the background issues that motivated this
research, illustrated with a recent example from the COVID-19 pandemic
(Section 1.1). Then, in Section 1.2 I state the rationale and aims of the research
and in Section 1.3 I outline the structure of this dissertation.
1.1 Background
Data visualizations (DVs), that is, visual representations of quantitative data, are
central in this dissertation. They are used for conveying quantitative data with
social, political and economic relevance, and have been used for several centuries
(Tufte, 2001). Recently, the use of DVs has grown and they have come to play an
increasingly important role in society (Engebretsen & Kennedy, 2020). The
significance of DVs in contemporary society was evident during the COVID-19
pandemic, when mainstream media channels regularly published DVs of
statistics such as number of infected and deaths. To stay informed about the state
of the pandemic, knowing how to read DVs such as cartesian graphs and maps
was essential (Aguilar & Castaneda, 2021). For example, choropleth maps such
as Figure 1 were instrumental in describing the severity of the pandemic across
the globe. Such DVs demand that readers can make sense of maps and color
codes that translate death rates to colors (e.g., darker red = higher death rates).
The increasing use and importance of DVs in journalistic media reflects societal
processes of mathematization, that is, processes in which phenomena are
increasingly understood and described in mathematical and quantitative terms
(Jablonka & Gellert, 2007). Processes of mathematization have far-reaching
implications for society and peoples’ lives (e.g., Jablonka & Gellert, 2007;
O’Neil, 2016).
During the COVID-19 pandemic, DVs were used for multiple purposes. For
example, they functioned to justify policies (Jablonka & Bergsten, 2021) and
create trust in mitigation measures such as social distancing and vaccinations (Li
& Molder, 2021; Riggs et al., 2022). It was also observed that some of the
COVID-19 DVs appearing in newspapers contained errors and misleading
designs (Kwon et al., 2021) and that the DVs relied on mathematics at the higher
end of the curriculum or beyond mandatory education (Aguilar & Castaneda,
2021; Kwon et al., 2021). Hence, we know that DVs were used by news media to
2
Figure 1. Choropleth map of the world, showing the number of COVID-19 related deaths per
100 000 inhabitants as of April 21st, 2020. Retrieved from vg.no. Reprinted with permission.
convey crucial information with important socio-political implications, and there
are indications that many of these DVs posed significant challenges to readers.
However, there are still gaps in our knowledge concerning implications of the
growing use of DVs in news media (Heyd-Metzuyanim et al., 2021; Kwon et al.,
2021). Central to the implications of the growing use of DVs is the demands they
put on readers and the role of schools in preparing students for such
mathematical demands in everyday life. Therefore, a core and recurring concern
in this dissertation is the non-trivial connection between processes of
mathematization and the mathematical literacy expected of citizens.
DVs are complex communication artefacts. To make them, it is necessary to
compile, structure and model data, decide which aspects of the data should be
visualized, and choose a visualization form that suits the intended audience and
the intended message (Allen, 2018; Kirk, 2019). Furthermore, designers can use
colors, arrows and other effects to accentuate certain aspects and downplay
others (Kirk, 2019). Thus, multiple actors play a role in making DVs, including
but not limited to data collectors, modelers, designers, distributors and the DVs
3
themselves (Allen, 2018). Once the DV is published, the readers’ role is to make
sense of it. Sensemaking involves decoding the DV – what information does it
contain? – and contextualize this information – what does it mean, and for whom
(Engebretsen, 2020)? Further, several researchers argue that it is important that
citizens develop their critical and reflective literacy to question the
trustworthiness of the data and their representation, and the validity of the
implications that are drawn from the DVs (Gal & Geiger, 2022; Geiger et al.,
2023; Gray et al., 2018; Jablonka & Bergsten, 2021; Rubel et al., 2021;
Tønnessen, 2020). Citizens’ capacity to decode, contextualize and critically
reflect on DVs, such as Figure 1, is central in this dissertation.
1.2 Rationale and aims
The rationale for this dissertation is to give empirical and theoretical
contributions to research on the reading and sense-making of journalistic DVs,
with a focus on their properties and implications as mathematical objects in
citizens’ everyday life. An overarching aim was to explore connections between
mathematical literacy, everyday mathematics and mathematization as a social
process. I decided to focus on journalistic DVs because journalism is essential to
democracy and reaches out to large and heterogenous audiences.
1.2.1 Empirical aims
This dissertation presents empirical findings from two studies, described in four
papers, whose data differ in their place in time and their socio-political relevance.
Therefore, the empirical aims of each study differ.
First, I aimed to study journalistic DVs with a close connection to a scientific
practice that has become increasingly mathematized. This way, potential
implications of the mathematization of the underlying sciences for the way the
data are presented in newspapers can be studied, in tandem with the
characteristics of the DVs themselves. For this aim, I decided to study weather
forecasts in Norwegian newspapers because the underlying science of weather
forecasting has become mathematized (Bauer et al., 2015; Harper, 2008;
Kristiansen, 2017). Further, newspaper weather forecasts (NWFs) have been a
regular feature in newspapers for a long time, which means that they enable
studying long-term trends in the design and use of journalistic DVs. Reading
NWFs is a mundane everyday activity undertaken by many citizens around the
world and is easily taken for granted. To study NWFs and capture historical
4
developments, it was necessary to obtain a manageable corpus. Therefore, I
sampled from the most read Norwegian newspapers in the period that captures
the mathematization of weather forecasting from 1945 up to the present. NWFs
are the focus of Study A, which is reported in Paper I and Paper II (see Figure 2).
Next, I wanted to study the demands and opportunities of journalistic DVs as
they appear in current practices. For this part of the project, the COVID-19
pandemic yielded a golden opportunity. Many news outlets across the world used
DVs when presenting news about the pandemic, and some outlets even
developed specialized online resource pages in which they gathered relevant DVs
that were updated regularly. A collection of such DVs offers an opportunity to
explore the literacy that is expected of lay readers of journalistic DVs and the
opportunities and challenges that readers encounter during such reading
experiences. I opted to use the online resource page developed by Verdens Gang
(VG), the most read Norwegian newspaper (Mediekatalogen, 2022). Study B
centers on this web page, and Figure 1 is an example of a DV from this web
page. Study B consists of an analysis of this web page, which is reported in Paper
III, and interviews with young adults on their experience from reading this web
page, which is reported in Paper IV (see Figure 2). Epidemiology, the science of
infectious diseases, has undergone a mathematization (Davey Smith, 2019),
which means that Study B too offers opportunities to study implications of
processes of mathematization.
Figure 2: Overview of the studies reported in the respective papers in this dissertation.
Dissertation
Study A:
Newspaper Weather
Forecasts
Paper I
DVs historical development
regarding their demands on
readers
Paper II
How scientists' 'speaks' in
newspapers
Study B:
COVID-19 DVs
Paper III
DVs' demands on readers Paper IV
Two adults reading DVs
5
Due to the differences between Study A and B, they can give complementary
and contrasting insights on the roles that journalistic DVs play in terms of
mathematical literacy, everyday mathematics and mathematization as a social
process.
1.2.2 Theoretical aims
In the process of conducting these studies, there were underlying theoretical
issues. The growing use and importance of DVs is part of a larger societal trend,
in which various phenomena are increasingly described and understood in
mathematical terms. This trend is captured by the term mathematization as a
social process (Jablonka & Gellert, 2007). Processes of mathematization have
complex implications for society, citizens’ everyday life and education, and
literacy. The studies described in this dissertation offer new empirical
perspectives on mathematical literacy, everyday mathematics and
mathematization as a social process, which assist to further develop the
theoretical understanding of these phenomena. In particular, I aimed to develop a
conceptualization of mathematical literacy that focuses on the particular
opportunities and challenges of journalistic DVs. Further, to explore journalistic
DVs as a form of everyday mathematics, it was necessary to conceptualize ‘the
everyday’. Lastly, to explore journalistic DVs as a form of mathematization as a
social process and its implications it was necessary to find and adapt a suited
theoretical framework and connect this framework with mathematical literacy
and everyday mathematics. Paper III and IV address mathematical literacy; Paper
I, III and IV address everyday mathematics; and all papers, directly or indirectly,
address mathematization as a social process. In the discussion section of the
dissertation, the findings from all papers are discussed in relation to these three
main perspectives.
The research presented in this dissertation occurred at the intersection of
mathematics education and media studies with a special focus on journalistic
DVs. Therefore, this project bridges discussions in these domains regarding the
roles and implications of journalistic DVs in everyday life. The findings can be
of interest to teachers and researchers in mathematics education; practitioners
and researchers interested in DVs; and anyone interested in the sociological
dimensions of mathematics in society.
6
1.3 Overview of the dissertation
This dissertation is structured as follows:
In Chapter 1 the background and rationales for this PhD are outlined, as well
as a brief explanation of the studies and aims.
In Chapter 2 I present reviews of relevant literature. The literature review is
divided into four sections. There is first a review of the literature on data and
DVs, in which some key studies on this topic are presented. Next, I present a
review of literature pertaining to the literacy of reading DVs. In the third
subsection, I review literature on the mathematization of society and its
implications. This includes conceptual and empirical papers from mathematics
education and other relevant fields. Finally, I present a review of the literature on
everyday mathematics that focuses on the mathematical skills needed and
developed in everyday life. Chapter 2 concludes with the research questions.
Chapter 3 is devoted to the theoretical perspectives used in this research. The
main theoretical perspective, social semiotics, is presented first. Then, I have
sections presenting supporting theories and a discussion of potential issues
associated with combining theories.
In Chapter 4, containing methodological and ethical reflections, the studies
are positioned in terms of research paradigm and epistemological and ontological
assumptions. Further, the methods used are outlined and justified, focusing on
the choices that are not discussed in the articles. Finally, relevant ethical issues
are discussed.
In Chapter 5 extended abstracts for the four research papers produced during
this PhD are presented.
Chapter 6 presents a discussion of the main findings from all four papers in
light of relevant literature. The discussion is structured according to the three
perspectives of mathematical literacy, everyday mathematics and
mathematization as a social process.
This is followed by Chapter 7, in which I present conclusions, reflections on
the use of theory and research designs, strengths and limitations of this research
and recommendations for practice and further research.
7
2 Literature review
In this literature review, I aim to present the state of the research relevant to this
dissertation. The main themes in this literature review will be revisited in the
discussion chapter of the dissertation (Chapter 6). Owing to the interdisciplinarity
of this research, several strands of literature are relevant. I have organized the
literature review in four main subchapters. In Section 2.1 I review the literature
that specializes on the use of quantitative data and data visualizations (DVs) in
journalism. Next, in Section 2.2, I take a closer look at the particular forms of
literacy that are associated with reading DVs. Then, in Section 2.3, I start by
clarifying the meaning of this term and thereafter present a review of relevant
literature. In Section 2.4, I present research on mathematization as a social
process, with an emphasis on the implications of mathematization processes. In
2.5, the research questions for this dissertation are stated in relation to gaps in the
literature.
2.1 Data and DVs in journalistic settings
In this section, I will review key literature about data and DVs in journalism.
First, I define DVs and situate them in contemporary society. Then, I present
empirical and theoretical research about how data and DVs come into being, how
and to what extent they reflect the phenomena they claim to describe, and their
social roles and significance. I will place particular emphasis on journalistic DVs.
My definition of DVs builds in the first place on the term data, which refers
to quantitative data, that is, numbers indicating an amount or magnitude of
something. Inspired by the definitions by Engebretsen and Weber (2017) and
Kirk (2019), I developed a definition of DVs that highlights their composition
and intended function: a DV is a representation of data using visual semiotic
resources such as lines, bars, dots and colors to present data in a way that is
accessible and socially applicable. This definition includes for instance graphs
and maps.
According to several studies, the use of quantitative data in journalism has
been in rapid growth and development in recent decades (Allen, 2018;
Coddington, 2015; Engebretsen & Kennedy, 2020; Engebretsen et al., 2018).
This means that journalism fits within the broader patterns of datafication and
mathematization of society, in which phenomena are increasingly quantified,
formatted in tables and graphs, and analyzed according to mathematical rules
(Jablonka & Gellert, 2007; Mayer-Schönberger & Cukier, 2013).
8
Research on journalistic DVs can roughly be divided into two categories. The
first category is the design processes ‘behind’ DVs, that is, the processes in
which a DV is planned, data is generated and processed, visualized and
integrated into news stories (Kirk, 2019). This category is the focus of this
section. The second category is the reading of DVs, which is the focus of Section
2.2.
Two early studies on mathematics in journalistic media found high
frequencies of misleading and erroneous mathematics in news stories, as well as
high levels of mathematics anxiety and variable numeracy skills among
journalists (Maier, 2002, 2003). More recent studies have focused specifically on
DVs and give a more detailed and nuanced picture. According to The Data
Journalism Handbook 1 (Gray et al., 2012), methods journalists use for obtaining
data include browsing in databases, asking for data or data sources in relevant
fora and mailing lists, contacting relevant experts and crowdsourcing. It can also
happen that the relevant data is not yet assembled, and the journalists must
compile a dataset themselves. Relevant skills for obtaining data include meta-
knowledge of relevant databases such as knowing its purposes, how it is
managed and who to contact, and how to effectively operate search engines
(Gray et al., 2012). Journalists working with data often show willingness to be
transparent about their data, data sources and methods, which gives readers
opportunities to check and verify the underlying data and the methods to obtain
and manage them (Weber et al., 2018). However, Lowrey et al. (2019) suggest
that data journalists are unlikely to scrutinize their data deeply and can be
uncritical of the categories that are developed from the data. In recent years,
newsrooms often have dedicated teams working with designing DVs, in which
labor is divided between people with different expertise such as “visual
communication, digital interaction, data handling and coding with journalistic
insights in news values and storytelling” (Engebretsen et al., 2018, p. 14).
Several authors have noted the crucial role of digital tools in contemporary
DV design (Engebretsen et al., 2018; van Geenen & Wieringa, 2020).
Engebretsen et al. (2018) positions digital tools as one of the key drivers of
development in design practices. Van Geenen and Wieringa (2020) highlight
how the affordances and features of the digital tools used for design purposes
shape both the design process and the result.
9
To my knowledge, there are no studies that document how journalistic DV
practices have changed over time. Also, I am not aware of studies documenting
how DV design and relevant scientific practices interact.
According to Engebretsen et al. (2018), some developers of journalistic DVs
take on an educational role by exposing their readers to novel DV forms, even
though challenging readers with unfamiliar formats may discourage some
readers. DVs, novel or not, pose unique challenges and opportunities for readers,
which is the focus of the next section.
2.2 Literacy regarding DVs
In this section, I present a review of literature on the literacy related to reading
DVs. Some of the literature presented here also appears in Paper III (Section
5.3). By ‘literacy’ I refer to the capacity to read, design, make sense of, use,
critically reflect upon or otherwise engage with DVs. Therefore, I will not
distinguish between related terms ‘numeracy’ and ‘mathematical literacy’.
Literacy is not regarded as something innate, but something that is learned and
developed through experience. I will first summarize review articles in this field.
Then I will review frameworks that have been developed to describe how people
read and design DVs and present some key empirical findings.
I found three review articles about DV reading (Friel et al., 2001; Glazer,
2011; Shah & Hoeffner, 2002). Friel et al. (2001) focused on the skills needed to
read DVs of statistical data and found that these skills have many aspects, for
example: recognizing the components of DVs (e.g., lines, coordinates and
colors); knowing how to read off information from a DV; find relationships
within the data; and understand how a DV relates to a table of the same data set.
Shah and Hoeffner (2002) focused their review on cognitive psychological
research. Among their findings is that contextual knowledge is very important in
making sense of DVs, and that DV reading skills learned in an abstract school
context may not be applicable to DV reading in out-of-school contexts.
In the review article by Glazer (2011), it is stressed that the reading of DVs is
a learned skill and “should be an explicit focus of instruction” (p. 201). Glazer
claims that most studies on DV reading have focused on DVs of abstract
mathematical functions, and that there is a need for further research on DV
reading in other domains. Although many studies on DV reading have been
published since the publication of Glazer’s article, it is still the case that more
10
research on reading DVs outside the mathematics classroom is needed
(Engebretsen, 2020).
Olande’s (2013) framework describes two qualitatively different reader
approaches to DVs, the identification approach and the critical-analytical
approach. In the former approach, the reader is focused on the perceived visual
elements of the DVs (e.g., bars, lines, colors) and familiar operations and forms
of expression (e.g., addition, cartesian coordinates). In contrast, in the second
approach, the reader is critical, reflective and analytical about the production and
design of the DV. According to Olande (2013), readers using the critical-
analytical approach can easily see beyond “misleading” DVs (e.g., DVs with
truncated axes) and make correct interpretations. Olande’s (2013) analyses of
interview data and selected PISA results showed that students in secondary
school generally struggle to employ the critical-analytical approach. However,
his study does not address what approaches students and adults would employ in
out-of-school contexts.
Börner et al. (2019) developed a cyclic framework describing a process in
which DVs are needed, designed, read and evaluated. This framework gives a
holistic perspective on processes in which DVs are read and designed. A
potential weakness with this framework is that it relies on a pre-defined
inventory of DV forms which limits the space for variation and innovation.
Vos and Frejd (2020) have a novel approach to analyzing processes in which
DVs are read and designed. Their analytical construct distinguishes between DVs
as objects, with a cultural and historical background and culturally agreed-upon
properties that can be learned; and DVs as tools that can be used to communicate
messages to others. Vos and Frejd’s analysis of a teaching session with grade 8
students showed that the students could successfully learn to use Sankey
diagrams as tools before they became aware of their mathematical properties.
Tønnessen (2020) also studied how students read DVs in school contexts. She
coined a concept, visual-numeric literacy (VNL), and developed a framework for
it. This framework is grounded in the larger theoretical perspectives of social
semiotics and socio-cultural learning theory. The framework consists of three
aspects, which are recognition literacy (to recognize semiotic resources such as
lines, bars and colors, and their intended meaning), action literacy (interpret DVs
in their social context and use them to reach personal goals) and reflection
literacy (being critical, analytical and reflective about DVs and their
sociopolitical role). I will revisit Tønnessen’s (2020) theoretical framework for
11
VNL in later chapters and use and further develop this framework. Regarding her
empirical results, Tønnessen (2020) observed that the students in her study
(Norwegian upper secondary school) had few struggles with the recognition
aspect, but needed guidance in the action aspect. The reflection aspect was barely
observed. She also found that the students’ VNL was aided by their digital media
literacy, which helped the students to effectively navigate the interactive features
of the DV and locate relevant information. Because her study was conducted in a
school context, it remains open how the results would differ if a similar study
was conducted in an out-of-school context.
Engebretsen (2020) explored how young adults read journalistic DVs in non-
school settings. He found that the informants generally interpreted line graphs
correctly and effortlessly, but that less common DV forms were more prone to
misinterpretation. Like Tønnessen (2020), he found that digital media literacy
habits aided the participants to navigate the DVs. Further, he found that
motivated reasoning, feelings such as responsibility and surprise, and prior
knowledge and engagement with the topic at hand were important aspects of
people’s interactions with journalistic DVs. Reimann et al. (2022) found that
more common DV types are more readily identified and interpreted. Kennedy
and Hill (2018) also explored how people interact with DVs in non-school
settings and found that emotions are “vital components of making sense of data”
(p. 830). Similarly, Didier (2024) found that emotions and reason work in
harmony when numbers were used to justify political actions during the COVID-
19 pandemic and acted as “a means for implicating emotions in the vast social
aggregate by correlating them to reason” (p. 18).
Yates et al. (2021) have identified six different profiles for citizens engaging
with data (not necessarily quantitative data) in digital media: extensive political
users; extensive users; social and media users; general users; limited users; and
non-users. However, their research revealed that none of these reader types
engage deeply with data in their personal and civic lives, and the participants’
understanding of digital data infrastructures – for example how Google and
Facebook use user data to personalize commercial content – was limited in all
groups.
Aguilar and Castaneda (2021), Gal and Geiger (2022), Kwon et al. (2021) and
Rubel et al. (2021) studied the expectations raised by journalistic DVs related to
the COVID-19 pandemic. They all conclude that these DVs raised high
expectations on readers, and that making sense of the DVs involved an
12
understanding of their significance in the social context of the pandemic. Hence,
the capacity to make sense of these DVs involves both a capacity for decoding
mathematical symbols (e.g., cartesian coordinate systems) and situating the
meaning of these symbols in their social context (i.e., the pandemic).
As the above examples show, frameworks used for analyzing processes of
reading or designing DVs highlight various aspects of such processes. A
common theme is that the design of the DV is important – how are semiotic
resources such as lines, bars, colors and so on used by designers to create
meaning in DVs, and how do readers make sense of them. Other aspects that are
highlighted are how readers use DVs to reach certain goals, and how readers can
be critical and analytical when reading DVs. While many of the studies in this
field are conducted in school settings, the studies on out-of-school engagements
with DVs have highlighted emotions and contextual knowledge as key aspects of
how people read DVs. Further research is needed on how people engage with
data and DVs in contemporary media. In particular, Yates et al. (2021) call for
in-depth studies on the role of data in everyday life to inform educational
initiatives and DV design practices.
2.3 Everyday mathematics
In this section, I will focus on everyday mathematics. The meaning attributed to
this term varies between different authors, and there are many related terms
appearing in the literature. Therefore, I will start by clarifying the meaning of this
concept before proceeding to a review of empirical research on this topic.
2.3.1 Conceptual clarification
A striking feature in the literature on everyday mathematics is the lack of
coherent and unified terminology. Terms with a related meaning include “real-
world mathematics” (Gainsburg, 2008), “out-of-school mathematics” (Wager,
2012), “workplace mathematics” (Williams & Wake, 2007), “street
mathematics” (Nunes et al., 1993), “numeracy” (Wedege, 1999), “health
numeracy” (Heilmann, 2020), “ethnomathematics” (d’Ambrósio, 2006),
“informal mathematics” (Bonotto, 2005) and, of course, “everyday mathematics”
(Carraher & Schliemann, 2002; Civil, 2002; Lave, 1988; Presmeg, 2007). This
list is not exhaustive but illustrates the multiplicity of terms related to the
practical use of mathematics. To start organizing the terminology and clarify the
meaning I attribute to ‘everyday mathematics’, I look to Wedege (2010) who
13
reviewed literature on “knowing mathematics in society” (p. 31). She makes
three distinctions: first, the terminology can to various degrees highlight
functionality or contextuality. For example, health numeracy highlights the
context of health and street mathematics highlights the context of life in the
streets, whereas numeracy and mathematical literacy are often conceptualized in
more general terms and describe functional mathematics with less emphasis on
specific contexts (e.g., OECD, 2019). Next, some terms describe mathematical
skills and knowledge that are developed in everyday life, whereas others describe
mathematical skills and knowledge that are wanted or expected in everyday life.
For example, street mathematics and ethnomathematics are geared towards the
mathematics practiced by people in the streets and in different cultures; and
OECD’s definition of mathematical literacy describes mathematical skills and
knowledge wanted in industrialized countries for developing human capital. The
final distinction is between capacity and performance. Mathematical capacity
describes what people can do mathematically in different contexts, while
mathematical performance relates the individuals’ mathematical competence to
predefined mathematical requirements. While ethnomathematics explicitly
embraces the capacity perspective (d’Ambrósio, 2006), concepts like
mathematical literacy and numeracy are often used to inform testing schemes and
therefore relate to performance (e.g., OECD, 2019).
These three distinctions assist in outlining the meaning of everyday
mathematics as used in this dissertation. First, everyday mathematics captures
function and context alike. It is about mathematical functionality in everyday
contexts. I will return to the meaning of ‘everyday’ in the next paragraph.
Second, it can be used both to describe the mathematics developed (e.g., the
mathematical skills and knowledge developed from reading COVID-19 DVs)
and the mathematics wanted in everyday life (e.g., the mathematical literacy
wanted for reading newspaper weather forecasts (NWFs)). Third, everyday
mathematics is about capacity and not performance: the usefulness of the concept
lies in describing the mathematics developed and wanted to manage everyday
situations.
Next, I will delineate the meaning of ‘the everyday’ as a context. Among the
authors who have used the term ‘everyday mathematics’, the scope of what is
included in “the everyday” varies. The first major publication using this term,
Lave (1988), defined the everyday thusly:
14
“Everyday” is not a time of day, a social role, nor a set of activities, particular social
occasions, or settings for activity. Instead, the everyday world is just that: what
people do in daily, weekly, monthly, ordinary cycles of activity. A schoolteacher
and pupils are engaged in “everyday activity” in the same sense as a person
shopping for groceries after work and a scientist in the laboratory. (Lave, 1988, p.
15)
Carraher and Schliemann (2002) have a slightly more restricted definition, “the
mathematics outside school settings” (p. 133). This definition includes the
mathematics of the workplace, the mathematics of grocery shopping and the
mathematics of reading the news, but not the mathematics of schools. Civil
(2002) suggests a threefold compartmentalization of mathematics into school
mathematics, mathematicians’ mathematics and everyday mathematics, each
with distinct characteristics. In her use of the term, everyday mathematics is “the
mathematics learning that occurs outside school” (p. 43) and, although not
explicitly stated, presumably does not include the mathematics of
mathematicians.
Because the research literature varies on how this concept is defined, I have
found it useful to consult a different source. The sociologist Lefebvre
(1947/1991) defines the everyday very much in the same way as Lave (1988). To
Lefebvre, the everyday is the space in which all of life occurs, and therefore
exists across and between institutions such as schools, work and politics. At the
center of the everyday are mundane routines activities such as waking up in the
morning, reading the news, commuting to work and watching television in the
evening; but it also contains transitions between contexts and life stages. This
definition of the everyday does not privilege schools or workplaces and avoids
the difficulties of separating school activities from other activities. As I see it,
this definition has two strengths: it puts the living, acting human in the center
without cutting them off the cultural and historical context in which life happens;
and it positions institutions as secondary. This underlines an important feature of
everyday mathematics: it is not so that everyday mathematics, school
mathematics and workplace mathematics are somehow subsets of one another.
While school mathematics and workplace mathematics are institutionalized
bodies of knowledge that belong to specific school systems and workplaces,
everyday mathematics is the mathematics of the human moving across and
between institutions through life. Everyday life is a site of ‘public pedagogy’ in
which people learn in all kinds of social practices and contexts, often without a
15
teacher present (Giroux, 2004). Thus, everyday mathematics is learned in all
kinds of ways, including but not limited to teaching, apprenticeship,
observations, practice and experience. When people engage in everyday
mathematical activities such as reading DVs as part of news stories, they may or
may not rely on mathematics learned in schools. The focus of my dissertation is
limited to a very particular arena of everyday mathematics, namely the
mathematics of reading the news. However, this sociologically oriented
definition of the everyday enables me to see the news and the citizen reading the
news as constituent parts of a much larger system.
Attending school is part of everyday life, and so people encounter and engage
with school mathematics as part of everyday mathematics. However,
mathematical activities in everyday situations are often not recognized as
‘mathematical’, perhaps because school mathematics and out-of-school
mathematics may not resemble one another (Wedege, 1999). Also, many people
experience a gap between school mathematics and workplace mathematics
(FitzSimons & Björklund Boistrup, 2017). Thus, different mathematical
everyday practices can be experienced as disconnected.
With these conceptual clarifications of the everyday and everyday
mathematics, I will move on to reviewing empirical research findings on
everyday mathematics.
2.3.2 Research findings regarding everyday mathematics
The early studies (roughly speaking, those published between 1967 and 2000) on
everyday mathematics would typically describe the mathematical practices of
particular groups of people in particular contextual and non-formal activities –
adults grocery shopping, children selling candy in the streets, carpenters at work,
indigenous peoples’ practices – and would typically have a critical discussion of
the transferability of mathematical skills between schools and out-of-school
activities (Greiffenhagen & Sharrock, 2008). Carraher and Schliemann (2002),
Civil (2002) and Presmeg (2007) reviewed this literature, documenting that
children or adults with little or no formal schooling can master arithmetic,
measurement, geometry and probability in out-of-school contexts. There is also
evidence that people who master mathematical problems in everyday situations
struggle to solve problems with a similar mathematical structure in school
contexts. Carraher and Schliemann (2002) suggest that the main reason for this
gap is that even though the problems have a similar mathematical structure and
16
are presented in a similar context, “[b]y indicating in the school-like setting that
they should write out their answers, we induced them to use computation routines
that involved features, such as borrowing from neighboring columns, that they
understood poorly and did not employ in their work as vendors” (p. 134). Thus,
the struggles may be born from the different expectations in the two
environments: the school(-like) context comes with an expectation that certain
procedures must be used, while those same procedures were not routinely applied
in out-of-school practices. Does this mean that schools are wrong in prompting
students to use certain methods when solving mathematical problems? Carraher
and Schliemann (2002) answer this by pointing out a tension between school
mathematics and out-of-school mathematics. While schooling educates students
for an unknown future that may include becoming mathematicians, engineers,
economists or any other endeavor involving mathematics to an unknown extent,
the practices of everyday mathematics tend to have a much shorter horizon.
Schools should introduce students to novel content with potential future
usefulness, and everyday mathematics practices are primarily aimed at solving
concrete, practical problems on the spot or in the near future. Presmeg (2007)
reaches a similar conclusion and adds that school incorporation of out-of-school
ideas runs the risk of positioning school mathematics as superior and trivializing
everyday mathematics. This leads her to formulate a challenge: how can out-of-
school mathematics be incorporated into school mathematics without trivializing
the ideas that the students bring from out-of-school experiences?
Civil (2002) outlines four characteristics of everyday mathematics learning:
• It mainly occurs by apprenticeship.
• It involves contextual problems.
• The person working on the task is in control over the task and choice of
strategies.
• It often involves hidden mathematics.
Thus, school mathematics and everyday mathematics are two related yet different
categories of mathematical knowledge. Students encounter school mathematics
during school as part of their everyday life and may to various degrees engage
and immerse themselves in this institutionalized body of mathematical
knowledge. At its best, students’ enculturation into school mathematics will be
supported by the mathematical knowledge and experiences they have from other
arenas of life. School mathematics can feed back into their everyday
17
mathematics, guiding them to useful insights that would be difficult to develop
without schooling, so that they can be active and reflective participants in society
and engage in lifelong learning.
The research literature on everyday mathematics has yielded long lists of
mathematical practices in everyday situations (Carraher & Schliemann, 2002;
Civil, 2002; Presmeg, 2007). These long lists can give the impression that
everyday mathematical practices are relatively stable within a given cultural
context. However, Jorgensen Zevenbergen (2011) suggests otherwise. In her
study of workers in the retail industry comparing younger to older adults’
practices, she found differences between age cohorts. The younger generation
were
[m]ore likely to approach tasks holistically, to use estimation, to problem solve, to
use technological tools to support their work and thinking, to use intuitive methods,
and to see tasks esthetically. (Jorgensen Zevenbergen, 2011, p. 99)
In contrast, the older generation valued mental and accurate calculations higher
and did not value the use of technology. This study has two important
implications. First, the different practices and attitudes in the younger and older
generations meant that some of the older workers would not recognize and value
the contributions of the younger workers. Second, it indicates that young workers
have developed new ways of working that are adapted to post-industrial
workplaces which suggests that generational differences can be understood as
adaptations to shifting cultural and technological conditions.
A key concern in studies on everyday mathematics is how and to what extent
the mathematics learnt in school supports mathematical practices outside of
school, and the other way around: how can the mathematics learned outside of
school support the learning of mathematics in school? According to the review
by Wake (2014), the literature on this theme is conflicting both on how such
processes are conceptualized and to what extent such transfer or
recontextualization occur. Despite the controversy surrounding this issue, there
appears to be a consensus that workplace mathematics, irrespective of where and
when it occurs, is fundamentally different from school mathematics due to the
differing aims of these two broad categories of mathematical practices (Wake,
2014). Wake summarizes this difference as such:
In school, mathematics is mainly the object of study, whereas in the workplace, it is
used as a tool for mediating activity that is inevitably focused on the productive
outcomes that are the raisons d’être of the workplace. … Mathematics in
18
workplaces takes on different formulations from those familiar in schools because of
the very different role it plays for those who engage with it and also because of the
diverse range of technologies (considered in the widest sense) available and used.
(Wake, 2014, p. 272)
This conclusion is also supported by the reviews on everyday mathematics by
Carraher & Schliemann (2002) and Presmeg (2007). Due to the importance of the
issue of the gap between school mathematics and everyday mathematics, I will
highlight the methodological complexities in this research. Lave (1988)
presented several cases of everyday mathematics and their (lack of) connection
with school mathematics. She reported that people who could perform virtually
error-free mathematics in everyday situations (e.g., grocery shopping, dieting)
had significant difficulties when solving problems of a similar mathematical
structure in a school-like context. This, she argues, supports the view that there is
little transfer between school and everyday mathematics. However,
Greiffenhagen and Sharrock (2008) pointed out that the shopper’s problems were
not comparable to the tasks in the standardized arithmetic test. The arithmetic
done at the supermarket was mostly of a very basic sort, predominantly involving
doubling, tripling, and estimating differences. In contrast, the problems in the test
were more complex and of a form not routinely performed in everyday situations.
This meant that the performances in the shopper’s problems and in the school-
like test required very different competences and the participants were unequally
prepared for the two kinds of tests. Also, in the school-like tests, it is likely that
the participants would attempt to use the kinds of formal procedures that are
typical of school mathematics and, insofar that they are not accustomed to these
methods, a lower performance ratio on this test is exactly what you would expect,
regardless of how much transfer there is between school and the everyday. Thus,
Lave’s (1988) conclusions may be overstated.
Another investigation of the relationship between mathematical competences
developed in different contexts (not necessarily schools) found that people who
are experts in designing and interpreting graphs in one specialized domain can
have significant struggles with making sense of DVs from other domains, which
indicates that the mathematical skill of graph reading is highly dependent on
knowledge and experience, not only with graph reading, but from the field that
the graph pertains to (Roth, 2003). Other studies have documented that the
mathematical practices in workplaces, such as engineering, can be remarkably
sophisticated (Frejd & Bergsten, 2016; Gainsburg, 2006), and the practices are
19
highly adapted to the concrete tools and conventions that are available (Roth,
2014). Thus, out-of-school mathematics encompasses the full range from the
mundane to highly complex mathematical practices and tends to be intimately
adapted to its concrete circumstances.
The reading of DVs in journalistic media is a form of everyday mathematics
that can occur at any time of the day and in any location and is not limited to a
particular institution. The recontextualization of DV reading skills from one
context (e.g., school, work) to another (e.g., domestic life) might be problematic
or demanding. However, research on DV reading as a form of everyday
mathematics is scant and needs further research.
In the next section, I review literature pertaining to mathematization as a
social process, which connects everyday mathematical practices (or the absence
thereof) with cultural-historical processes of increasing use of mathematics.
2.4 Mathematization as a social process and its implications
An underlying premise for this dissertation is the view that the growing use of
DVs in journalistic media is part of a larger process, namely a mathematization
of society. This chapter will start by summarizing the literature on
mathematization as a social process and its implications (Section 2.4.1). Then, I
will present relevant literature on mathematization processes related to Study A
and Study B. For Study A, I take a closer look at the mathematization of
meteorology and weather prediction (Section 2.4.2). For Study B, I take a closer
look at the mathematization of epidemiology (Section 2.4.3).
2.4.1 Overview of the literature on mathematization
Mathematization is a term describing processes in which something becomes
more mathematical (Chevallard, 1989; Davis & Hersh, 1986; Keitel, 1989). To
illustrate the idea of mathematization and its implications, I will use the example
of the mechanical clock, which Keitel (1989) borrowed from Lewis Mumford.
This example illustrates the main points made in these early publications
(Chevallard, 1989; Davis & Hersh, 1986; Keitel, 1989). The mechanical clock
yielded a formalized quantification of time, and people started to identify time
with clocks rather than the sun’s position in the sky. This illustrates the first
implication of mathematization: when mathematics is applied to practical
phenomena, its application frames and shapes the way the phenomenon is
understood. Further, Chevallard (1989) distinguished between two modes of
20
presence of mathematics, explicit and implicit. Explicit mathematics is the
mathematics that is handled and used visibly, for example by the clockmaker
when designing gearwheels. On the other hand, there is implicit mathematics
which was previously explicit mathematics that has become ‘frozen’ in objects
and routines and is no longer visible. This is the case of the inner workings of the
clock as it appears to the user. Much implicit mathematics can be found in
digitalized modern society. Some authors have also noted that the highly
specialized division of labor that characterizes contemporary working life is a
driving force in making applied mathematics invisible, in which mathematical
modelling is only handled by a few experts in complex multidisciplinary
organizations (Jablonka, 2003; Williams & Wake, 2007). When mathematics is
applied to a human activity, it tends to have a formatting effect because the
activity will be organized around the mathematical models (Skovsmose, 1994).
The formatting effect of mathematics and the processes in which explicit
mathematics is transformed into implicit mathematics are the driving forces of
what Keitel (1989) and Chevallard (1989) call demathematization.
Demathematization refers to processes in which mathematics in use becomes so
crystallized in routines that it is taken for granted; embedded in technology to the
extent that it becomes opaque and invisible; and so commonplace that it becomes
devalued. Demathematization is a challenge for mathematics education because
it creates a gap between school mathematics and out-of-school mathematics.
Chevallard goes further and suggests the following:
No modern society can live without mathematics. [However, i]n contradistinction to
societies as organized bodies, all but a few of their members can and do live a
gentle, contented life without any mathematics whatsoever. (Chevallard, 1989, p.
49, emphasis in original)
Thus, for Chevallard, most people live their lives entirely disconnected from
mathematics, except for school mathematics, which suggests that no
mathematical literacy is needed in everyday life in the modern world. On the
other hand, Jablonka (2003) discusses the mathematical literacy needed in a
contemporary society and suggests that “mathematical literacy for critical
citizenship” (p. 90) is crucial in a demathematized world so that citizens can
critically evaluate applications of mathematics.
Jablonka and Gellert (2007) provide an extensive literature review and
discussion to clarify the meaning of mathematization and demathematisation. Of
21
interest to this dissertation is their treatment of trust and citizenship. First, their
treatment of trust in relation to demathematisation:
The process of demathematisation affects strongly the values associated with
different kinds of knowledge and skills. For the user of technology it becomes more
important to, first of all, simply trust the black box and, then, to know when and
how to use it – for whatever purpose. If it turns out to be inefficient, ineffective,
erroneous or disastrous, nobody can be blamed – it was the technology’s fault. From
this perspective, demathematisation reduces the feeling for, and the acceptance of,
responsibility: a car’s antilock braking system is taken as a licence for driving fast;
formal assessment of personal creditworthiness ensures that a loan is not approved
to the wrong people. When faults still occur, then this is a request for a better
technology – technology that is designed to be foolproof. (Jablonka & Gellert, 2007,
p. 8)
Hence, Jablonka and Gellert extend the implications of demathematization to
include a transfer of trust from people to the technology imbued with implicit
mathematics. Second, the implications of demathematization to citizenship:
being a critical, reflective and participatory citizen demands not only following
the rules made by authorities, but also critical evaluation of rules and responding
with criticism when needed. As mathematics becomes more implicit, it becomes
exceedingly difficult to assess its validity and implications (Jablonka & Gellert,
2007; see also Skovsmose, 1998).
Straehler-Pohl (2017) presented three examples of (de)mathematization and
suggests that the form of demathematization has changed since the early work on
this topic. He argues that the formatting power of mathematics has received
public attention over the past years in the form of critical attention to the power
of enterprises using Big Data computer algorithms, such as Facebook and
Google. But even though this formatting power may have become more explicit,
he suggests that the increased awareness of the power of algorithms has not
translated into critical action.
The growing use of DVs in journalism is sometimes connected to the Big
Data revolution, that is, computer-driven analyses of large data sets (Coddington,
2015; Engebretsen & Kennedy, 2020; Kennedy & Hill, 2018). Jablonka (2017)
discusses the role of (de)mathematization in the context of Big Data. According
to her, Big Data represents a shift in mathematization: while mathematization
traditionally occurred by people constructing mathematical models to which
mathematical procedures were applied, Big Data enabled manufacturing models
22
directly from data. Furthermore, the emergence of machine learning means that
models can be derived from data without a human modeler. In other words, the
work of modelling can be outsourced from human actors to machine actors.
Another interesting observation is that Big Data models typically rely on vast
amounts of data, sometimes all the data on a subject. The backing from extensive
data changes the perception of validity and uncertainty associated with models
generated by machine learning: “more data” is regarded as closer to a truth.
Categories and connections produced by Big Data analyses such as correlations
between Facebook likes and intelligence (e.g., that liking “curly fries” is
positively correlated with high intelligence) derive their value from their
predictive value which can be used in for example marketing and give no clues to
why these correlations exist (or if they are spurious variables). In sum, Jablonka
(2017) argues that the discourse around Big Data constitutes a particular truth
discourse in which “theory” is replaced by “data”.
The idea that mathematization occupied a special role in the history of science
has received attention in science and technology studies (Bourdieu, 2004;
Ferreira & Silva, 2020; Gingras, 2001). I will review some of this literature to
illustrate other potential implications of mathematization. Bourdieu (2004)
positions the mathematization of physical science as one of the most important
factors in the scientific revolution that started with Copernicus in the late
medieval period. Ferreira and Silva (2020) offer a summary and comparison of
four different accounts of processes of mathematization in physical science, the
accounts of Koyré, Dijksterhuijs, Burtt and Kuhn. They conclude that
mathematization alone cannot give a satisfactory account of the evolution of
science. Furthermore, some scientists and scientific branches that relied more on
experimental methods to build theory, as opposed to using experiments to
confirm mathematical predictions, do not fit the narrative that mathematization
was the main transformative event. Nevertheless, the mathematization of physics
was transformative, as researched by Gingras (2001). According to Gingras
(2001), the work of Newton accelerated the mathematization of physics and he
identifies three key implications. First, it created a strong separation between
professionals who were fluent in mathematical methods and could read and
contribute to the field, and amateurs who were effectively excluded from the
field due to lacking mathematical schooling. Eventually, the social exclusion
became so strong that only a relatively small group of practitioners with
relatively heterogenous competence remained active in the field. Next,
23
mathematization changed the epistemology of physics: to explain gradually came
to mean explain in terms of Newtonian calculus and not to explain in terms of
mechanical causes. Third, there was an ontological change which was closely
connected to the epistemological change: the mathematization of physics moved
the focus away from substances and mechanical causes, over to quantitative
relations between objects, space and time. This he called desubstantialization.
Wigner (1960) praises the contributions from mathematics to the natural
sciences under the famous title the unreasonable effectiveness of mathematics in
the natural sciences. Halevy et al. (2009) paraphrase Wigner (1960)’s famous
statement as the unreasonable effectiveness of data and argue complex problems
such as language processing cannot be approached by simple mathematical
formulae, but that algorithms generated by machine learning from vast amounts
of data have proven surprisingly effective. In a similar fashion, Debreu (1991)
summarized the successful contributions of mathematics in economic theory.
Some related studies have also been conducted that center on the concept
datafication, that is, the process of putting a phenomenon “in a quantified format
so that it can be tabulated and analysed” (Mayer-Schönberger & Cukier, 2013, p.
78). I regard datafication as a form of mathematization that is closely related to
Big Data. With datafication, phenomena such as friendships, interests and
emotions are quantified, yielding new ways of perceiving them and introduces
new challenges and tensions (Mau, 2019; Mayer-Schönberger & Cukier, 2013).
Challenges associated with datafication include that data sets tend to be opaque
and difficult to scrutinize (Cushion et al., 2017; Lowrey et al., 2019) and
inaccessible (Snaprud & Velazquez, 2020).
Bourdieu (2004) and Ferreira and Silva (2020) caution that the implications
of processes of mathematization vary between fields, contexts, and moments in
history. This literature review indicates qualitative differences between
mathematization processes in physics, Big Data, and journalism. The literature
from mathematics education research holds the trivialization, devaluation and
opacification of mathematics and the three main implications of
mathematization. Researchers from other fields also highlight other implications,
such as ontological, epistemological and social restructuring. Whether and to
what extent these implications apply to the growing use of DVs in journalistic
media remains an open question.
24
2.4.2 The mathematization of meteorology and weather forecasting
The science of meteorology has undergone a process of mathematization that has
changed the way the atmosphere is understood, and the tools used to model and
predict it (Bauer et al., 2015; Harper, 2008; Kristiansen, 2017). Quantitative
reasoning in meteorology has a long history. Already in the fourth century BC,
Aristotle wrote the Meteorologica in which phenomena such as cloud formation
and precipitation were explained in terms of air movement and temperature
(Frisinger, 1972). In 1904, Bjerknes envisioned a numerical approach to weather
forecasting and formulated a set of equations and fundamental concepts that are
still in use today (Kristiansen, 2017). However, the development of mathematical
weather models was halted by the immense computational power required
(Harper, 2008). During the interwar period and the second world war, increased
political pressure due to military and economic factors led to increased interest in
numerical weather prediction (Benjamin et al., 2019; Harper, 2008). The
resulting efforts and improved computer technology led to the first runs of
computerized numerical weather predictions in the US in the 1950’s (Harper,
2008), and by 1970, computers were in use in daily weather prediction in
Norway (Kristiansen, 2017).
The mathematization and computerization of meteorology, as well as
continued improvements in atmospheric models, computational power and input
data, has gradually improved the accuracy and time span of weather predictions
(Bauer et al., 2015; Benjamin et al., 2019). In the late 1990’s, global weather
models using global satellite atmospheric data were implemented and this
resulted in a leap in prediction accuracy (Bauer et al., 2015). The
mathematization of meteorology implied that the theoretical understanding of the
atmosphere became based on measurable entities such as humidity, the
movements of air masses and temperature; and that the labor of weather
prediction moved from human hands to computer programs (Harper, 2008). In
Norway, a later consequence of the mathematization and computerization of
meteorology is that commercial weather forecasting becomes feasible, which
transform weather forecasts into a commodity (Nilsen & Vollset, 2016).
What the mathematization of meteorology implies for the readers of weather
forecasts – how forecasts change in response to the mathematization of
meteorology and what expectations these changes put on readers – remains in
need of further research and is approached in Study A reported in this
dissertation. Figure 3 presents some of the symbols used to report cloudiness and
25
precipitation in the data that were analyzed in Study A, and gives some cues to
the ways that weather forecasts in Norway have changed during the
mathematization of meteorology. These excerpts indicate that symbols have
become more pictographic, and that the “pie charts” symbolizing cloud coverage
have gone out of use. Hence, reporting cloud coverage appears to have
transformed from a fractional representation to a pictographic representation.
This can be interpreted as a mathematization followed by a demathematization:
First, the phenomenon “how cloudy it is” was represented by a visual fraction.
Later, this fractional representation was replaced by an iconic representation of
clouds which suppresses the quantitative interpretation of cloud coverage and
foregrounds the sensory experience of cloudiness.
Figure 3. Selected excerpts of signs from weather forecasts. From left to right: Legend from
Aftenposten Aften, 1950; legend from Aftenposten Aften, 1980; Aftenposten Morgen, 2000;
Aftenposten, 2015. Reprinted with permission.
2.4.3 The mathematization of epidemiology
Like meteorology, epidemiology has also undergone a mathematization during
the twentieth century (Davey Smith, 2019). The science of epidemiology, that is,
the study of the spread of diseases, was crucial in managing the COVID-19
pandemic. Epidemiological methods were used to monitor and simulate the
spread of the virus and its mutations, the ways it spreads, and suggest mitigation
measures (Lipsitch et al., 2020). The choropleth map in Figure 1 (see section 1.1)
is an example of an epidemiological DV, in which death rates (per 100 000
inhabitants) of the worlds’ countries are visualized. This DV suggests that
epidemiology is mathematized – the use of such DVs indicates that quantitative
26
data and statistical methods are in the foreground of this field. It also shows signs
of demathematizaton. For example, statistical frequencies are translated into
color codes which enable readers to make comparisons without reading the actual
numbers. Hence, the mathematical skill involved is downplayed and trivialized,
and the methods used to generate the data are opaque.
According to Davey Smith (2019), a methodological and epistemological
shift occurred in epidemiological research and practice that started around 1970.
This shift was constituted by an “increasingly formal and mathematized” (p.
1411) approach to modelling diseases, an increasing belief that biological causal
mechanisms could be studied through statistical methods, and that more
statistical data on risk factors will inevitably lead to better assessments of risk.
This development was met with resistance from senior researchers in the field,
who argued that the heavy reliance on statistics and mathematical models made
the field drift away from a biological understanding of diseases (Davey Smith,
2019). However, while the mathematization was met with reluctance, a different
implication was that epidemiological research was met with much more respect
and trust in the scientific community at large – once statistical methods were the
norm, epidemiological findings became credible. Epidemiology’s embrace of
“formal language and graphical representation of the causal inference
movement” is, according to Davey Smith (2019, p. 1414), unequivocally
positive, primarily because it has accelerated the knowledge accumulation in the
field and strengthened the credibility of the field among scientists and lay people.
More recent developments in the field include integrating statistical methods
with biological and sociological knowledge, which have widened the scope of
the field so that factors such as medical history, poverty and religious attendance
are regularly considered (Davey Smith, 2019).
Among the mathematical models used to model the COVID-19 pandemic are
nonlinear systems of difference equations and nonlinear systems of ordinary
differential equations that are within reach of many undergraduate mathematics
courses (Meyer & Lima, 2022).
2.5 Research questions
In the above, the literature on data and DVs in journalism, literacy for reading
DVs, mathematization and everyday mathematics have been reviewed. I have
identified some gaps in the literature. For example, there is a need for further
research on how people read DVs in out-of-school contexts (Engebretsen, 2020;
27
Kennedy & Hill, 2018), the demands that journalistic DVs put on readers
(Aguilar & Castaneda, 2021; Jablonka & Bergsten, 2021; Kwon et al., 2021),
more research on everyday mathematical activities outside of schools and
workplaces, and more empirical research on the implications of processes of
mathematization. Journalistic DVs offer a chance to shed more light on these
gaps, because reading journalistic DVs is an example of a mathematical activity
in everyday life that is often taken for granted and that demands a certain
mathematical literacy and reflect processes of mathematization in journalism and
society. They can also reflect processes of mathematization in other fields, such
as meteorology when weather data are reported, and epidemiology when
pandemic data are reported. Therefore, the overarching research question for this
dissertation is:
What do DVs in journalistic media imply for readers, from perspectives of
mathematical literacy, everyday mathematics and mathematization as a
social process?
What journalistic DVs imply for readers is a broad question and will in this
dissertation encompass several topics. A core concern in this dissertation is the
non-trivial connection between processes of mathematization and the
mathematical literacy expected of citizens. This dissertation includes descriptions
of the mathematical literacy that journalistic DVs demand of readers – what kind
of mathematical literacy is expected of citizens to be productive readers of DVs,
and how have these expectations changed over time? Further, it includes an
analysis of how the scientists behind DVs are represented and how their
representation changes over time vis-à-vis processes of mathematization. Finally,
it includes a study of challenges and opportunities encountered when adults read
socially relevant journalistic DVs and thick descriptions of their VNL.
The three perspectives – mathematical literacy, everyday mathematics and
mathematization as a social process – are ordered so that the perspective most
closely tied to the empirical research presented in this dissertation comes first,
mathematical literacy. This is also the perspective with the smallest scope, which
is peoples’ actual or expected engagement with mathematics within specific
social contexts (e.g., the COVID-19 pandemic) and specific means of
representation (e.g., the visual and numeric modes of communication). The next
perspective, everyday mathematics, has a larger scope which is the mathematics
28
wanted and developed by people in all venues of everyday life, which can
happen within or between institutions such as workplaces, schools and
bureaucratic agencies. Finally, the perspective of mathematization as a social
process has the largest scope as it deals with cultural and historical processes in
society in which phenomena become more mathematical, and the implications of
such processes.
The first study reported in this dissertation, Study A, is based on analyses of a
corpus of NWFs retrieved from the most read newspapers in Norway, spanning
the period from the second world war until the present day. This corpus offered
opportunities for exploring historical changes in the way that socially relevant
quantitative data was reported to lay audience in popular newspapers, and what
these changes implied for readers. Second, NWFs was an arena in which the
science of meteorology met journalistic media. Therefore, this corpus also
offered an opportunity to explore how scientists and lay audiences interacted.
Because meteorology and operational weather forecasting were mathematized
and computerized during this period (Bauer et al., 2015; Harper, 2008;
Kristiansen, 2017), an analysis of how meteorology and weather prediction was
presented to lay audiences can give indications to implications of the
mathematization of meteorology as well as implications of the growing use of
DVs in journalism.
The issue of how data reporting has changed historically and what this
implies for the expectations that are put on readers is approached in Paper I
(extended abstract in Section 5.1). Here, I asked two research questions. First, to
elaborate on how the data reporting changed over time, I asked how has the use
of semiotic resources in newspaper weather forecasts changed from 1945 to
2015? Also, I ask the second RQ which was how do the changes in newspaper
weather forecasts change the readers’ role, and what do the changes indicate for
everyday mathematics? These questions targeted the mathematization of NWFs
after the second world war, and what this mathematization implies for everyday
mathematics and the demands put on readers’ mathematical literacy.
The issues of how meteorology and weather prediction is represented to lay
audiences and how meteorologists are presented is approached in Paper II
(extended abstract in Section 5.2). The research question for this paper was how
have the identities, relations and discursive roles in newspaper weather forecasts
changed since 1945? This question targeted implications of the mathematization
29
of NWFs after the second world war, and how the mathematized science behind
NWFs was represented, popularized or obscured.
The second study reported in this dissertation, Study B, was based on
journalistic DVs related to the COVID-19 pandemic. The abundance of DVs that
were used in the journalistic coverage of the COVID-19 pandemic offered
opportunities to explore the implications of the increasing use of DVs in
journalism in the present moment, such as the expectations that the DVs put on
readers and the challenges and opportunities that readers encounter when reading
them. The journalistic DVs that were used to report on COVID-19 reflected both
the mathematization of epidemiology, and the mathematization of journalism.
The issue of the expectations that COVID-19 DVs in journalistic media puts
on readers is studied in Paper III (extended abstract in Section 5.3). Here, an
online collection of such DVs is analyzed with a framework developed from
Tønnessen’s (2020) concept VNL to answer the research question what
characterizes the VNL expected of readers of COVID-19 DVs in online news
media? This question targets the VNL expected of citizens to stay updated on the
COVID-19 pandemic.
Next, the issue of the opportunities and challenges encountered by readers of
journalistic DVs related to the COVID-19 pandemic is studied in Paper IV
(extended abstract in Section 5.4). Two young adults were observed while
reading an online collection of journalistic COVID-19 DVs and interviewed
about their reading experience. The sessions were analyzed according to a second
framework based on Tønnessen’s (2020) concept of VNL to answer the research
question what characterizes adults’ VNL when reading and making sense of
journalistic COVID-19-related DVs? This question targeted the opportunities
and challenges that adults can encounter when reading journalistic COVID-19
DVs.
Journalistic DVs of two kinds – NWFs and COVID-19 DVs – were studied
through the lenses of everyday mathematics, the mathematical literacy expected
of citizens, and the implications of processes of mathematization. At the same
time, these DVs acted as windows into these same phenomena. Thus, NWFs and
COVID-19 DVs were studied through everyday mathematics, mathematical
literacy and mathematization; and everyday mathematics, mathematical literacy
and mathematization were studied through NWFs and COVID-19 DVs.
30
31
3 Theoretical perspectives
The main theoretical approach in this dissertation is social semiotics (Halliday,
1978; Kress, 2003; Van Leeuwen, 2005). Social semiotics allows me to see data
visualizations (DVs) as socially situated semiotic artefacts. However, social
semiotics does not offer theoretical perspectives on literacy, human activity,
mathematization and sociology. Therefore, I also drew on other theoretical
traditions including a theory of literacy developed by Hasan (1996, 2003) and
developed into a framework for DV reading by Tønnessen (2020). This
framework is known as visual-numeric literacy (VNL). Further, I drew on the
sociology of Giddens (1984, 1990, 1991) which I used to theorize processes of
mathematization. In each of the four papers, the most relevant theoretical
concepts are presented. Therefore, the purpose of this chapter is to give a
coherent and comprehensive overview of the theoretical perspectives in this
dissertation with an emphasis on their epistemological and ontological
assumptions, and a discussion of advantages and challenges related to how the
theories are applied. First, I give an introduction to social semiotics (Section 3.1)
after which I present the supporting theories (Section 3.2) and a discussion of the
combination of different theories (Section 3.3).
3.1 Social semiotics
Social semiotics is a theory of how human actors use semiotic resources (words,
numbers, images, etc.) to make meaning in particular social situations. I follow
van Leeuwen’s formulation of social semiotics (Van Leeuwen, 2005). Van
Leeuwen (2005) traces the roots of social semiotics to two main sources. First,
Ferdinand de Saussure’s formulation of semiotics was an important inspiration
and foundation for social semiotics. Second, the work of Michael A. K. Halliday
was essential in, among others, laying out the ontological and epistemological
foundations of semiotics as something fundamentally social and cultural: people
use language for meaning making, and the meaning of any given coherent
utterance (e.g., a word, a sentence) derives from all its past uses and its potential
meanings under the concrete circumstances it was used. In the 1980’s and later,
Van Leeuwen and other researchers extended the work of Halliday to include
multiple modes of communication, such as images, music, toys and DVs (Aiello,
2020; Engebretsen, 2013; Kress & Van Leeuwen, 2020). Social semiotic research
often focuses on social issues of accessibility and power related to meaning
making in various contexts (Van Leeuwen, 2005). Van Leeuwen (2005)
32
describes four main semiotic principles that practically and theoretically frame
social semiotics. In the following, I will use a brief outline of these principles as
an introduction to social semiotics. For a more detailed account, see Van
Leeuwen (2005).
The first principle is the understanding of semiotic resources as the things
people use to express meanings. Semiotic resources are actions and objects and
are always material (e.g., ink on paper, pixels on a screen, soundwaves, gestures).
What makes them interesting is how they are used to create meanings. A semiotic
resource has a theoretical semiotic potential which derives from its past and
potential uses, and an actual, situated semiotic potential derived from the
situation of use and the semiotic potentials known and imagined by the user.
With this formulation, the crucial roles of both the cultural and the social
environment in social semiotics are highlighted: the potential meanings of a
semiotic resource is not intrinsic to it, but derives and evolves from the cultural,
social, situational and material conditions under which it is used. To concretize,
consider the map in Figure 1 (see Section 1.1). This map is materialized as pixels
on a screen or scribblings on a piece of paper, depending on the platform you use
when reading this dissertation. Its theoretical meaning potential derives from all
the past uses of maps. As such, it derives its meaning from a culturally
conventionalized and formal sign system. In its original context, it was used to
report on the state of the COVID-19 pandemic. In the particular social context of
this dissertation, however, it is used to motivate, justify and explain issues that
underly my research, and to connect the reader’s meaning making of these DVs
to the studies on DVs in this dissertation. Its actual, situated semiotic potential
depends on the reader’s specific needs, past experiences and interests. For a news
reader, it may for example, imply that mitigation measures in Europe were not
severe enough.
The second principle is semiotic change and transitions. Social semiotic
research is interested both in how semiotic systems appear and work at a
particular moment in time, and how semiotic systems evolve, change and
transition over time and between cultures. Semiotic changes also reflect social
changes, and social semioticians are interested in capturing the interplay between
semiotic and social change. In Figure 1, this principle surfaces in multiple ways.
First, this DV could only be produced and disseminated in a time when computer
technology is sufficiently advanced to support interactive DVs. Further, the
editorial decision to publish this DV tells of how quantitative data was valued
33
during the pandemic, and in contemporary times more generally. The text about
the map explains that the map is updated several times a day, which reflects the
urgency of the situation. Cultural, technological and situational conditions are
constantly changing, and so journalistic DV practices change with them.
The third principle is semiotic rules. While social semioticians recognize
rules that govern the association between semiotic resources and their meanings,
and rules that stipulate how semiotic resources should be combined to make
messages, they also recognize that these rules are malleable and depend on
context. In Figure 1, several semiotic rules are in play. There are the
conventional rules of maps – that the outline represents the countries of the
world, that north is upwards, and that the map is centered on Europe. Further, the
map has a more concrete situated meaning in the COVID-19 pandemic: it tells
about how the reported rate of deaths associated with the virus is very high in
some parts of the world and low in other parts. This distribution is told through
interaction between two rules: locations in the map and the color code. The
meaning of the color code – that darker shades of red correspond to higher death
rates – is made accessible both by the conventional use of colors (red=danger),
and through the legend above the map that connects each color to a numerical
meaning.
The fourth principle is semiotic functions. Social semioticians regard coherent
utterances as always fulfilling three functions: the ideational function, which is
how an utterance describes the world (who, what, to whom, where, etc.), the
interpersonal function, which is how an utterance creates and enacts social roles
and relations, and the compositional function, which is how semiotic resources
are arranged into coherent wholes that we recognize as texts or communicative
events. Applying these three functions to Figure 1, we can infer that the
ideational function here is primarily about the geographical distribution of death
rates – how many people are reported dead from COVID-19 and where? We
further note that the DV represents death rates in a very abstract way – we don’t
see people, but a representation of relative numbers of deaths. This abstraction
has an interpersonal function regarding the role of the reader. It positions the
reader as a dialogue partner with high cognitive abilities – a person capable of
perceiving and processing complex visual messages. Finally, we have the
compositional function. The map is composed so that metainformation – about
how often the numbers are updated, what data should be shown (death rates or
infection rates) and how the color codes should be interpreted – is at the top. In
34
this position, the information is given high salience. According to western
reading conventions, it thus becomes a natural starting point for the reading
process.
In the four papers, I use several additional concepts from social semiotics for
analysis and discussion, including, but not limited to, genre, style and social
actors. These concepts are explained in the papers and not repeated here.
There were several reasons for choosing social semiotics as the main
theoretical approach. An important reason was that social semiotics offers a well-
developed perspective on multimodal communication, which means that it
offered relevant analytic concepts for the study of DVs (Jewitt et al., 2016; Kress
& Van Leeuwen, 2020). These analytic concepts include style, which arises in
modes such as gestures, text, images and graphs (Van Leeuwen, 2005). Further,
according to van Leeuwen (2005), social semiotics is best used when it is
combined with relevant concepts from other theories of social interaction. This
openness to integration with other theoretical perspectives further enhanced the
usefulness of social semiotics for my research project. Further, several studies
relevant to this dissertation have already been conducted within social semiotics
(e.g., Engebretsen, 2020; Engebretsen et al., 2018; Tønnessen, 2020), which
enabled me to build on their work.
3.2 Supporting theories
While social semiotics provides the main theoretical basis for the studies
presented in this dissertation, I complemented it with perspectives on psychology
and sociology to position the studies in a broader framework. In this section, I
present accounts of the theoretical perspectives employed insofar as they are
relevant to the present studies (Sections 3.2.1–3.2.4) and a commentary on the
way I have combined theories in this dissertation (Section 3.2.5).
3.2.1 Hasan’s theory of literacy and Tønnessen’s VNL
The concept VNL was developed by Tønnessen (2020) to capture the form of
literacy related to reading and making sense of visually represented numerical
data, hence the name. This concept was developed from a concept of literacy
presented by Hasan (1996, 2003), which again was built on social semiotics
(Halliday, 1978, van Leeuwen, 2005) and Vygotsky’s learning theory (Vygotsky,
1978). In this dissertation, VNL is developed further to conceptualize and
analyze DV reading and sense-making. Because Tønnessen (2020) developed her
35
concept from Hasan (1996), I will start this chapter by clarifying the meaning of
Hasan’s (1996, 2003) original concept. Then, I present the basic tenets
underlying the concept and the aspects of VNL as adapted by Tønnessen (2020).
Hasan presented her concept of literacy on different occasions, each time with
different emphasis. In Hasan (1996), the emphasis is on the role of pedagogical
institutions in producing and reproducing individuals’ literacy. Here, she posits
that there are three qualitative different aspects of literacy. These three aspects
are presented as stages in a developmental process where the first can develop
into the second and the second can develop into the third. Hence, the latter aspect
includes the former. The first aspect is recognition literacy, which concerns the
recognition of signs and their meaning. Here, language is regarded as an
inventory of forms, codes and rules, and language as social action is ignored. The
next aspect is action literacy which is about using language to reach personal
goals – language as social action. The third and final aspect, reflection literacy, is
about exploring the boundaries of existing conventions and knowledge through
reflective enquiry on how language forms create power relations, serve the
interests of certain groups, and so forth. Hasan (1996) sees reflection literacy as a
necessary precondition for the evolution of knowledge. The three aspects of
literacy are presented with a transformative undertone: As recognition literacy is
concerned only with the formal meaning of signs within a sign system,
recognition literacy in isolation cannot prepare students to use their language to
express their opinions, goals, and viewpoints. For these tasks, action literacy is
needed. Therefore, action literacy emerges as the synthesis of recognition literacy
and peoples’ need to use language productively in society. Further, action
literacy alone does not suffice to explore the limits of existing conventions or
generate new knowledge; reflection literacy emerges as the synthesis of action
literacy and the need to explore and expand the existing culture.
In Hasan (2003), the emphasis is on the role of literacy in a globalized world
saturated with conflicting ideologies. Here, the three aspects of literacy retain
their essence from Hasan (1996) – recognition literacy as meaning making
confined to the sign system, action literacy as meaning making for self-
expression, and reflection literacy as exploring and challenging conventions.
However, as her focus in this article is on the role of literacy in the globalized
world, she has shifted the focus from the individuals’ literacy to literacy
education, and in this perspective, the three forms of literacy emerge as historical
stages in educational institutions. Recognition literacy is the first and oldest
36
stage. Action literacy emerged as a development over recognition literacy
prompted by the need to ‘write to mean’, that is, writing to express an opinion, a
meaning or a purpose, within established genres. Reflection literacy is presented
as a necessary next step for education, a synthesis yet to come. Hence, Hasan’s
(1996, 2003) three forms of literacy can be regarded as developments in the
literacy of the individual or as developments in educational institutions. In either
case, each aspect is a synthesis of the former. So, Hasan’s (1996, 2003) concept
of literacy can be applied both at the level of the individual as developmental
stages, or at the level of institutions as historical stages. According to Hasan
(1996), reflection literacy includes a well-developed action and recognition
literacy.
There are four tenets underlying Hasan’s (1996) concept of literacy. These
tenets arise from the combination of social semiotics (Halliday, 1978; van
Leeuwen, 2005) and Vygotsky (1978).
• The first tenet concerns the things people use to convey meaning. Semiotic
resources are material elements that can be interpreted (e.g., sound waves
making words, ink on paper making DVs, pixels on screens making
maps). The rules that connect semiotic resources with their potential
meanings are called codes. For example, in Figure 1, colors are a semiotic
resource (because it is something material that can be interpreted), and
some of the codes that connect this semiotic resource to a meaning is
elaborated in the legend.
• The second tenet concerns the nature of codes. When people interpret a
semiotic resource, meaning making is both a formal and a situated act. It
is formal because semiotic resources have accrued meaning in formal sign
systems that are agreed upon within a cultural group. For example, maps
have evolved over a long time into a group of related sign systems, such
as the Mercator projection, that are recognized globally. This formal sign
system is not sufficient for giving the map its meaning within the social
context. Therefore, meaning making is also situated because it arises from
participation in social events. In the case of Figure 1, its meaning derives
both from the formal codes in play; as well as from the context of
COVID-19.
• The third tenet concerns the role of the individual. People have different
life stories and different goals. Therefore, they have learned different code
37
systems and may have their own idiosyncratic codes; and they may have
different ways of participating in social contexts which influence the
situated meanings they make. This tenet implies that people can have
unequal access to DVs.
• The fourth tenet concerns the multiplicity of sign systems. Most of the
time, sign systems interact to make semiotic wholes. For example, in
Figure 1, the sign system of maps interacts with a conventional use of
colors (e.g., red=danger) and a color-coded scale. This interaction yields
opportunities for innovation as novel combinations can yield novel
meaning potentials or better accessibility. An implication of this tenet is
that DV interpretation is deeply context sensitive: it is sensitive to the
context of its interacting sign systems, and it is sensitive to the social
context of the reader.
These tenets form the basis of Hasan’s (1996) concept of literacy, which consists
of the three aspects mentioned before. These aspects were adapted to DV reading
by Tønnessen (2020), and I here present Tønnessen’s adaptation.
• The first aspect is called recognition literacy and addresses semiotic
resources and codes within formal sign systems. To make sense of DVs, it
is necessary to recognize relevant semiotic resources and their meanings
within the sign systems, such as axes, bars, lines, bubbles, colors, labels
and legends.
• The second aspect is called action literacy and extends from the sign
system context to the situated meaning of DVs. Action literacy refers to
the capacity to use DVs to reach personal goals. Key skills include
understanding how different design options create different impressions of
the data, producing DVs for specific aims, and making sense of DVs as
texts connected to a social situation.
• The third aspect is called reflection literacy. Hasan included this aspect in
her framework because sign systems, knowledge and social norms evolve,
and this aspect of literacy targets the capacity to participate in the
production of knowledge. Key aspects in this capacity are the ability to
use DVs to reflect, enquire, analyze and critique. She does not suggest that
everyone participates in knowledge production, nor that holding these
abilities guarantee that the actor contributes to new knowledge. However,
insofar as knowledge and practices evolve, Hasan (1996) attributes this
38
evolution to reflection literacy. Reflection literacy includes critiquing
design choices, sources and data handling; enquiring into the socio-
political impact of DVs; and exploring alternative ways of conveying
relevant information.
My purpose with adopting Tønnessen’s (2020) adaptation of Hasan’s (1996,
2003) concept is to use it for analysis of VNL at the level of individuals. It has
been used to analyze the form of literacy that was demanded and invited by
readers of journalistic COVID-19 DVs, and young adults’ actual VNL when
reading journalistic COVID-19 DVs. Hence, the focus in this dissertation is on
the level of individuals.
3.2.2 Giddens’ sociology
Processes of mathematization extend into the domain of sociological phenomena
such as division of labor, trust and technological developments. Therefore, a
sociological framework of living in contemporary society that integrates these
phenomena was useful. The sociological understanding in this dissertation is
grounded in Anthony Giddens’ account of ‘late modernity’ (Beck et al., 1994;
Giddens, 1984, 1990, 1991). According to Bryant and Jary (2003), Giddens’
sociological works are remarkable for the critical integration of many of the main
ideas from modern and classical sociology and are regarded as one of the most
holistic and applicable accounts of contemporary society. It is particularly
relevant for my research because of the many parallels between his framework
and the concepts of (de)mathematization, which I will return to shortly. His
perspective on modernity builds on a dialectical approach to ontology (categories
are evolving and co-constituent in dialectical relationships) and hermeneutical
approach to epistemology (understanding and knowledge is achieved through
iterative processes). Giddens’ sociology is extensive, and I will limit my
presentation here to a few important principles. In The Consequences of
Modernity (Giddens, 1990), he provides a pertinent framework of
phenomenology in late modernity that positions individuals in a series of four
dialectics. I will present them briefly here, draw parallels to the literature on
mathematization, and return to them later in discussing my findings.
The first dialectic is the dialectic of displacement and re-embedding. On the
one hand, globalized economy and modern communication technologies have
yielded a peculiar experience of proximity and distance that Giddens calls
displacement. The local shopping mall, however familiar and grounded in the
39
local environment, is modeled on principles of commerce that were developed
elsewhere and was placed into the local community. Because it is modeled on
generic principles, you can expect to find similar layouts and similar products
elsewhere, even on different continents. Infectious diseases that emerge in a
different continent can within weeks become a health hazard in the local
community. Telephones, e-mail, chat applications and global news can bring us
‘closer’ to people and events on the other side of the globe than our neighbors.
Such instances of displacement, of warping of time and space, are amplified in
late modernity compared with previous eras. On the other hand, the displacement
mechanisms have yielded opportunities for new ways of creating connections – it
is possible to cooperate with people and companies with similar interests in far
off countries, and work experience gained in a local business can be applicable in
similar businesses elsewhere, making it easier to start a new life in a different
location. Thus, displacement is not a one-way path of increasing alienation, but
creates new opportunities for re-embedding. An example of this dialectic can be
found in the mathematization literature in the discussion of the role of global Big
Data technology companies such as Facebook and Google which, one the one
hand, displaces relations and information, but, on the other hand, radically re-
embeds people and organizations in ways that were unimaginable only a few
years ago (Straehler-Pohl, 2017). Another example lies in the mechanical clock,
which yielded a standardized and formalized concept of time that enabled trade
and cooperation over long distances at the expense of subjective, sensory and
local concepts of time (Keitel, 1989). Journalistic DVs can play a role in this
dialectic in multiple ways. On the one hand, DVs often acts as abstract and
impersonal representations of events which distances the readers from the event
and the people involved (Kennedy & Hill, 2018). On the other hand, DVs have
become a staple of journalism internationally and DV producers (e.g.,
weathercasters) can have an international user base, which means that journalistic
DVs create coherence between international news outlets.
The second dialectic is the dialectic of intimacy and impersonality. Late
modern society is one in which we encounter more people than ever before, often
in brief and superficial ways such as economic exchanges, work experiences and
casual encounters with neighbors and people on the streets. Despite the many
impersonal social encounters, Giddens maintains that we do not live in a world
of strangers. Rather, he describes a complex “transformation of intimacy”
(Giddens, 1990, p. 142) which is prompted by the dynamics of displacement and
40
re-embedding: Communication technologies have enabled intimate relationships
to be sustained at distances and social media constitute spaces where people can
nurture their identity. Such technologies create many opportunities for forming
intimate bonds. This dialectic also has a parallel in the literature on
mathematization: Illouz (2007) suggests that digital dating has transformed and
formalized the search for a partner. This formalization takes the form of
intensified use of data to marketize the individual in a competitive dating market.
Likewise, COVID-19 mitigation measures informed by mathematical
epidemiological models created new barriers for intimacy and social activities,
but also prompted people to use digital platforms for social interaction such as
social group calls and outdoors walks (Deejay & Henne, 2024; Skinner, 2020).
The third dialectic is the dialectic of expertise and reappropriation. Still
following Giddens (1990), contemporary society is immersed in expert systems
such as law, crafts, finance, psychology, and so forth. Expertise in such systems
is increasingly difficult to obtain as these systems become more abstract,
specialized and multinational, which have yielded a radical and unprecedented
division of labor. One consequence of this radical division of labor is that the
inner workings of the things we are surrounded by are specifically opaque – just
think of all the electric apparatuses used in everyday life whose inner workings
are incomprehensible to most users. However, processes of what Giddens call re-
appropriation also occur as citizens interact with various expert systems:
although no one can be an expert within more than a few expert systems, lay
people must obtain a rudimentary understanding of the systems with which they
routinely interact. Such processes of re-appropriation can relate to all aspects of
life, such as child rearing, romantic relationships, economy – and journalistic
DVs. As DVs become a regular feature in journalism, the capacity to make sense
of them becomes an important skill and so lay readers need to learn the basics of
DV reading. The dialectic of expertise and reappropriation creates particular
bonds of trust, because when you rely on others’ solutions to problems, such as
an engineers’ solution to car brake systems, you need to trust that the system
works. These bonds of trust appear when the actors are separated in time or space
or when one is particularly ignorant of the systems at work. This dialectic, too,
has parallels in the literature on (de)mathematization. First, the radical division of
labor in contemporary society, in which advanced tasks are handled by experts or
automated machinery, is the main driver of the black-boxing and opacification of
mathematics in society (Jablonka, 2003; Jablonka & Gellert, 2007; Keitel, 1989;
41
Williams & Wake, 2007). Second, Jablonka and Gellert (2007) recognize the
trust bonds necessitated by the role of expertise in late modernity in relation to
the use of technology: mathematized expert systems such as antilock braking
systems in cars or loan credit assessments function as guarantees for fast driving
or dealing out loans. When these guarantees don’t measure up, the systems are at
fault and not the users. Likewise, contemporary meteorological models are far
too complex for lay people to grasp, so when a weather forecast is wrong, the
visible expert – the forecaster – becomes the scapegoat (Compton, 2018).
The fourth dialectic is the dialectic of privatism and engagement. A
characteristic of the globalized late modernity is the presence of high-
consequence risks for the global or local community. This was evident in the
COVID-19 pandemic, when the risk of global mass infection was on the agenda.
Also, the risk of ecological collapse and financial crises are continually looming.
The presence of such high-consequence risks, in which the actions of any
individual are negligible, puts the individual in a dialectic of pragmatic
acceptance of issues that are out of their reach and activism for collective
solutions to these problems. However aware people are of the risks of the use of
fossil fuels, or international travel in pandemic times, they are torn between
private and collective interests. This dialectic also has a parallel in the literature
on (de)mathematization. According to Straehler-Pohl (2017), many people today
are aware of how big-tech companies like Facebook and Google are exploiting
personal data for commercial purposes but show few signs of critical action in
response to this critical awareness. Straehler-Pohl (2017) asks why this critical
awareness does not translate effectively into critical action. Giddens’ (1990)
dialectic of privatism and engagement indicates an explanation to this lack of
critical action: while critical action, such as boycotting or new legislation, could
benefit the collective, any one individual does not have the power to impose such
action in a way that will significantly change the status quo. Therefore, people
may choose to pragmatically accept some level of exploitation in exchange for
the benefits these companies offer. In the context of journalistic DVs, it has been
observed that DVs pertaining to high-consequence risks such as global warming
and pollution, the reader may dismiss the intended message through motivated
reasoning, thereby justifying a passive approach (Engebretsen, 2020).
42
3.3 A reflection on combining theories
The aim of this dissertation was to study journalistic DVs from the perspectives
of mathematical literacy, everyday mathematics and mathematization as a social
process. Therefore, it was necessary to apply theories that enabled me to
understand journalistic DVs through these three perspectives. First, it was
necessary with an overarching theory that allowed me to conceptualize and
analyze journalistic DVs as semiotic artifacts. The social and situational aspect of
meaning-making was important for this project. Social semiotics (van Leeuwen,
2005) was an appropriate choice as it met these needs. Next, I needed theories
that could address the perspectives of mathematical literacy, everyday
mathematics and mathematization as a social process. In selecting theories to
address these perspectives, an important consideration was that the theories must
provide relevant concepts. Further, it was important that the theories meet some
requirements for theory networking, as I will connect the three perspectives in
Section 6.5.
According to Jungwirth (2010), two requirements for theory networking are
(1) that the theories have concordant paradigmatic positions so that the theories
align in their assumptions about basic phenomena such as cognition and the role
of culture, and (2) that the theories complement each other by allowing the study
object to be investigated in units of different size. In my case, the study object is
journalistic DVs, and the three perspectives offer analytical units of various sizes
to be investigated.
The first perspective is mathematical literacy. The unit of analysis here is the
individual reader in interaction with specific journalistic DVs in a particular
situation. As there is already significant work on literacy within social semiotics
(Hasan, 1996, 2003; Kress, 2003), it was natural to look for a theory of literacy
here. I selected VNL (Tønnessen, 2020). VNL is based on a theory of literacy
developed by Hasan (1996, 2003). Hasan synthesized ideas from social semiotics
with dialectical learning theory and sociology. Hence, Hasan’s (1996, 2003)
theory of literacy is concerned with the relationship between semiotic signs,
social action and ideology, inspired by the concept of dialectics.
The second perspective is everyday mathematics. The unit of analysis here
encompasses aspects of peoples’ mathematical practices in everyday life. As the
literature was torn on how everyday mathematics should be defined, the
sociological literature offered a definition of the everyday that clarified the
43
meaning of this perspective for this dissertation. Lefebvre’s (1947/1991) work
also builds on the concept of dialectics.
The third perspective is mathematization as a social process. The unit of
analysis here is the role of mathematics in society, seen from a structural
perspective (Jablonka & Gellert, 2007). This unit required a holistic and
structural theory of contemporary society. This was catered for by Giddens’
(1984, 1990) sociology, which too has roots in the concept of dialectics. Further,
Giddens (1990, p. 139) offers “a phenomenology of modernity in terms of four
dialectically related frameworks of experience”. These four dialectics, elaborated
in Section 3.2.1, offer an opportunity to systematically interpret the implications
of processes of mathematization, seen through the case of journalistic DVs. Thus,
Giddens’ (1990) theory can offer a historical and structural perspective on
mathematization.
Hence, the theories used fulfill Jungwirth’s (2010) two requirements as they
all share a paradigmatic position in dialectics, and they offer units of analysis of
different sizes to understand the phenomenon of journalistic DVs. The
philosophical roots of social semiotics are not explicitly built on dialectics, but
the perspectives’ notion of semiotics as a cultural, social and evolving
phenomenon is compatible with this idea. Van Leeuwen (2005) explicitly advises
social semiotic analyses to be combined with other theories to reach its full
potential. Moreover, Parkin and Harper (2020) and Wells (1994) provide
thorough arguments for the compatibility between social semiotics and
dialectical theory.
Later, in the discussion of the empirical findings in light of relevant literature
(Section 6.5), I will bring together the three perspectives by positioning
mathematical literacy and everyday mathematics as implications of processes of
mathematization. In the theory networking framework of Bikner-Ahsbahs and
Prediger (2010), this networking can be characterized as a local integration. By
local integration, it is meant one or more theories or theoretical concepts are
“integrated into an already more elaborate dominant theory” (Bikner-Ahsbahs &
Prediger, 2010, p. 496). In this dissertation, the three perspectives are combined
in Section 6.5. Here, mathematical literacy and everyday mathematics are
integrated into the overarching perspective of mathematization as a social
process, positioned in Giddens’ (1990) framework. Meeting Jungwirth’s (2010)
two criteria supports the validity of this theory integration.
44
45
4 Methodology
In this chapter, I will explain how I studied data visualizations (DVs) and related
phenomena such as visual-numeric literacy (VNL), everyday mathematics and
processes of mathematization. I start by positioning my work in a research
paradigm and explore what this paradigm implies for the ontological and
epistemological understanding of the objects of study, which are journalistic DVs
and people reading and making sense of journalistic DVs (Section 4.1). Then, I
use this perspective to explain how the methodologies of my studies shed light on
mathematical literacy, everyday mathematics and different aspects of processes
of mathematization. As mentioned before, a core concern in this dissertation is
the non-trivial connection between processes of mathematization and the
mathematical literacy expected of citizens. An approach to this concern will need
to start from thick descriptions of particular forms of mathematical literacy and
everyday mathematics.
In Section 4.2 and 4.3, I elaborate on methodologies of the studies. I will
explain the research designs, data collections and data analyses, and elaborate on
trustworthiness and validity. Finally, I address relevant ethical and legal
considerations. This includes using journalistic content for research purposes and
maintaining the integrity of interview participants.
4.1 Research paradigm
According to Mertens (2016), a research paradigm is “a way of looking at the
world” (p. 8) and can be summarized by a researcher by describing one’s
axiological, ontological, epistemological and methodological position. Some
authors have described social semiotics as a paradigm in its own right (e.g.,
Bezemer & Kress, 2020). However, I have used ideas and concepts from multiple
theories, and therefore, I position this dissertation in a paradigm that aligns well
with all the theories and aims of this dissertation. Therefore, I position this
dissertation in the paradigm that Wellington (2015) calls interpretivism, in which
the aim is “to explore perspectives and shared meanings and to develop insight
into situations” (p. 26). This is a broad paradigm that allows some freedom in
outlining the position. Below follows a summary of how the dissertation is
positioned along Mertens’ (2016) four axes.
Axiology refers to the study of value and ethics. The core ethical concern in
this thesis is that the research should contribute to knowledge that can promote
citizenship and empowerment. This aim is approached by studying the role of
46
mathematical knowledge and mathematical literacy in everyday life, how
processes of mathematization transform society, and how people can be invited
to participate in mathematical activities in their everyday lives.
Ontology refers to the study of what exists and how it exists. In line with
social semiotics and dialectics, I hold that bodies of knowledge and human
practices – mathematics, literacy, journalism, meteorology, epidemiology, etc. –
are cultural entities that arise from needs and tensions in history and will
continue to develop for as long as they exist (Van Leeuwen, 2005). Interaction is
realized through material, culturally shaped mediators such as gestures, words
(spoken or written), colors, and so forth. (Kress, 2003; Van Leeuwen, 2005;
Vygotsky, 1978). The driving forces of change and development are historically
arising dialectical contradictions, causing tension, negotiation and change
(Giddens, 1984; Lüders et al., 2010; Vygotsky, 1978).
Epistemology refers to the study of how knowledge develops or is developed.
I hold that knowledge is socially, culturally and historically situated and
constituted, continually being produced and reproduced through cultural
mediation. Every individual carries a unique perspective from their unique life
history. In any situation of social interaction, including a research situation,
knowledge, as a shared cultural resource, is co-produced (Mertens, 2016).
Methodology refers to the study of how methods and research designs can
lead to knowledge within a research paradigm. I hold that the social world has
both qualitative and quantitative aspects, and research can therefore involve both
aspects. Because research situations generate co-production of knowledge
between researchers and participants, it is an aim to keep the research context as
naturalistic as possible while acknowledging that a perfect reproduction of the
natural environment is impossible. A key aspect of such a methodological
approach is the understanding of the unit of analysis. Regarding this aspect, I
follow Vygotsky’s (cited in Jornet & Damşa, 2021) dialectical line of reasoning,
which he expressed by the metaphor of water molecules: if you want to study
water, a study of oxygen atoms and hydrogen atoms will not suffice. Likewise, to
study how people read journalistic DVs, the unit of analysis should be the reader
and the journalistic DV in interaction. To study how DVs change over time, they
should, as far as possible, be situated in their proper historical context. By
studying the components in isolation, the characteristics of the whole get lost.
Thus, to capture the characteristics of the whole, it is the whole that must be the
unit of analysis. In practical research, it is rarely an option to study very large
47
units of analysis. Therefore, as I elaborate below, the scope of the unit of analysis
in the studies reported in this dissertation is a compromise between what is
practically feasible, and the reach of the phenomenon being studied.
4.2 Study A: Newspaper weather forecasts
Processes of mathematization unfold and accumulate historically (Jablonka &
Gellert, 2007). In Study A, implications of two parallel historical processes of
mathematization were studied. The first was the mathematization of meteorology
and weather prediction, which unfolded during the 20th century, and accelerated
after the second world war when new technologies enabled the development and
use of computation-intensive methods (Harper, 2008). The second historical
process was a mathematization in news discourse, whereby journalistic media
carried more mathematical aspects such as DVs and mathematical language such
as ‘percent’, ‘correlate’ and ‘significant difference’ (Coddington, 2015). In the
intersection of these two processes, I studied whether newspaper weather
forecasts (NWFs) showed characteristics of mathematization. Readers are crucial
actors in this interaction, whereby NWFs mediate between readers and
meteorological scientists and determine the expected literacies (Figure 4). So, the
mathematization of NWFs has direct implications for readers and shapes the
mathematical literacy expected of citizens in everyday situations.
Study A is based on one corpus of NWFs which were analyzed in different
ways. The data collection is described below (Section 4.2.1). In Section 4.2.2 and
4.2.3, the analyses that are reported in Paper I and Paper II, respectively, are
described. Finally, relevant ethical issues are discussed (Section 4.2.4).
Figure 4: The mediation of NWFs between weather science and readers.
4.2.1 Data collection for Study A
To conduct a feasible study, it was necessary to sample so that the corpus had a
manageable size. I limited the data in three ways. First, I only used weather
forecasts from printed newspapers, thereby excluding those broadcasted on radio,
television and the internet. This decision ensured that all the data have a similar
format, which made it easier to make comparative analyses of different time
Meteorology and
weather
prediction
Newspaper
weather forecasts Readers
48
periods. Second, I decided to limit the scope to the two most read newspapers in
Norway. This ensured that the NWFs would have reached a large audience while
keeping the size of the corpus relatively small. Third, I limited the corpus to the
period 1945–2020. This time frame was chosen because it captures the main
milestones in the mathematization of operational weather forecasting, from the
first experimental attempts of running computer models in the 1950’s (Harper,
2008), via the early implementation of computers in day-to-day weather
forecasting in the 1970’s (Kristiansen, 2017), to the implementation of global
satellite systems, continued improvements in computer technology and
atmospheric models, and commercialization of weather services in more recent
decades (Bauer et al., 2015; Nilsen & Vollset, 2016). The year 1945 was also a
practical starting date because weather forecasting was on hiatus during the
second world war and resumed operation in public newspapers in 1945. Within
the given time period 1945-2020, I retrieved one sample from each newspaper
every five years and chose the sample date March 1st or the closest available date,
for the years 1945, 1950, …, 2020.
By these criteria, the newspapers used were Aftenposten and Verdens Gang
(VG) because these were the most read newspapers in the time frame
(Mediekatalogen, 2022). Until 2012, Aftenposten came in two daily issues.
Aftenposten morgen was a national newspaper and Aftenposten aften was a
regional newspaper for the capital region. Because the regional issue shows some
interesting deviations from the overall patterns, it was included in the analysis.
The corpus consisted of 46 samples. Note that the latest two samples were not
included in Paper I because the paper was written in 2019, before the 2020
samples were available.
Hence, the unit of analysis for Study A was a collection of historical
newspaper weather forecasts taken in the historical context of meteorology and
weather broadcasting. The historical collection of NWFs was retrieved through
archives; and the historical context of meteorology and weather broadcasting was
reconstructed from historical sources. Using terminology from social semiotics,
this unit of analysis can be called the NWF genre, that is, a class of texts unified
by a common communicative goal, disseminating weather predictions, and a
recognizable form and content (Van Leeuwen, 2005). Due to the historical
character of the corpus, it was not possible to study readers’ interaction with
these NWFs.
49
4.2.2 Data analysis reported in Paper I
The research questions for Paper I were how has the use of semiotic resources in
newspaper weather forecasts changed from 1945 to 2015? and how do the
changes in newspaper weather forecasts change the readers’ role, and what do the
changes indicate for everyday mathematics? The first research question was
approached by analyzing the corpus according to a two-way framework. First,
the semiotic resources in each NWF were categorized in one of four categories
(verbal-numeric text; maps; graphs; tables). Second, I opted to categorize the
degree of salience (high; middle; low) because this allowed me to indicate which
semiotic resources are more or less prominent in each NWF. The categories of
semiotic resources, based on Few (2012), were defined thusly:
• Verbal-numeric text is text based on written words and numbers that is a
separate compositional element in the NWF and not embedded in a table
or graph.
• Maps are geospatial representations of (a part of) the world.
• Graphs are defined according to three characteristics, “[v]alues are
displayed within an area delineated by one or more axes”, “[v]alues are
encoded as visual objects positioned in relation to the axes” and “[a]xes
provide scales (quantitative and categorial) that are used to label and
assign values to the visual objects” (Few, 2012, p. 45). This includes line
graphs and histograms.
• Tables are defined by two characteristics, (1) “[i]nformation is arranged in
columns and rows” and (2) “[i]nformation is encoded as text (including
words and numbers)” (Few, 2012, p. 43), colors, or icons.
For levels of salience, the following operational definitions based on Van
Leeuwen (2005) were used:
• High salience means an item clearly stands out, for example, by being
placed partly over other components, conspicuous colors, large size, and
so forth.
• Semiotic resources that are neither standing out nor being significantly
downplayed are categorized as middle salience.
• Low salience is used for components that are much downplayed, for
example, by being small, blending into the background color, and so forth.
50
This is a rather quantitative approach because it involves counting frequencies of
nominal variables and assessing ordinal degrees. This illustrates how
interpretative research can merge quantitative and qualitative aspects.
In the following, this analysis is illustrated with an example, and the example
analysis is summarized in Table 1. The exemplary analysis is of the NWF shown
in Figure 5 and 6. Figure 5 shows the ‘raw’ NWF; and Figure 6 shows the same
NWF but with the relevant semiotic resources highlighted in different colors. The
first stage in this analysis consists of registering and counting the semiotic
resources used. Here, there is one piece of verbal-numeric text, highlighted in
blue; one map, highlighted in yellow; zero graphs; and four tables, highlighted in
red. Next, the salience of each semiotic resource is categorized. The piece of
verbal-numeric text is placed at the bottom of the page, with no conspicuous
colors and relatively small size. It has a large and clear headline and is not
disturbed by other components. Thus, it is not highly salient, nor inconspicuous.
Therefore, it is categorized as middle salience. Next, the map is mostly green,
which stands out among the rest of the color palette which is mostly grey, black
and red. It is placed above the other elements, which is emphasized using
shadows. It is also large. Therefore, this is regarded as the most conspicuous
semiotic resource in this NWF and is therefore categorized as ‘high salience’.
Finally, there are four tables. The table in the upper left corner has a black
background which almost merges with the dark blue background upon which it is
placed. It is small and uses small, thin typesetting. Therefore, this table is
categorized as low salience. The other three tables are made using the same
colors, which are white, grey, black and yellow, and are of medium to large size
relative to the other components of this NWF. Therefore, these three tables are
neither particularly conspicuous nor inconspicuous, and are therefore categorized
as middle salience.
The second research question in Paper I was approached by identifying trends
in the findings from the analysis of the use of semiotic resources as described
above. These changes were discussed in light of other relevant events, such as
weather forecasts broadcast on television, to characterize trends in the everyday
mathematical practice of reading NWFs. The key affordances of each semiotic
resource are as follows:
• Verbal-numeric text primarily affords narration (Kress, 2003).
51
• The non-verbal semiotic resources (maps, graphs tables) afford to show
larger amounts of data (Few, 2012).
• Graphs and maps afford displaying relationships among and between data
(Few, 2012).
• Maps afford showing the geospatial distribution of data (Few, 2012).
• Tables afford looking up individual data values (Few, 2012).
The semiotic resources coupled with their affordances assisted me to identify
how the role of the reader changed over time.
Figure 5: NWF from VG, March 1st, 2015. Reprinted with permission.
52
Figure 6: NWF from VG, March 1st, 2015, with the analyzed semiotic resources highlighted.
Verbal-
numeric text
Map
Graph
Table
High
salience
Middle
salience
Low
salience
Table 1: Overview of the analysis scheme for Paper I, with the outcome of the exemplary
analysis inserted.
53
4.2.3 Data analysis reported in Paper II
In Paper II, the aim was to study how NWFs changed during the mathematization
of weather prediction, aiming for an analysis to reveal further implications of
mathematization. In this paper, the interpersonal aspects of semiotics were
targeted. The research question for Paper II was how have the identities, relations
and discursive roles in newspaper weather forecasts changed since 1945? To
answer this question, the main analytic device was Van Leeuwen’s (2005)
concept of style. He defines style as “[t]he manner in which a semiotic artefact is
produced, or a semiotic event performed, as contrasted with the discourse and
genre it realizes” (p. 287). Hence, style has to do with features in semiotic
artefacts and events such as typefaces, ornamentation and adjectives that tells
something about who the discursive participants are and how they are positioned
toward one another.
Van Leeuwen (2005) presents a framework for styles, in which each style is
defined according to certain characteristics. These styles were developed to
analyze content in the magazine Cosmopolitan and therefore needed adaptation
before they could be applied to the analysis of NWFs. Among the styles in van
Leeuwen’s (2005) framework, I deemed three of them as relevant to my analysis
and used them as a starting point. These styles were expert style, conversational
style and advertising style. I used theoretical and empirical work on science
communication to adapt the framework to my purposes. Further details on this
adaptation can be found in Paper II. Because the ‘expert’ in science
communication is a scientist, I renamed ‘expert style’ to ‘scientific style’.
The use of DVs is not covered in Van Leeuwen’s (2005) framework.
However, Van Leeuwen (2005) sees the use of scientific and formal jargon as a
characteristic of scientific (or ‘expert’) style. DVs with scientific content
(meteorological maps and graphs) can be regarded as part of the scientific jargon,
particularly when scientific concepts are encoded into these. Further
extrapolating from the characteristic verbal registers of each style to the non-
verbal register of DVs, it is reasonable to expect that DVs in conversational style
exhibit less formalism and more playfulness (e.g., sun icons with smiles and
sunglasses); and that DVs in advertising style exhibit more aesthetic approaches
to DVs. The characteristics of each of the three styles are summarized in Table 2.
The next step in the analysis was applying this framework to the corpus and
determining what or which style each NWF exhibits. This analysis was
performed by taking the NWFs one-by-one and catalogue which characteristics
54
they exhibit. Once I got an overview of the characteristics for a NWF, I would
determine which style(s) it exhibits. In most cases, the NWF would be consistent
with one of the styles in all aspects. However, there were some cases when
characteristics of multiple styles were used in one NWF. In these cases, the NWF
would be categorized as a mix between these styles. To concretize the analysis
process, and illustrate the issue of style multiplicity, I proceed to present an
illustrative analysis of a NWF exhibiting multiple style characteristics, Figure 7.
Styles
Aspects
Scientific style
Conversational style
Advertising style
How are social actors
represented?
Describes social actors in
third person or passive
voice
Includes social actors,
often in first and second
person.
Direct address from
sender to reader
How is the data
focused?
Data is concerned with
accuracy and uncertainty
Data is concerned with
sensory experience
Data is concerned with
consumer interests
Data appearance
Data appears objective
Data appears subjective
Data appears attractive
Evaluation
No evaluative statements
Uses evaluative
statements
Uses positive adjectives
Verbal registers
Uses technical language
from the meteorologists’
register
Uses vernacular language
Uses poetic language
DVs
Technical DVs
Informal DVs
Polished DVs
Relationships
Top down
Equality
Bottom up
Table 2: Characteristics of three different styles.
I will present the analysis in the same order that the characteristics appear in
Table 2. Therefore, I start with the representation of social actors, proceed to data
focus and appearance, and so on.
Social actors. In Figure 7, the following social actors are represented: (1) The
Norwegian Meteorological Institute, represented with their name, logo and an
URL. Their role is not stated, but they presumably serve as data sources. (2)
Three online weather services, all represented with a URL and a very brief verbal
presentation. Their role is to serve as sources of additional information. None of
the social actors are personalized. The social actors who designed the NWF are
not represented.
Data focus and data appearance. All data are represented without an explicit
margin of error, and no data is represented as uncertain. Because the data are
represented as facts, and because all data are represented via non-sensory metrics
(e.g., millimeters of precipitations; degrees centigrade), the data appears
objective.
55
Evaluation. There are no evaluative statements. All adjectives are from the
meteorologists’ register.
Verbal registers. As mentioned, the verbal register employed is consistently
scientific and formal. Aside from the explanation of geostrophic wind, the
terminology is only descriptive of weather conditions (precipitation, cloud
coverage, wind, etc.) and does not include causal explanations of the predicted
phenomena (that would include, for example, high-pressure zones, fronts or air
streams).
DVs. This NWF includes a map (with one region enlarged), a satellite image
that also functions as a map, three line graphs and three bar charts. In the map
and satellite image, information is encoded with icons (vectors for wind, cloud,
sun and raindrops for cloud coverage and precipitation) and numerals (numerals
embedded in the vectors for wind strength; numerals embedded in circles for
temperature). The map has a clean design, showing a relatively detailed coastline
of Norway. The satellite image is more naturalistic. The line graphs and bar
charts also have a clean, minimalistic design.
Relationships. The sender takes an authoritative stance by asserting the
weather as objective facts and positioning him/herself as an expert in
meteorological theory. This is indicative of a top-down relationship. On the other
hand, the sender goes a long way to personalize the forecast: the reader can look
up weather predictions for all major cities in Norway as well as timely vacation
destinations, domestic and foreign. If the presented information does not suffice,
the reader can consult the online services that presumably contain weather
predictions for even more locations. Hence, the sender also strives to
accommodate the needs and desires of the reader, as in a salesperson-customer
relationship.
Conclusion of the analysis: This NWF exhibits many characteristics of
scientific style (few social actors are represented, and their representation is
impersonal; the data are represented as objective and certain; there are no
evaluative statements; the sender assert themselves as highly knowledgeable;
uses scientific terminology and relatively technical DVs; and offers a top-down
relationship with the reader). On the other hand, there are some characteristics of
advertising style (the icons used for representing weather data in the maps and
tables have an aesthetically appealing design, and the extensive use of tables and
links to additional weather services gives good opportunities for personalization).
56
Thus, the style of this NWF can be viewed as a hybrid of scientific style and
advertising style.
Figure 7: NWF from Aftenposten Morgen, March 1st, 2005.
4.2.4 Ethical and legal issues
Study A is what Webb et al. (1966, as cited in Bryman, 2016) call a ‘non-reactive
method’, i.e., a method “that do not entail participants’ knowledge or their
involvement in research” (p. 277) because, likely, the data were generated
without an assumption that it would one day be analyzed. This means that the
57
research setting did not change the artifacts studied, so it was not necessary to
take reactive effects into consideration.
The main ethical issue in research analyzing published documents concerns
ownership of the material (Jewitt et al., 2016). Even though the NWFs I use are
available in online archives, they are the property of the newspaper editorials and
are protected by Norwegian law (Åndsverkloven, 2018, § 1-2). Therefore, I have
asked for permission to use the material for research purposes and for reprinting
excerpts of the material in academic publications. Both editorials agreed to my
request.
A second ethical issue arose on the few occasions when NWFs included
portrayals of an individual meteorologist. Because this person is a public figure
in Norway, I deemed it acceptable to not anonymize her in publications.
4.3 Study B: COVID-19 DVs
In the late winter and early spring of 2020, a new opportunity to study
journalistic DVs associated with processes of mathematization presented itself.
As the COVID-19 pandemic started to spread across the globe, DVs showing
descriptive and projective statistics related to the virus (death rates, infection
rates, predicted death rates, etc.) were instrumental in introducing and
maintaining mitigation measures (Jablonka & Bergsten, 2021). The journalistic
use of DVs during COVID-19 forms the basis of Study B which is reported in
two papers, Paper III and Paper IV. Both papers utilized a collection of web-
based journalistic DVs related to the COVID-19 pandemic developed by the
most read Norwegian newspaper, VG. This web page
(https://www.vg.no/spesial/corona/) was launched in March 2020 and was
frequently updated.
Study B approached this web page in two ways. Paper III reports on analyses
of a collection of screenshots from this page to document the expectations it put
on readers. Paper IV reports on analyses of in-depth interviews with two young
adults reading and interacting with this web page. In the following, I describe the
collection of the data reported in Paper III (4.3.1), the data analysis reported in
Paper III (4.3.2), the data collection for the data reported in Paper IV (4.3.3) and
the data analysis reported in Paper IV (4.3.4); and discuss ethical and legal issues
related to Study B (4.3.5).
Both papers have in common that the concept of VNL is used to analyze the
capacity to read and make sense of DVs. However, because the two papers are
58
based on different data materials and different aims, for each paper the concept
was developed into a different analytical framework.
4.3.1 Data collection used for Paper III
During the first wave of the pandemic, when mitigation measures made
interviews and observation research difficult, I opted to study the mathematical
literacy expected of readers by analyzing a corpus of DVs. Inspired by the
method of studying DVs suggested by Aiello (2020), the first step in Study B
was collecting a representative sample of DVs from this webpage. We observed
that the design of the webpage changed frequently during this early period.
Therefore, to get a sample that represented the page in this early stage, all the
content on the page was captured via screenshots on three different occasions.
These occasions were March 27th, April 8th and April 21st, 2020. To ensure that
all content was captured, one of my supervisors and I made independent captures
each time. The captures were made with multiple digital tools: full-page screen
captures; full-page print-to-pdf, screen recordings as the page was scrolled
through, and screenshots of the individual DVs in all available variations. This
yielded a rich data set that was representative of how the webpage appeared
during the first wave of the pandemic.
To limit the corpus to the most used DVs, we only included the DVs that
were present on all three captures. After duplicates were removed, this data
collection yielded a corpus of 24 DVs in the following formats:
• Ten line graphs
• Three bar charts
• Two choropleth maps
• One histogram
• One sector diagram
• One Sankey diagram
• Three DVs in which the reader can choose between line graph and
histogram
• Two DVs in which the reader can choose between area graph and
histogram
• One DV in which the reader can choose between area graph, line graph
and histogram
4.3.2 Data analysis reported in Paper III
In Paper III, the aim was to describe what the COVID-19 DVs on VGs web page
expects of readers in terms of VNL. This led to the research question: What
characterizes the VNL expected of readers of COVID-19 DVs in online news
59
media? To answer this RQ, it was necessary to adapt Tønnessens’ (2020) concept
of VNL into a suitable framework. This framework was developed by first
thoroughly studying the original work by Tønnessen (2020) and Hasan (1996,
2003), and then find relevant analytic concepts suited for analyzing web-based
DVs. The main source for analytic concepts was Kirk (2019). The theoretical
framework for Paper III is presented in Table 4.
Each DV in the corpus was analyzed according to this framework. Hence, for
this study, the unit of analysis was individual DVs in isolation. To show how this
framework was applied, an exemplary analysis of Figure 8 is now presented.
The analysis starts with recognition literacy, in which the aim of the analysis
was to find the semiotic resources and codes that readers were expected to
recognize. First, the semiotic resources were listed. These are: curved lines of
varying widths; rectangles; colors; written words and written numbers. Second,
codes are listed. The most important codes at work here are those of the sign
system of Sankey diagrams: the curved lines show the flow from the categories
on the left to the categories on the right; and the width of each line corresponds
to the magnitude of that flow. Color codes are used to group the main categories
on the left – Italy is grouped with turquoise, Norway with blue and Austria with
lilac – and the other categories are grey. The categories are labeled with written
words telling what they represent; and a written number telling the number of
people in each category. This DV offers little explanation – there is no text
offering reading guidelines and no legends. However, the design is conventional
for Sankey diagrams. Hence, access to this DV is restricted to readers who can
decode Sankey diagrams without aid.
In the next step, action literacy expectations were analyzed. In this step, the
aim is to find the intended message of the DV and how this message is realized.
First, the angle of analysis in this DV visualizes where did the COVID-19
infected people in each Norwegian county contract their infection; and how
many were they? The framing for this DV is stated in the meta-text above and
describes that the DV shows the number of cases of infection that VG has been
able to connect to both county of residence and the provenance of their infection.
The DV does not have a pre-given focus; instead, the reader can highlight parts
of the DV by hovering the cursor over a provenance of infection, a county, or one
of the curved lines. This action reduces the color saturation of the chosen element
so that it can be explored without the ‘noise’ from the other curves. This
mouseover action also revealed the number of people in the emphasized curved
60
line. The absence of a pre-given focus and multiple opportunities for mouseovers
affords exploration and makes this an exploratory DV.
Aspect and central question
Concepts and categories
Recognition literacy
What semiotic resources
and codes are readers
expected to recognize?
- What semiotic resources and codes are used?
Examples of semiotic resources: Lines, bars, dots,
colors, written words, written numbers
Examples of codes: Coordinate systems, color codes,
- Are the codes restricted (the reader is expected
to recognize them without guidance) or
explained (the reader is guided to understand
them)? Restriction is the absence of
elaboration.
Examples of explanation: explaining text, legends,
adherence to conventions
Action literacy
What is expected to
understand the intended
message of the DV?
- What is the angle of analysis?
Angle of analysis refers to the aspects of the data that
are highlighted. This is an affordance of the format
used and variables displayed (e.g., a map highlights
the geospatial distribution of a given variable)
- What is the framing?
Framing refers to what data are included in the DV
- What is the focus?
Focus refers to the items of the data that are
emphasized (with conspicuous colors, arrows,
magnified size, etc.) to attract the readers’ attention
- Is the DV explanatory or exploratory?
The DV is explanatory if it suggests a particular
interpretation of the data.
The DV is exploratory if it aids readers to make their
own interpretations, for example by allowing
interaction (mouseovers, alternative angles, frames or
foci, etc.). Explanatory/exploratory form a spectrum.
Reflection literacy
What critical and reflective
engagement does the page
invite for?
- Are there issues related to trustworthiness?
Are there missing data?
Are data collection methods and data sources stated?
Have the data been modified?
Are the errors in the data or their representation, and
how are errors handled?
Table 4: Theoretical framework for Paper III.
61
Figure 8: Sankey diagram connecting, for infected Norwegians, the region of provenance of
their infection to the county of residence.
In the third step, the DV was analyzed for the reflection literacy that it
invited. There are two reasons why the data were incomplete: first, the data were
limited to the cases that VGs journalists were able to connect to both a county of
residence and a provenance of infection. Hence, all the cases that VG were
unable to connect are missing. Second, the second largest category on the left-
hand side is called ‘abroad’, meaning that the provenance for these cases is not
determined further than ‘not Norway’. Some of the remaining categories on the
left-hand side are overlapping, such as ‘Hong Kong’, ‘China’ and ‘Asia’, which
opens the possibility that some of the cases in the larger category Asia belong to
62
Hong Kong or China. VG stated that they collected the data themselves, but the
exact methods and sources remain unstated. There are no cues suggesting that the
data have been modified. However, in the change log for March 9th, 2020, an
entry explains that these data were gathered from municipal medical doctors and
cross-checked with updates from the Norwegian Institute of Public Health. Thus,
a reader could find information about data collection methods although it is not
easily accessible. There are no cues suggesting that there are errors in the data.
However, an attentive reader may notice that there are two curved lines between
England and Rogaland; that these two lines are not merged into one suggests that
there may be an error in the representation of the data. The presence of the
change log indicates that errors would be corrected shortly once discovered.
Hence, this DV invites some forms of critical reflection regarding the
trustworthiness of the information, such as irregular and overlapping categories,
opaque methods and sources, an irregularity in the data representation, and the
absence of logged errors.
The research reported in this paper has some limitations. Because the unit of
analysis here is DVs, it was impossible to study readers’ interaction with these
DVs. In Van Leeuwen’s (2005) terms, it was a study of the theoretical semiotic
potential of the page and not the situated semiotic potential of the page. This
limitation is addressed and complemented in the next part of Study B, which is
described in the next sections.
4.3.3 Data collection used for Paper IV
In the winter and early spring of 2021, pandemic mitigation measures were
eased, making it feasible to conduct physical research interviews. Therefore, I
opted to complement the textual analysis of journalistic COVID-19 DVs with an
analysis of how people read and make sense of such DVs and of the challenges
and opportunities encountered in this process.
For comparability, I decided to use the same web page from VG as used in
Paper III again. This time, almost a year later, the page had evolved with some
new DVs and some DVs were removed. I developed an interview guide that
targeted some main points of interest, gave the interviews a similar structure that
made them comparable and ensured that all participants were asked the same
main questions. The interview guide consisted of a front page reminding the
researcher of some key principles for high quality research interviews and the
practical frames and purposes of the interview. After this, the themes and key
63
questions for the interview were stated. First, the participant was informed about
the purposes of the interview. Then, the rest of the interview was divided into
seven main parts:
1. Introductory background questions. The participant was asked relevant
questions about their background such as academic specialization, self-
identified gender, familiarity with VGs COVID-19 web page, and what
COVID-19 has implied for the participant personally.
2. Exploration phase. The participant was asked to choose one tab on the
page and explore it at their own pace. The participant was asked to
comment as they read. For practical reasons, the exploration was limited
to one of the five main tabs on the page.
3. The participant was asked about the immediate reaction to reading the
page. This included very open questions (e.g., what do you think about the
web page) and closed questions (e.g., have you seen this before?).
4. Recognition literacy. This part contained questions that targeted
recognition literacy. The participant was asked to explain the meaning of
concrete semiotic resources, if some things were unfamiliar, and whether
some things were easier to understand than others.
5. Action literacy. This part contained questions that targeted action literacy.
The participants were asked about how they used interactive features
when exploring the page (e.g., why did you do X?), if they could see
connections between different DVs (e.g., why do the waves in different
line graphs have different shapes?) and the contextual meaning of the DVs
(e.g., what does this content mean for you?).
6. Reflection literacy. The participant was asked questions intended to
prompt reflection, analysis, enquiry and critique of the DVs and their use.
For example, the participants were asked about their opinion of the way
the data had been visualized and if they thought the representation would
be different if it was produced by a different media insitution.
7. Concluding background questions. The participants were asked if they
think of DV reading as a mathematical activity, their experience with
mathematics, statistics and DVs, and if they had any concluding questions
or reflections.
Because this study involved gathering audio recordings and personal data, it was
necessary to get approval from the Norwegian Centre for Research Data (NSD).
This is elaborated in 4.3.5 on ethical and legal issues.
64
Because I aimed at an in-depth analysis of the participants’ reading and sense-
making of DVs, I opted for a low number of participants, with balanced gender-
representation. For convenience, I recruited participants at the university campus
by approaching them, informed them about the purposes and implications of the
project as required by NSD, and asked if they would participate. Participants
were given a small symbolic gift of a 200 NOK gift card for the university
canteen after the interview was completed. I interviewed two young adults, one
self-identified as male and one self-identified as female.
The low number of participants enabled me to do in-depth analyses that can
complement similar studies with higher numbers of participants (e.g., Yates et
al., 2021; Engebretsen, 2020).
4.3.4 Data analysis reported in Paper IV
The research question asked in Paper IV was:
What characterizes adults’ VNL when reading and making sense of journalistic
COVID-19-related DVs?
To analyze the interview data for this question, it was first necessary to develop a
suitable theoretical framework for VNL. The framework that I developed had
two main inspirations. First, it was based on Tønnessens’ (2020) concept of VNL
which yielded three main codes, recognition literacy, action literacy and
reflection literacy. Second, it was based on Hasan’s (1996) third tenet that
meaning-making is both formal, emerging from the context of agreed-upon sign
systems, and social, emerging from the readers’ social context. From a
preliminary analysis, it was evident that these two contexts for meaning-making
interacted and had implications for the participants’ capability to make sense of
the DVs. Therefore, I developed two sub-codes for each of the main codes
corresponding to these two contexts for meaning-making. This yielded a
framework with six codes. The codes are summarized in Table 5.
To apply this framework, the transcripts were first inserted into a table in
which each statement was numbered. In this table, a column for assigning codes
and a column for inserting comments were added. Then, each statement was
coded. Because the aspects of VNL are intertwined (Hasan, 1996; Tønnessen,
2020), some statements were assigned multiple codes. When I was in doubt,
supervisors were consulted to discuss, validate or change the coding. Often, it
was necessary to interpret the statements in light of the screen activity to make
sense of them. After both interviews were coded, the statements were grouped
65
according to code so that it was possible to summarize all statements pertaining
to each of the six aspects and use this to build a thick description of the
participants’ VNL.
Code
Subcode
Example of utterance
Recognition
literacy
Context of sign system: decoding without
relating it to the experienced world
When I look at the numbers,
I see that the increase is even
more here.
Context of social situation: decoding
while relating it to the experienced world
Here in the graph, the
infections are going up. I
remember that happened.
Action literacy
Context of sign system: action within the
DV
I move the mouse over the
graph.
Context of social situation: action beyond
the DV
I cancelled a trip to my
parents after I saw these
types of graphs.
Reflection
literacy
Context of sign system: reflection on the
DV
The logarithmic vertical axis
makes the graph misleading.
Context of social situation: reflection on
the impact of the DV
DVs do not show how people
experience the pandemic.
Table 5: Theoretical framework used for the analysis presented in Paper IV.
To apply this framework, the transcripts were first inserted into a table in which
each statement was numbered. In this table, a column for assigning codes and a
column for inserting comments were added. Then, each statement was coded.
Because the aspects of VNL are intertwined (Hasan, 1996; Tønnessen, 2020),
some statements were assigned multiple codes. When I was in doubt, supervisors
were consulted to discuss, validate or change the coding. Often, it was necessary
to interpret the statements in light of the screen activity to make sense of them.
After both interviews were coded, the statements were grouped according to code
so that it was possible to summarize all statements pertaining to each of the six
aspects and use this to build a thick description of the participants’ VNL.
To show how the framework was applied, I now present an exemplary
analysis of an excerpt from one of the interviews. This excerpt, reproduced in
Table 6, is from the interview with one of the participants.
In line 200 and 202, the participant explains that she used and compared data
from different media sources in the beginning of the pandemic because they
published different figures. This statement pertains to action literacy within the
66
sign system because the participant shows that she can navigate between DVs
from different sources. It also pertains to reflection literacy within the social
context because when she observes that different media sources present different
data, it implies that the data are not reliable.
Who
#
Statement
Code
Comment
Int
199
Do you ever use other sources than VG’s
information pages?
Bea
200
As I said, I use Fædrelandsvennen [local
newspaper] for Kristiansand. But that was more
in the beginning of last year so that I could
compare with Aftenposten [national newspaper],
for example, the graphs and infections and …
For they were not always the same. So …
Ac(sign)
Rf(social)
Rf: critical
reflection on
the reliability
of the
numbers
Int
201
Right. So to see different …
Bea
202
… sides of the matter.
Rf(social)
Int
203
Yes yes. Can you say a little about how you use
the information you get when you look at this
kind of page?
Bea
204
Well … as I said, I mostly use it to pay attention.
And I use it a lot in everyday conversation and
small talk.
Ac(social)
Int
205
Yeah, ok. So, it is correct to say that you use it to
keep updated on the world around you?
Bea
206
Yes, that would be correct.
Ac(social)
Table 6: Exemplary coding of an excerpt from the transcripts.
In line 204 and 206, the participant explains how she uses the information she
gets from reading journalistic COVID-19 DVs. Her two main rationales are
staying updated on the state of the pandemic and using it for everyday
conversations. Hence, she uses journalistic COVID-19 DVs for making sense of
the world and interacting with peers, which are forms of action literacy in the
context of the social situation.
When the transcripts for a participant were fully coded, they were reorganized
so that all statements pertaining to each code were grouped. These groups of
statements were then used to obtain thick descriptions of each aspect of the
participants’ VNL so that the research question could be answered.
67
4.3.5 Ethical and legal issues
Like with NWFs, the COVID-19 DVs are property of the newspaper editorials
and are protected by Norwegian law (Åndsverkloven, 2018, § 1-2). Therefore, I
asked VG’s editor in chief for permission to use the material for research
purposes and for reprinting excerpts of the material in academic publications,
which was granted.
The interviews were recorded with audio and screen casts, and the
participants’ e-mail addresses were collected. Current Norwegian legislation
requires that I obtain voluntary, specified, informed and unambiguous consent
from the participants, which must be documentable and must enable participants
to withdraw at any moment they wish (SIKT, n.d.). Therefore, the participants
were exhaustively informed about the purposes of the project and data handling,
and they consented in written form. The information letter and consent form were
assessed by NSD. The participants received a copy of the information letter and a
copy of the consent form. I retrieved signed copies of the consent form, which
were stored in a safe facility. The participants also wrote their e-mail addresses in
the consent form.
For the interviews that are reported in Paper IV, I opted to use participant
validation (Slettebø, 2021). This meant that I offered the participants the
opportunity to read through the transcripts and a preliminary analysis, and I
asked them if they felt accurately represented and if they had any comments. The
correspondence went through e-mail. Even if they did not have any comments, I
requested that they confirmed that the e-mail was received and that they still
wished to be part of the study. One participant replied that they were accurately
represented in the transcripts and preliminary analysis and still wished to
participate; the other participant only replied that they had received the e-mail
and still wished to participate. Although this participant validation did not
generate new research data, it strengthened the validity and trustworthiness of the
already gathered data. Further, I consider this as a successful means of
maintaining a high ethical standard by giving the participants the opportunity to
check the data and the analysis, and if they felt fairly represented, empowering
the participants and reinforcing their integrity.
68
69
5 Extended abstracts of papers
The purpose of this chapter is to give an overview of the main content in each
paper, with an emphasis on findings and conclusions.
5.1 Paper I: Trends in everyday mathematics: The case of newspaper
weather forecasts
Introduction: Reading and making sense of newspaper weather forecasts (NWFs)
is a widespread form of everyday mathematics. This is because it involves
interpreting quantitative information from data visualizations (DVs), tables and
text to be used in everyday activities such as choosing an apt transportation form
or deciding how the weekend should be spent. In this paper, the aim is to explore
long-term trends in everyday mathematics through a corpus of NWFs from the
period 1945-2015.
Theoretical framework: NWFs are conceptualized as a genre, meaning that is
has recognizable content, form and function (Miller, 1984; Van Leeuwen, 2005).
To identify the types of mathematics that NWF reading involves, I was interested
in indicating their form – what kinds of semiotic resources are used to convey
meteorological data, and what kind of mathematical activity do they afford? This
led to a theoretical framework consisting of different semiotic resources that
afford different engagements (verbal-numeric text, graphs, maps and tables) and
the relative salience of each semiotic resource which says something about how
much attention they attract (low salience, middle salience and high salience).
Methods: Data were collected from the two most read newspapers in Norway,
Verdens Gang (VG) and Aftenposten. NWFs were sampled from each newspaper
for March 1st every five years in the period 1945-2015. The analysis was based
on social semiotics. First, the semiotic resources in each NWF were categorized
into four categories, verbal-numeric text; maps; graphs; and tables. Next, these
same semiotic resources were categorized into three degrees of salience to
account for how some occupy a more prominent role than others.
Findings and conclusions: The analysis showed that the semiotic resources
used in NWFs changed over time. The main trends were: (1) Verbal-numeric text
dominated in the earlier NWFs, but tables and maps came to dominate in the later
decades. This shift, from verbal to non-verbal forms of communication, occurred
around the year 2000. (2) From ca. 1980, maps started to become a regular
feature in NWFs; and from ca. 2000, tables are the most frequent category.
Tables continued to rise in frequency until 2015. According to Kress (2003), a
70
shift from verbal to non-verbal forms of communication implies a change in the
readers’ role: in verbal communication forms, readers are told how the weather
will be, and the readers’ role is to interpret what is told. In non-verbal
communication forms, however, the information is shown to the reader, and the
readers’ role is to organize this information so that it becomes meaningful and
useful.
Discussion: This paper indicates that the everyday mathematics of reading
NWFs has become more complex over time. Today it requires a more complex
form of mathematical literacy than in earlier times. Although we do not know
whether this finding is generalizable to other forms of everyday mathematics, it
raises the question of schools’ accountability in preparing students for new and
rising mathematical demands in everyday life. Kennedy and Hill (2018) suggest
that school mathematics can be counter-productive in learning students to relate
to data in everyday life because it can make them anxious and unconfident. Other
studies suggest that interdisciplinary school activities that are closely connected
to relevant applications from everyday life are promising ways of making school
mathematics more relevant and applicable for students (Vos & Frejd, 2020).
5.2 Paper II: From scientist to friend to advertiser: Norwegian
newspaper weathercasters’ identities, roles, and reader relations 1945–
2020
Introduction: Meteorology and weather prediction have made substantial
advances over the past century, and these advances have had implications for
peoples’ lives. At the heart of these advances is the mathematization of
meteorology, which entailed the development and implementation of numerical
models and methods for understanding and predicting the weather and climate
(Bauer et al., 2015). Weather forecasting is an arena in which lay people and
meteorological scientists meet, and therefore it offers an opportunity to study
social aspects of meteorologist-reader interaction and how this interaction has
developed vis-à-vis developments in meteorology. This way, potential
implications of the mathematization of meteorology can be explored.
Literature review: According to Wilson (2008), weather casters are highly
visible scientists, and they may use their position to educate their audiences on
relevant topics such as climate change and how weather data are collected.
However, despite their salient position, weather casters’ role as science
communicators is poorly understood (Wilson, 2008). Key issues regarding the
71
social aspects of weather forecasting that remain to be investigated include how
the sender consolidates their identity and conveys trust, which roles the sender
takes, and how the relationship between sender and reader is.
Theoretical framework: The theoretical concept style captures many aspects
of the interpersonal function of communication including identity, roles and
relationships (Van Leeuwen, 2005). Therefore, the theoretical framework for this
paper centers on this concept of style. Because NWFs are forms of science
communication, Van Leeuwen’s (2005) concept was adapted to studying NWFs
by enriching it with ideas from relevant research (e.g., Bucchi, 2013; Gilbert &
Mulkay, 1984; Plough & Krimsky, 1987). On this foundation, a framework of
three different styles (scientific style, conversational style and advertising style)
was defined according to a list of characteristics. Multiple styles can coexist. In
Table 3, the connections between the three styles and identity, roles and
relationships are outlined.
Scientific style
Conversational
style
Advertising style
Sender identity
Scientist
Friend
Salesperson
Sender role
Conveyor of facts
and science
Mediator who
makes weather
predictions
accessible to lay
readers
Making the
weather forecast
attractive and
personalized
Sender-reader
relationship
Top down
Equality
Bottom up
Table 3: Relationship between styles, identity, roles and relationships.
Methods: A corpus of NWFs from the two most read newspapers in Norway,
VG and Aftenposten, sampled at five-year intervals for the period 1945-2020,
was collected through digital archives. The 46 NWFs in this corpus were
analyzed individually to determine the style of the sender. The analysis consisted
of ascertaining which characteristics the NWF exhibits and to see which style
these characteristics are consistent with.
Findings and conclusion: The analysis of styles yielded two main findings.
First, from 1945 to 1995, the scientific style dominated, interspersed with a few
cases of conversational style. Conversational style mainly occurred before 1970.
72
In particular, the NWFs from VG in 1955, 1960 and 1965 all exhibited
conversational style, but VG reverted to scientific style in 1970. Hence, until
1995 the sender mainly took a scientist identity whose role is to convey facts and
scientific explanations, and who relates to the reader in a top-down fashion.
When conversational style was used, the sender takes the very different identity
of a friend, whose role is to make the scientific facts and explanations from the
meteorologists accessible and who therefore relates to the reader as equals. The
role of making the NWF trustworthy was realized in different ways. For the
scientist before 1970, this was realized by asserting oneself as a scientific
authority by using scientific jargon, as well as by providing scientific
explanations to the predictions. After 1970, scientific explanations no longer
appeared. Second, from 2000 onwards, all NWFs exhibited a combination of
scientific and advertising style. Scientific style was realized by using scientific
terminology, non-evaluative presentation of weather predictions, popularized
scientific explanations of meteorological phenomena, and increased use of maps,
graphs and tables. Advertising style was realized by design choices that made the
graphics appear more visually appealing (e.g., colors, shadow effects, semi-
naturalistic weather icons) and by increased personalization and attention to the
needs and desires of the reader (e.g., tables with weather predictions for major
cities, weather predictions for popular holiday destinations, links to commercial
weather services, Q&A columns in which reader’s weather-related questions
were answered). In this scientist/advertiser hybrid style, the senders’ identity
appeared as a hybrid between scientist and advertiser, appearing as a competent
meteorologist while also being a salesperson. Consequently, the role of the
sender is both to convey weather predictions and scientific content, while at the
same time making the content appealing and personalized and providing the
reader with links to further commercial weather services. Because the scientific
content tends to be wrapped in a personalized framing, the relationship between
the sender and reader appears bottom-up: the central focus is tending to the needs
and desires of the reader. In the scientist/advertiser hybrid style, the main
strategies of establishing trustworthiness are appealing to the authority of the
scientist and referencing trusted data sources.
Discussion: Two years stand out as watersheds in the analysis of styles, 1970
and 2000. It is therefore pertinent to ask what historical events in weather
prediction occurred in or around these years that can give cues to why stylistic
changes occurred. 1970 was an important year in the Norwegian meteorological
73
community because the first computer dedicated to numerical weather prediction
was purchased. This purchase which marked the start of operational numerical
weather prediction in Norway. How can this event explain the end of scientific
explanations of weather predictions and the reversion to scientific style? It seems
that the emergence of computerized numerical methods reinforced the trust in
prediction methods, which took away a need to use scientific explanations of the
predictions. Further, this first computer seems to have restored the status of the
scientists who no longer needed to be mediated by a conversationalist. The
second watershed, in the year 2000, also stands out in the history of meteorology
in Norway. Aided by advanced atmospheric models, computerized prediction
methods and computer technology, the first Norwegian commercial weather
company was established in 1998 and rapidly became a competitor to the state-
owned meteorological institute (Mahroum, 2016). The commercial competitor
was likely a driver of advertising style. Hence, the mathematization of weather
prediction was likely an important factor in both events when there were
significant changes in the style of NWFs.
5.3 Paper III: Visual-numeric literacy: The case of COVID-19 DVs for
news media and their expectations of readers
Introduction: During the COVID-19 pandemic, DVs of many forms (graphs,
maps, charts) were used by journalistic media to report quantitative data. While
DVs offer opportunities to convey data in compact ways, research has shown that
people encounter many obstacles when trying to decode and make sense of them
(Engebretsen, 2020; Shah & Hoeffner, 2002). Researchers have also found that
COVID-19 DVs can contain errors (Kwon et al., 2021), and media coverage of
the pandemic sometimes expects a sophisticated form of mathematical literacy
from readers (Aguilar & Castaneda, 2021). In this paper, we aim at
characterizing the mathematical literacy expected of readers of COVID-19 DVs
so as to be informed, active, critical and reflective citizens. To do this, we
conceptualized the capacity to read and make sense of DVs as visual-numeric
literacy (VNL) and developed a theoretical framework for analyzing the VNL
that DVs expect of readers.
Literature review: The concept of mathematical literacy has received much
attention in research literature. A recurring finding is that there are various gaps
between curricula, teaching and the mathematical literacy expected in everyday
life. For example, Bolstad (2020) found that teaching for mathematical literacy
74
focuses too much on procedures and neglects other important aspects of
mathematical literacy that are prescribed by the curriculum. Crucially, too little is
known about the mathematical literacy that is needed and expected in everyday
life (Niss & Jablonka, 2014). This study aims to shed light on this gap.
Theoretical framework: The framework was based on Tønnessens’s (2020)
concept VNL, which distinguishes between three qualitatively different, but
intertwined, aspects of literacy. The aspects are recognition literacy (the capacity
to recognize relevant semiotic resources and their meaning), action literacy (the
capacity to make sense of DVs in their social context and use them to reach
goals) and reflection literacy (the ability to reflect, analyze, enquire and critique
DVs and their socio-political role). For each aspect, appropriate analytic concepts
were employed to enable textual analysis of journalistic COVID-19 DVs.
Methods: Inspired by Aiello (2020), we compiled a corpus of journalistic
DVs pertaining to the COVID-19 pandemic from a web page developed by the
most read Norwegian newspaper, VG. We sampled DVs on three different days
during the first wave of the pandemic in the spring of 2020. To capture the core
content, we limited our corpus to those DVs that occurred on all three captures.
This yielded a corpus of 24 DVs that were analyzed one by one according to our
theoretical framework.
Findings and conclusion: For recognition literacy, the analysis showed that
readers of these DVs were expected to be able to recognize and decode a wide
range of DVs, including (from most to least common) line graphs, bar charts,
choropleth maps, histograms, a sector diagram and a Sankey diagram. To make
sense of these, readers must be familiar with several code systems including
cartesian coordinates, geographical maps, and relative numbers (percent, per
mille and per 100 000). The DVs often expect the reader both to read off specific
values (e.g., how many infected were there on April 14th?) and trends (e.g., when
did the first wave of infections start to taper off?). A few codes were presented
with explanatory text that elaborated its meaning, for example for logarithmic
scales.
For action literacy, the DVs offered a wide range of different perspectives on
the pandemic. The majority of the DVs were exploratory: it was up to readers to
explore the data and assess what they tell about the state of the pandemic,
policies and future decisions. The action literacy also entailed that the readers
were expected to interpret the DVs themselves.
75
For reflection literacy, readers were given some basic information about data
sources, data collection methods, data handling and how projections were
calculated. Further, VG appeared to be open about errors and corrections. This
meant that interested readers were offered opportunities to evaluate VGs
methods, assess their trustworthiness and reflect on alternative approaches to
reporting on the pandemic.
The analysis showed that VGs COVID-19 DVs expected an advanced form of
VNL from their readers, with a broad repertoire of recognition literacy, high
levels of independence and judgement in action literacy, and the ability to
critically evaluate methods for data collection, data handling and data
manipulation.
Discussion: The high expectations raised by journalistic COVID-19 DVs
raised questions of equity: who has access to the mathematized discourse on the
management of the COVID-19 pandemic, and how can mathematics educators
best prepare students and adults for these aspects of everyday mathematics?
Further research is needed on the mathematical literacy expected in everyday
situations; and on the role of everyday mathematical practices as different sites of
learning that can complement school-learned mathematics.
5.4 Paper IV: Making sense of journalistic COVID-19 DVs: An in-
depth study of two adults’ VNL
Introduction: The extensive use of DVs in the journalistic coverage of the
COVID-19 pandemic raised new mathematical expectations on readers. This
paper reports on an in-depth study of two young adults as they read a collection
of such DVs and reflected on their reading experience. The paper aims at
characterizing the particular form of mathematical literacy related to reading
DVs, which we conceptualize as VNL, and for which we adapted an existing
theoretical framework.
Literature review: DVs showing statistics pertaining to the COVID-19
pandemic were vital constituents of a public discussion about political decisions
such as mitigation measures and social and economic policies during the
pandemic (Da Silva et al., 2021; Jablonka & Bergsten, 2021). However, DVs
pose certain challenges for readers and are prone to miscommunication,
misunderstanding and misinterpretation (Engebretsen, 2020; Kwon et al., 2021;
Li & Molder, 2021). Such shortcomings constitute a democratic challenge as it
restricts access to these important discussions, and the growing use and
76
importance of DVs in journalistic media suggests that VNL will be an
increasingly important skill for citizenship in the future.
Theoretical framework: The theoretical framework for this paper is based on
the concept VNL (Tønnessen, 2020). This concept consists of three qualitatively
different aspects, namely recognition literacy, action literacy and reflection
literacy. To adapt the concept to analysis of the data reported in this paper, we
have opted to make a further distinction between the context of the sign system,
which refers to reading and sense-making that does not relate the DV to the
experienced world; and the context of the social situation, which refers to reading
and sense-making that relates the DV to the experienced world. When we applied
this distinction to Tønnessens’ (2020) three aspects, we got six analytical codes.
Methods: Two young adults were interviewed following an interview guide.
During the interview, the participants read an online collection of journalistic
COVID-19 DVs and were interviewed on the reading experience. Screen activity
and audio were recorded. The audio recordings were transcribed and analyzed
according to the theoretical framework. When necessary, the screen recordings
were used to interpret statements. Participants’ names were anonymized as Abe
and Bea.
Findings and conclusion: The analysis of the two interviews revealed two
very different reader profiles. Abe avoided DVs and other numerical content
when possible. Nevertheless, he showed a flexible and functional, albeit non-
formal, approach to decoding DVs that relied on his extensive knowledge of the
pandemic. He struggled with mathematical concepts such as per 100 000 and
logarithmic scales. He was able to see how events were reflected in DVs, such as
vaccinations and holidays, and saw how DVs can inform action. He could be
critical and reflective on some issues when prompted. The second participant,
Bea, was an experienced reader of journalistic COVID-19 DVs, even though she
had little experience with DVs before the pandemic. Although she considered
herself weak in mathematics, she had elaborate knowledge of many relevant
mathematical concepts. Among others, she was flexible and curious towards
logarithmic scales, which was an unknow concept to her. She was quick and
flexible in navigating the page and frequently used interactive features to
personalize the content. She could monitor the quality of her own interpretations
and could resolve a contradiction in her own interpretation. She was critical and
reflective of the DVs and their content.
77
Discussion: This study has documented hurdles and opportunities associated
with reading journalistic DVs; and even though there were only two participants,
it illustrates some of the variability in how adults approach journalistic DVs in
everyday life. Abe avoided DVs in everyday situations and reading them was
therefore not part of his everyday mathematics, whereas for Bea, the pandemic
prompted her to learn how to read DVs and she had developed a set of strategies
that allowed her to quickly and flexibly read DVs. Therefore, this study provides
evidence that the interaction between reader and journalistic DVs can constitute a
zone of proximal development (Abtahi et al., 2017; Vygotsky, 1978) in which the
newspaper, sometimes together with elaborations, constitute the more
knowledgeable other. It also shows that some adults are excluded from
participating in news discourse involving DVs and numbers. This constitutes a
democratic challenge and raises the question of how more people can be invited
to participate. This paper also contributes to theoretical development because it
has demonstrated that the three aspects of VNL mutually support and reinforce
one another, contrary to earlier assertions (Hasan, 1996; Tønnessen, 2020).
78
79
6 Discussion
The aim of this project was to contribute with theoretical and empirical
perspectives on the roles of journalistic data visualizations (DVs) as
mathematical entities in peoples’ everyday lives. As such, journalistic DVs carry
implications for everyday mathematics, mathematical literacy, and broader social
processes of mathematization. The overarching research question was:
What do DVs in journalistic media imply for readers, from perspectives of
mathematical literacy, everyday mathematics, and mathematization as social
process?
In Section 6.1, I focus on mathematical literacy, and in Section 6.2, I focus on
everyday mathematics. In Section 6.3, the findings from this dissertation are
discussed and interpreted in light of Giddens’ four dialectics of late modernity
(Giddens, 1990). Then, in Section 6.4, the interpretations from Section 6.3 are
discussed in light of the literature on mathematization as a social process.
Because there are overlaps between the three themes, some findings will be
discussed in two or three sections. Finally, in Section 6.5, I discuss the
overarching connection between the three perspectives of mathematical literacy,
everyday mathematics and mathematization as a social process.
6.1 Discussion of results in light of the concept of mathematical literacy
In this dissertation, a focal point is visual-numeric literacy (VNL). This concept
is regarded as a specific form of mathematical literacy, related to reading and
making sense of DVs. A recurring theme in the literature on mathematical
literacy related to DVs is the multiplicity of conceptual and theoretical
frameworks. They vary in the theoretical and ontological approach, where some
take a cognitive approach (e.g., Friel et al., 2001; Shah & Hoeffner, 2002), others
take a dialectical approach (e.g., Roth, 2003; Vos & Frejd, 2020), and yet others
take a semiotic approach (e.g., Tønnessen, 2020). Further, the frameworks vary
in the aspects they highlight. Most frameworks include an aspect of knowing
how to extract information from DVs. Other common aspects include the ability
to contextualize information from DVs, use DVs for informing actions and
judgements, critique, reflection, and so forth. The framework of literacy used and
developed in this dissertation, VNL, has a broad scope in the sense that its three
aspects cover many of the themes highlighted in the reviewed literature. A
finding that emerged from in-depth interviews in Study B was that the three
aspects (recognition, action and reflection) mutually support one another when
80
adults read and make sense of DVs, and the two relevant contexts – the context
of the sign system and the social context – were both used to make sense of one
another. This is not documented in the literature before and is therefore a
contribution to new knowledge. This finding implies that a broad-scoped
framework, that allows the researcher to look beyond the cognitive aspects of
meaning-making is needed for grasping the complexities of making sense of
socially situated DVs. It also indicates that VNL can serve as a fruitful
framework for further fine-grained research on the complexities of how people
read and make sense of DVs in various social contexts. Such fine-grained
analysis cannot substitute the insights from large-scale surveys, so these two
approaches should be used and developed as complementary and not competing
research strategies (Carmi et al., 2020). I also expect that Hasan’s (1996, 2003)
three aspects of literacy, on which the concept of VNL is built, can be applied to
other forms of mathematical literacy in everyday life.
In the interviews in Paper IV, it was observed that the participants’ VNL was
intertwined with related literacies such as digital literacy (e.g., knowing how to
use internet browsers and typical interactive features) and verbal literacy (e.g.,
making sense of verbal texts accompanying DVs). Earlier work on DV-reading
conceptualized this type of reading as a relatively isolated skill (e.g., Friel et al.,
2001). However, Gal and Geiger (2022) found that the media coverage of the
COVID-19 pandemic demanded blended knowledge that integrates different
types of mathematical and statistical knowledge with verbal literacy skills and
demographical and contextual knowledge. This aligns well with the
multiliteracy-approach developed by The New London Group (Cazden et al.,
1996). This perspective asserts that modern communication is increasingly
multimodal, globalized and occurs in multiple channels, and this complex form
of communication demands a multitude of intertwined literacies from the
reflective and participating citizen (Cazden et al., 1996). In this view,
mathematical literacy in general and VNL in particular is one of the forms of
literacy demanded for participatory and reflective citizenship. Similar
observations of blended knowledge are also found in other studies on how people
read DVs (Engebretsen, 2020; Tønnessen, 2020).
Some studies have drawn attention to the relationship between people’s
experience with school mathematics and affective aspects of relating to data and
DVs in everyday life. Kennedy and Hill (2018) suggest that formal mathematics
education can be counterproductive because it can make people unconfident and
81
anxious when engaging with numbers later in life. Although Kennedy and Hill
(2018) do not specify how formal mathematics education makes people
unconfident and anxious, other researchers have indicated that certain aspects of
how mathematics is typically taught can be drivers of anxiety and negative
attitudes towards mathematics. These aspects include teaching methods that offer
limited learning opportunities and obscure the importance and relevance of
mathematics (Andersson et al., 2015; Brown et al., 2008). Heyd-Metzuyanim et
al. (2021) found that poor confidence in one’s mathematical literacy can be a
substantial obstacle for adults’ engagement with numerical COVID-19 data in
journalistic contexts. Findings from this dissertation complement the findings
from the studies of Kennedy and Hill (2018) and Heyd-Metzuyanim et al. (2021).
The two participants interviewed in Study B, Abe and Bea, had in common that
they had negative feelings and low confidence with school mathematics.
However, their approaches to journalistic COVID-19 DVs were very different.
Abe would systematically avoid DVs and other numerical content whereas Bea
regularly used and explored journalistic COVID-19 DVs and had developed a
sophisticated VNL. There are at least two possible explanations to the
discrepancy between Bea’s negative feelings towards school mathematics and
her skillful engagement with DVs. First, it is possible that she did not consider
DV reading as a mathematical activity and therefore did not connect this skill
with her experience with school mathematics. Second, Bea’s interest in these
DVs stemmed from her need to be updated and her general interest in social and
developmental issues, which had exposed her to statistics and DVs. Thus, she
had developed her VNL despite her experience with school mathematics. These
reflections lead to two implications. First, as in the case of Abe, negative affect
towards mathematics can lead an adult to systematically avoid mathematical
content. This positions an adult in a negative spiral where few opportunities for
participating in everyday mathematical activities further hinders the development
of mathematical literacy and confidence. Second, as in the case of Bea, it is
possible to break the negative pattern when mathematics becomes a relevant and
useful instrument to participate in and make sense of society. This shows that
people can develop a functional mathematical literacy even when school
mathematics has left them with poor confidence in mathematics. A possible
explanation to Bea’s success with VNL in spite of her negative experience with
school mathematics could be that school mathematics and out-of-school
82
mathematics are different kinds of mathematics and may therefore be perceived
as disconnected practices (Lave, 1988).
Studies have found that journalistic DVs can be challenging to readers
(Engebretsen, 2020) and put high demands on them (Aguilar & Castaneda, 2021;
Kwon et al., 2021; Rubel et al., 2021). In terms of mathematical concepts, the
demands are often at the high end of the curriculum or beyond mandatory
mathematics education (Aguilar & Castaneda, 2021; Kwon et al., 2021). Study B
complements this picture by distinguishing between recognition, action and
reflection aspects in literacy. First, it showed how the recognition aspect of
literacy puts high demands on readers through the range of mathematical
concepts that are involved in contemporary journalistic DVs. Paper III describes
the complexities of decoding labels, colors, and scales within a broad range of DV
formats (e.g., bar charts, line graphs, sector diagrams, choropleth maps). The
meaning potentials include ‘how many’, ‘at what moment’, ‘for what category’
and ‘how quantities change over time’. This meaning-making required, among
other things, capabilities with relative numbers and logarithmic scales. Second,
Paper IV describes how the action and reflection aspect of VNL also puts high
demands on readers. For example, readers of the investigated journalistic
COVID-19 DVs were expected to explore DVs and make their own judgements
of the DVs’ social implications, validity and reliability. Study A adds to this
picture by documenting how the use of DVs and other data representations
(tables, verbal-numeric text) in newspaper weather forecasts (NWFs) have
changed over time. This study found that there was a general trend towards
increased use of non-verbal data representations, which indicates that VNL is an
increasingly important skill for contemporary citizenship.
According to Vos and Frejd (2020), DVs are at the same time both objects
with formal mathematical properties, and modelling tools. This duality closely
aligns with the distinction between the context of the formal sign system and the
context of the social situation, which was pointed out by Hasan (1996) and used
for the analysis of two young adults’ VNL in Study B. Here, the mathematical
properties of the DV and the DV as an object consisting of lines, shapes and
other visualizers pertain to the relevant sign systems. Likewise, the DVs’ use as
tools for modelling phenomena pertains to the social context. I will point out
similarities in the findings from these two studies regarding this distinction that
strengthen the conclusions from both. Vos and Frejd (2020) found that it is
possible for grade 8 students to appropriate and appreciate how DVs can be used
83
to model phenomena and communicate about these phenomena without fully
grasping the mathematical properties of the DVs. By learning about DVs starting
from socially relevant modelling and not from formal, mathematical properties,
the students gained a sense of relevance, purpose and motivation, which prepared
the students for learning more about the DV’s formal and abstract mathematical
properties later (Vos & Frejd, 2020). Regarding this connection between the
context of sign systems and the social context, I described in Study B how the
participants’ knowledge of the social context supported their ability to make
meaning of the sign system. For example, Abe used anecdotes from his
knowledge of the state of the pandemic to support his meaning-making and give
contextual meaning to the numerical information when his understanding of
formal concepts, such as relative numbers, was lacking. This approach led him to
some peculiar conclusions. For example, he claimed that it is nonsensical to state
the number of infected per 100 000 in a municipality with less than 100 000
inhabitants. But it also enabled him to informally make sense of several DVs and
related mathematical concepts, such as seven-day averages in death rates. Thus,
knowledge of the social context was a supportive vehicle for making sense of
DVs, and therefore also a driver for developing his VNL. Abe’s difficulties with
making sense of concepts such as logarithmic axes and relative numbers showed
that, if he were to further develop his VNL, he could benefit from guidance in
some form. Likewise, Engebretsen (2020) found that less common DV formats
are less readily understood, which indicates that people need a lot of exposure
and experience before they can effectively and confidently make sense of DVs.
Thus, VNL can be developed through everyday experiences with personally
relevant DVs, and some form of guidance from a more knowledgeable other (in
the form of teaching, a written or spoken guide for readers, etc.) (Abtahi et al.,
2017; Bonk & Kim, 1998; Vygotsky, 1978) is a way to accelerate the learning
process.
In closing this discussion, I will also make a comment regarding reflection
literacy. It should be noted that terms like ‘reflective’ and ‘critical’ are used with
different meanings in the literature (Weiland, 2017). In Hasan’s (1996, 2003) use
of the term, ‘reflection literacy’ refers to literacy “as a potent instrument of social
formation” (Hasan, 2003, p. 446) and entails skills such as enquiry, reflection,
and critique (Hasan, 1996). This is a form of literacy that goes beyond mere
recognition of established rules and conventions of signs and sign use, and the
use of literacy for goal-oriented action. In the study of Abe and Bea within Study
84
B, I observed that they were both able to be critical and enquiring of DVs, and
Bea was critical and enquiring towards her own interpretations of the DVs.
Despite shortcomings in their understanding of the mathematical aspects of the
DVs, they were able to critique the socio-political role of DVs. This contrasts
with Tønnessen’s (2020) findings where the participants barely showed signs of
reflection literacy. The difference may be due to the school context in
Tønnessen’s study, whereas Abe and Bea were adults who had left school life
behind them.
The main lessons from this discussion concern the ways that mathematical
literacy is intertwined with various factors. It is deeply intertwined with the sign
systems involved, which in the context of VNL is the sign systems of DVs. It is
also intertwined with the social context, as for example the COVID-19 pandemic
or the school setting. Further, it is intertwined with peoples’ identity, history,
interests, and aspirations. Yet further, it is intertwined with itself – the different
aspects (recognition, action and reflection literacy) and contexts (the social
context and the context of the sign system) of VNL are intertwined with one
another and mutually reinforcing. Therefore, mathematical literacy in general and
VNL in particular are best viewed as complex, context-dependent phenomena
that can only be properly understood when they are investigated in rich, holistic
contexts.
6.2 Discussion of results in light of the concept of everyday
mathematics
Another focal point in this dissertation is everyday mathematics. This concept is
described as the mathematics that people use throughout life. Salient themes that
emerged in the literature review on everyday mathematics included the limited
and troublesome transfer or recontextualization of mathematical skills and
knowledge between school and out-of-school activities (Carraher & Schliemann,
2002; Lave, 1988; Presmeg, 2007; Wake, 2014); qualitative changes over time
and between generations concerning how mathematical skills are valued
(Jorgensen Zevenbergen, 2011); the intimate connection between mathematical
knowledge and situational circumstances (Civil, 2002; Roth, 2003, 2014); and
the breadth of mathematical practices in out-of-school activities that range from
mundane estimation in grocery shopping to advanced mathematical modelling in
various workplaces (Frejd & Bergsten, 2016; Gainsburg, 2006; Lave, 1988).
85
Most studies on everyday mathematics have focused on activities in specific
moments in time and therefore could not indicate historical changes over time.
However, by studying generational differences within a workplace, Jorgensen
Zevenbergen (2011) described some differences in how different age cohorts
approach mathematics in everyday life, where people from the Baby Boomer
generation (those born in the first two decades after the second world war) prize
accurate mental calculations, whereas the Millennials (those born in the 1980s
and early 1990s) value estimation and problem solving, and willingly defer
calculation tasks to technology. Study A has added to the picture on historical
changes in everyday mathematics by documenting how weather forecasts have
changed over time. Early forecasts would typically present the forecast as a
ready-made, verbal interpretation, and would often involve scientific jargon such
as air streams and fronts that could function to make the forecast trustworthy.
When maps appear in early forecasts, from the 1950s and 1960s, the maps would
typically include several scientific symbols that indirectly inform the reader of
how the weather will be. These symbols include fronts, isobars and air streams.
Later weather forecasts would present the forecast in the form of extensive tables
and maps with little or no text. This shift implied that the work of organizing and
interpreting the information was moved to the reader. Later weather maps only
included symbols that more directly inform the reader about the predicted
weather, such as icons showing how much precipitation and what temperature
the reader could expect. These shifts indicate that everyday mathematics changed
its character. In the period when people routinely and tediously made mental
calculations based on drilled algorithms (Jorgensen Zevenbergen, 2011), they
also received ready-interpreted verbal messages of the weather forecast telling
them how the weather would become. For the subsequent generation of readers,
those who valued problem solving, the weather forecasts offered them ways to
organize and interpret information from tables and maps themselves. Further, the
more recent weather forecasts were more visually appealing and addressed more
the varied needs and desires of readers. On the one hand, these changes made the
forecasts more directly applicable to the readers’ lives by presenting only the
most immediately relevant meteorological information. On the other hand, these
changes contributed to obscuring meteorological science, thereby increasing the
gap between everyday mathematics and the scientific mathematics needed for the
production of weather predictions.
86
Several studies on everyday mathematics have documented a lack of
connection between school-learned mathematical skills and out-of-school
mathematical practices (Carraher & Schliemann, 2002; Lave, 1988; Presmeg,
2007). The cases of Abe and Bea confirm that school-learned mathematics can be
poorly connected to how people engage with mathematics in an out-of-school
context. In the case of Bea, it appears that her DV reading skills were mainly
developed through her personal interest in social situations described through
DVs, despite poor experiences with school mathematics. In the case of Abe, it
appeared that his poor experiences with school mathematics had the effect of
prompting him to avoid numerical and mathematical content in everyday life.
However, these connections with school mathematics remain speculative (cf.
Greiffenhagen & Sharrock, 2008), and the issue of how mathematical school
learning relates to mathematical out-of-school learning needs further research. In
light of the findings from this dissertation and Kennedy and Hill (2018)’s
contributions towards the role of emotions in the sociology of numbers, a focus
on emotions can be a pathway towards a better understanding of this complex
issue.
A recurring claim in studies on everyday mathematics in diverse contexts is
that knowledge, skills and practices are bound up in the concrete circumstances
in which they are used and developed (Carraher & Schliemann, 2002; Lave,
1988; Roth, 2003, 2014). The key role of the person’s familiarity with the social
context for making sense of DVs, and hence dealing with everyday mathematics,
was observed in Study B. For example, it was observed when Abe used his
knowledge and lived experience of the pandemic to make sense of COVID-19
DVs and validate his sense-making. Further research on how the social context
and the sign system context interact in everyday mathematics may shed further
light on how people can recontextualize their VNL across social contexts.
In terms of everyday mathematics, the findings from this dissertation indicate
that (1) journalism has become more mathematical, in the form of increased use
of DVs, tables, statistics and numerical information; (2) the mathematics in
journalism today is presented in more visually attractive and user-friendly ways
(e.g., through a functional use of colors and icons), and, mediated by digital
technology, it is more interactive; and (3) the underlying mathematical models
are often less visible or completely invisible. There could be a tradeoff between
these different developments. To make an increasing amount of data accessible
to diverse audiences, and in order to make these data visually attractive and user
87
friendly, it is necessary to limit the presentation to the most relevant aspects. In
many cases, the most relevant aspects are ready-made interpretations of the data
and just enough background information (data, models, methods) that the reader
feels that this information is trustworthy. Further, this dissertation has added new
evidence to the complex relationship between school mathematics and everyday
mathematics; Study B showed that negative experiences with school mathematics
can prompt people to self-exclude from everyday mathematics. This suggests that
there is more to be gained from developing educational approaches that foster
positive emotions towards mathematics than approaches that push more formal
mathematical learning in school. One such approach is suggested by the case of
Bea, and confirms Vos and Frejd (2020), namely that students can learn in school
that DVs can be tools for better understanding the world around us. This way,
school mathematics can move one step closer towards the goal of better
supporting people to participate in mathematical activities in diverse contexts
through life and thereby engage in lifelong mathematical learning as informed,
active and reflective citizens. Further, the findings from Study B indicate that
certain concepts, such as logarithmic scales, are difficult to learn without
guidance. Therefore, another key role of schools is to support students to learn
about mathematical concepts that are likely to become useful in everyday life and
are less likely to be learned in out-of-school life.
6.3 Main findings discussed in light of Giddens’ framework of late
modernity
As the literature review on mathematization and its implications has shown,
processes of mathematization have occurred in many aspects of society, and the
implications are branching in many directions. Because the implications differ
between fields and historical periods (Ferreira & Silva, 2020), it is useful to
understand processes of mathematization today in light of a framework of
contemporary society. As mentioned in the theory section, Giddens’ (1990) four
dialectics of late modernity offer such a framework, and this framework also has
several parallels with the literature on mathematization as a social process.
Therefore, I will in this section interpret the main findings from this dissertation
in light of Giddens’ framework. This will function as a starting point for the
discussion on mathematization as a social process and its implications, which is
presented in Section 6.4.
88
6.3.1 Displacement and reembedding
A characteristic of late modernity is globalization and the emergence of
technologies, such as the internet, telephones and television, that blur
geographical boundaries and restructures social networks. Giddens (1990) calls
this the dialectic of displacement and reembedding.
NWFs, which were the focus in Study A, embody this dialectic. On the one
hand, NWFs changed from being very local, typically narrating the forecast for
the most densely populated areas of Norway in the 1940s and 1950s, to
visualizing in tables and maps European and global weather predictions in the
decades after 2000. Hence, as weather prediction became a global collaboration
in the 1990s (Bauer et al., 2015), so did NWFs become globalized. On the one
hand, this process displaces the phenomenon of the weather –readers can check
the outlook for New York and Beijing while also checking if it will rain
tomorrow in their hometown. On the other hand, this displacement also brings
distant locations closer and opens new ways of perceiving the weather. In
particular, through the use of maps, NWFs elucidate how all weather on the
globe is part of the same system; they illustrate how the technologies of
monitoring and modelling the atmosphere now span the whole globe; and they
enable readers to check the weather in distant locations. Thus, the very
globalization of NWFs also re-embeds people and cultures.
Study B on COVID-19 DVs also resonates with the dialectic of displacement
and reembedding. The spread of the novel coronavirus was closely monitored
and modelled through extensive and detailed reports on the number of deaths,
infections and geographical distribution, as well as infection trajectories and
human interaction (Kucharski et al., 2020). Information on the spread was
conveyed to the public through extensive media coverage, which included the
use of DVs. The use of DVs in the media to describe the pandemic was a global
phenomenon. In many countries, global choropleth maps (e.g., Figure 1 in
Section 1.1) and other DVs played a key role in conveying the global nature of
the pandemic and assisted in making and validating policies to manage its spread
(Jablonka & Bergsten, 2021). At the same time, the very global nature of the
pandemic meant that mitigation measures and management experiences could be
exchanged across borders – experiments on the effects and perceptions of social
distancing from the US could be applied elsewhere (Li & Molder, 2021) and
vaccinations developed in Germany were applicable globally. Hence, the
commercial and social networks that made the pandemic a global phenomenon
89
were also instrumental in re-embedding solutions across borders. This endeavor
was also monitored and modelled, and vaccination statistics were soon published
in news media through DVs. In the DVs investigated in this study, DVs on
vaccinations were published together with DVs comparing serious illness
between vaccinated and unvaccinated groups and in different age cohorts,
thereby visualizing the effect of the vaccination programs. It was observed in
Paper III that innovative DV designs were exchanged across news outlets in
different countries. This shows that the COVID-19 DVs themselves were global
and were displaced and reembedded in new locations. As DVs became global,
VNL became a skill that could be displaced and reembedded across the globe.
6.3.2 Intimacy and impersonality
Giddens (1990) next posits the dialectic of intimacy and impersonality. On the
one hand, modernity is characterized by frequent impersonal and superficial
encounters with people, such as in economic exchanges. These encounters are a
direct outcome of displacement mechanisms. On the other hand, modernity has
also opened up new ways of being intimate and nurturing one’s identity, such as
social media communities where people with similar interests can connect.
The NWFs that were analyzed in Study A underwent changes in terms of the
relationships that were constructed between sender and reader. The analysis of
the forecasts indicates that in the period 1945-1995, the sender mostly took an
expert identity which constructed a professional, formal and top-down
relationship with the reader. However, there were a few scattered instances
before 1970 where the sender used a conversational style where the sender-reader
relationship was that of a friendship between equal parts. In Paper II, I have
argued that the end of this friendly relationship was partly due to the
implementation of numerical and computerized methods in day-to-day weather
prediction. As the methods used for making weather predictions were
mathematized, trust was vested in the methods and not in the presenter.
Therefore, the need for a friendly presenter who could explain, in vernacular
terms, how the predictions were created, diminished, and the sender reverted to a
scientific identity. This shift is connected to the increased use of maps. From
around the year 2000, the analysis of the forecasts indicates that the sender-
reader relationship changed again. This time, the sender started to take an
identity that mixed characteristics of advertising and science to produce a form of
scientific salesperson, who relates to the reader in a bottom-up fashion by
90
catering to their needs and desires. Both advertising style and scientific style
appear as impersonal sender identities, which stands in contrast to the intimate
and friendly conversational style that was identified in scattered examples before
1970. In this period, the data were mostly presented in tables with precise data
for particular locations. This format afforded readers to personalize the forecast
by locating the most personally relevant information in the tables. This suggests
that the mathematization of meteorology yielded a more impersonal relationship
between the newspaper weathercaster and the reader. At the same time, the
increased use of tables afforded readers to personalize the information in the
forecasts. This way, the forecasts could better function to find relevant
information about the weather, which can enable people to plan intimate
encounters or see what the weather will be in places where friends and family
live. Hence, the impersonal and intimate aspects of NWFs are intertwined.
The dialectic of intimacy and impersonality was also relevant in the context
of COVID-19. On the one hand, the high risks and strict mitigation measures
forced people to stay at home for long periods of time. This situation put many
intimate relationships on trial, for example between younger generations and
elderly relatives at risk of complications from COVID-19 infection. On the other
hand, new arenas for being social emerged such as family gatherings on
WhatsApp and outdoor walks with appropriate social distance (Deejay & Henne,
2024; Skinner, 2020). The role of DVs in the dialectic of intimacy and
impersonality was witnessed in the interviews in Study B. One the one hand,
both participants remark that the DVs and the webpage presenting the DVs are
impersonal, only showing numbers and not how people experienced the
pandemic. Further, Bea considers the page as having authority as it demonstrates
extensive knowledge and control of the pandemic. On the other hand, the DVs
helped the participants to understand the severity of the pandemic, offered a
feeling of control and power, and had real implications for how they should
manage social interactions, for example when visiting elderly relatives. Thus,
while the DVs were experienced as impersonal and abstracting people’s
experiences, they were also instrumental in understanding the severity of the
pandemic and informing how intimate relationships could be managed.
6.3.3 Expertise and reappropriation
Gidden’s (1990) third dialectic, expertise and reappropriation, is of particular
interest to the notion of mathematization as a social process because it deals with
91
the distribution of knowledge, expertise and learning in late modernity. The core
idea is that late modern society is immersed in abstract systems that demand high
levels of expertise to master. As people interact with these systems, they may
appropriate some level of insight into them, or altogether outsource the handling
of these systems to other people or automated machinery. Because insight into
these systems is often limited to a small group of experts, relying on them is a
matter of trust – the non-expert must trust that the expert makes appropriate
decisions or that the machine functions as intended. Examples of expert systems
include weather forecasting, epidemiology and journalism, all of which have
become mathematized during and after the second world war (Bauer et al., 2015;
Coddington, 2015; Davey Smith, 2019). Because these expert systems have
implications for most people today, a combination of trust and knowledge of
them is needed.
In Study A, it was possible to track long-term changes in the ways that
weather predictions were conveyed to lay audiences, in other words, how
messages from the expert system of weather prediction were conveyed to non-
experts. Hence, this study enabled me to investigate, at different historical
moments, what insights into the expert system were provided to readers and how
trust was established. One finding was that until ca. 1970, when numerical
weather forecasting was still an experimental science that had not yet been
implemented in daily weather prediction (Nilsen & Vollset, 2016), it was
common practice to provide insight into the meteorological observations that
were underlying the predictions, often phrased as causal events. For example, in
1945, a weather forecast from Aftenposten Morgen stated “A weak, southerly to
southwesterly air stream is approaching South Norway and will yield relatively
mild weather in the east”. This statement could prompt readers to mentally
visualize the geographical situation. It connected an observed meteorological
phenomenon (the air stream) as a cause to a prediction (mild weather) grounded
in meteorological science. This made the ensuing predictions, which are
consistent with “mild weather”, plausible and trustworthy. Further, it gave
readers a rudimentary insight into how the expert system of weather prediction
worked. This dynamic changed in 1970, at the time when numerical and
computerized weather prediction was implemented in operational weather
forecasting and started to replace manual methods (Kristiansen, 2017; Nilsen &
Vollset, 2016). From 1970 onwards, weather predictions were presented without
any causal explanations or background information. Even in the weather maps,
92
where it was possible to visualize these phenomena using symbols for pressure,
fronts and so forth, it was rare to see other information than temperature, cloud
coverage, precipitation and wind. In terms of Giddens’ (1990) framework, this
had two implications for the readers. First, the knowledge of the expert system of
meteorology needed to make sense of the forecasts was reduced as knowledge of
concepts such as air streams and barometric pressure were no longer needed.
Second, it meant that readers had to vest more trust in the abstract system as it
was increasingly opaque.
A second change that occurred in NWFs in terms of expertise and
reappropriation concerns the repertoire of signs used to describe the weather. As
indicated in Figure 3 (see Section 2.4.2), the signs for cloud coverage and
precipitation evolved from abstract symbols that needed a legend, to self-
explanatory iconic symbols. The symbols used in the early weather forecasts
were the same symbols as those used by meteorological scientists. The okta used
to represent cloud coverage in the examples from 1950 has a very precise
numerical meaning, expressed visually as a fraction of a circle. By contrast, in
the examples from 2000 and 2015, this visual fraction is replaced with icons of
clouds and the sun. Here, the scale is still ordinal (more-less cloudy), but the
numerical aspect is no longer visually evident to the reader. Instead, readers see
instantly recognizable clouds and suns, juxtaposed with raindrops or snow
crystals. Hence, the symbolic scale for cloud coverage transformed from
symbolically numerical-ordinal to visually ordinal-iconic.
In Study B, it was also possible to see how the dialectic of expertise and
reappropriation unfolded in relation to DVs. Bea, who paid close attention to
COVID-19 DVs and statistics, expressed that these DVs played little role in
shaping her everyday decisions. Rather, insofar that the pandemic affected her
life it was through policies and not her own interpretations of the data. However,
she could use the DVs to critically reflect on these policies, and even though she
had limited knowledge of the mathematical models and statistical procedures
underlying the data, she was able to reflect on the limitations and validity of the
data. Thus, to her, the COVID-19 DVs were part of a larger ecology, and Bea
was able to be reflective of this ecology and the role of mathematical models
there. Abe generally avoided DVs, likely due to poor confidence in mathematics,
but he was able to stay informed about the pandemic and was also aware of how
data were limited. Returning to Giddens’ (1990) dialectic, the pandemic with its
high stakes and mathematized means of monitoring, had multiple consequences.
93
The mathematized media coverage excluded some people, but less mathematical
alternatives were available, for instance in narrative formats. In the media
coverage, the mathematics and statistics underlying the DVs were semi-
transparent. This gave readers some opportunity to reflect on the validity of the
data and the validity of relevant political decisions based on these data and
provided some insights into the expert system of mathematical epidemiology.
Another finding from Study B was that even though some mathematical
models were semi-transparent, most of the time the numbers were presented
without detailed information about how they were compiled, and often the
numbers were only indirectly presented through color codes.
When the findings from Study A and Study B are compared, a pattern
emerges. In the NWFs, it was observed that the underlying mathematical models
became less visible while weather prediction became increasingly mathematized
and graphical. In Study B, however, the mathematical models underlying
COVID-19 data and extrapolations were semi-transparent. When underlying
models are made explicit, reflection literacy is invited. But why is the visibility
of mathematical models so different in these two cases? I suggest two key
reasons for this, both based on the works of Giddens (1984, 1990). The first
reason pertains to risk and accountability. In the case of weather forecasting, the
risks for an individual varies. Severe weather can negatively affect certain
weather dependent industries such as fisheries, farming and transportation. Many
other activities are relatively unaffected by regular weather fluctuations. Weather
forecasters are generally not held responsible for unfavorable weather. Due to the
uncertain nature of weather prediction, flawed forecasts are bound to appear.
When a weather forecast failed to report a severe storm in 2014, with dire
consequences, the forecaster acknowledged responsibility but added that similar
flawed forecasts will happen again even though they will learn as much as they
can from the mistake (Compton, 2018). Hence, even when flawed weather
forecasts have negative consequences for people, the accountability of the
forecasters is diminished by the inherent uncertainty of weather prediction.
COVID-19, on the other hand, involved high risks at a global scale and its spread
and severity could to a large extent be controlled politically through mitigation
measures (limiting global trade, social distancing, hygiene, etc.) and the
development and dissemination of vaccines. Due to the higher risks and greater
degree of controllability, it was much more important to establish trust in
COVID-19 data, and consequently, it was necessary to be more transparent about
94
how data were generated and how data informed policy. The second reason
pertains to the role of routine in peoples’ lives and the constitution of society.
Well-established institutions that reliably produce the same output over long
stretches of time, such as meteorological offices, become part of people’s
everyday routines and contribute to what Giddens (1984) calls ontological
security, that is, the sense that the individuals’ surroundings are stable and
reliable. As long as these institutions go on in the same predictable pattern, the
need to re-establish trust diminishes. For this reason, there is no need to
continually convince the audience that weather predictions are made on sound
methods. However, in highly unpredictable situations, especially ones involving
high risks and with a need to change people’s behavior, this sense of ontological
security is disrupted – the familiar conditions for life that we have today may be
fundamentally altered tomorrow. This is what happened during the COVID-19
pandemic, where well-known daily routines of commuting to work, grocery
shopping and social activity were interrupted by invasive policies. Under such
circumstances, it is crucial to re-establish conditions for trust. This was partly
achieved by making the models underlying the data reported in journalistic
COVID-19 DVs semi-transparent.
6.3.4 Privatism and engagement
Situations with high-stakes risks can put individuals in dilemmas where they
must weigh personal interests against collective interests. This dilemma is at the
heart of Giddens’ (1990) dialectic of privatism and engagement. In the case of
the NWFs of Study A, reading them is for the most part a routine and mundane
activity with low stakes. Still, they are read because they have implications for
our lives. The journalistic COVID-19 DVs in Study B, however, described an
urgent, high-stakes global pandemic with profound implications, fears,
engagement and emotions. Although Bea did not like mathematics, she had spent
considerable time and effort engaging with COVID-19 DVs. Abe, who avoided
DVs, was nevertheless aware of the urgency of the situation and knew a lot about
the impact of the pandemic on his own and other people’s lives. Further, both
Abe and Bea indicated that the dialectic of privatism and engagement was
relevant to their interpretation of the COVID-19 DVs. Abe pointed out a tension
between numbers and mitigation measures. At the time of the interviews,
infection numbers were rising in Kristiansand where he lived. The rest of the
country was in lockdown (according to him), but he could still go to the gym
95
every day, potentially spreading the virus. Interestingly, he did not scrutinize or
critique this situation further, but tacitly accepted the tension between reduced
mitigation measures and rising infection numbers. Bea reflected that the sheer
amount and richness of data on the page is more than anyone can consume. This
way, the authors of the page were showing off their excessive knowledge of the
situation, which positioned them as authorities on the matter. Thus, both
participants indicated that the page conveyed a sense of power and control that
left them as passive spectators who had to accept current policies.
Journalistic DVs with low-stakes such as NWFs and high-stakes such as
COVID-19 DVs are read because they are relevant to our lives. They can elicit
emotions, engagement and reflection. An urgent situation can even prompt
people who otherwise have low confidence in mathematics to engage with DVs,
and knowledge obtained from DVs can create awareness of and compassion for
impacted people. Nevertheless, journalistic DVs can also leave people feeling
powerless and unable to influence policy. Hence, the reflexive capacity to
contrast information from DVs with relevant policies and use this to take a stance
on collective decision making is a skill that may need more attention from
researchers and educators. This skill is relevant to democratic participation.
Journalistic DVs act as mediators between scientific and political institutions
(e.g., weather forecasters, pandemic modelers, policy makers), and lay readers.
This mediation crystallizes the tensions between the collective and the individual.
On the one hand, journalists provide information with implications for people’s
individual lives. On the other hand, the same information has implications for
policy making. This juxtaposition gives journalists opportunities to inform
people, justify and explain policies, and expose tensions between policy and data.
In the journalistic COVID-19 DVs researched in this dissertation, it seems that
one way of realizing these opportunities was by allowing the reader to explore
the data themselves and make their own conclusions.
6.3.5 Reflections on using Giddens’ framework to research mathematization
as a social process
The purpose of Section 6.3 was to interpret the findings from this dissertation in
a sociological framework. I chose to use Giddens’ (1990) four dialectics of
phenomenology in late modernity for this analysis for two reasons. First, because
it is a relevant and holistic perspective on contemporary society, and second, as
demonstrated in the theory section (Section 3.2.1), because it aligns well with the
96
themes brought up in the literature on mathematization as a social process, and
can therefore offer a complementary perspective to the existing research on
mathematization as a social process. The above interpretation shows that this
framework can be used to connect the use of DVs in settings such as NWFs and
journalistic COVID-19 DVs, to macro-structures such as the mathematization of
meteorology, the institutionalization of weather forecasting, and the abrupt onset
of COVID-19. Therefore, I suggest that this framework can also be useful for
further research on the role of mathematics in society and the implications, good
or bad, of processes of mathematization. This leads to the next part of the
discussion chapter, where the findings from this dissertation are discussed in
light of the literature on mathematization as a social process.
6.4 Discussion of the results in light of the concept of mathematization
as a social process
The final lens that I will view my findings through is mathematization as a social
process, which is a systems perspective on the role of mathematics in society. I
will use the four dialectics from Giddens’ (1990) as an organizing principle in
this part of the discussion (Sections 6.4.1–6.4.4) because these four dialectics
align well with the main themes brought up in relation to mathematization as a
social process, and because these dialectics offer a novel way of understanding
the role of mathematization in society that can complement other approaches to
this issue (e.g., Ferreira & Silva, 2019; Keitel, 1989). Finally, in Section 6.4.5, I
will look back at the findings and discussion in light of the dialectic of
mathematization and dematematization (Chevallard, 1989; Jablonka & Gellert,
2007; Keitel, 1989).
6.4.1 Mathematization in the context of displacement and reembedding
DVs from both study A and B show aspects of globalization. In Study A, the
more recent NWFs show weather information for major cities in all continents
through tables and maps. This shows, first, that global weather data is available
for weather forecasters, and second that weather forecasters expect that some
readers are interested in this global information. The first point is enabled by
global data exchange and collaboration between meteorological and weather
forecasting institutes, and global standardization of weather models and weather
data collection methods (Bauer et al., 2015; Benjamin et al., 2019). The second
point regarding readers being interested in global weather information, is not, to
97
my knowledge, mentioned in the literature relevant to mathematization as a
social process. Maps and tables afford publishing abundant information where it
is possible to look up data for individual locations (Few, 2012). Hence, these
formats enabled newspapers to make this data available to as many people as
possible. The NWFs act as interfaces between meteorological science and lay
readers, where journalistic expertise bridges these domains.
COVID-19 was a global phenomenon, which was evident in DVs such as
Figure 1 that show the global distribution of COVID-19 related deaths. Although
much of the modelling was done by experts using advanced statistical models,
DVs served in making these statistics accessible to lay audiences. Because the
pandemic was a global crisis, the methods used for monitoring and managing
could be applied everywhere; these were displaced and created opportunities for
re-embedding.
The journalistic DVs used during COVID-19 were also displaced and
reembedded. For example, a DV known as “flatten-the-curve” was widely used
across the globe (Heyd-Metzuyanim et al., 2021; Li & Molder, 2021). Hence,
both the COVID-19 virus, the data and models used to monitor its spread and the
vaccines and legislation used to manage it, and the DVs used to convey
information about it, were global phenomena.
Although globalization is not commonly mentioned in relation to
mathematization, formalization has been brought up by several authors (Davis &
Hersh, 1986; Jablonka & Gellert, 2007; Keitel, 1989). The formalization of
mathematical procedures, highlighting form at the expense of transparency or
content, has been pointed out as a prerequisite for these procedures to be
implemented at a large scale. For instance, the formalization of data is a
prerequisite for Big Data analyses (Mayer-Schönberger & Cukier, 2013), and
both atmospheric and epidemiological models run on large swaths of data in
standardized formats fed into algorithms. So, formalized mathematics can be
applied independently of location; it is displaced. People versed in formalized
mathematical procedures, such as weather prediction or public health monitoring
as described in paragraphs 2.4.2 and 2.4.3 of this dissertation, can in principle
apply their skills wherever there is need; these skills can be reembedded. In other
words, the general character of mathematical models makes them applicable in
multiple contexts and their formal character makes them exchangeable across
boundaries. So, in the dialectic of displacement and re-embedding, generality and
formalism are powerful features of mathematics. In terms of journalistic DVs,
98
this formalization and globalization was evident during the COVID-19 pandemic
as similar DVs were used in diverse locations such as Europe (Jablonka &
Bergsten, 2021), East Asia (Kwon et al., 2021), North America (Li & Molder,
2021) and Latin America (Aguilar & Castaneda, 2021); and it was observed in
Study 2 that DV designs were exchanged across news outlets on different
continents.
6.4.2 Mathematization in the context of intimacy and impersonality
Global and universal forms that can be re-embedded in locations distant from
their origin are characteristic of late modernity, and this is witnessed in
journalistic DVs. These general forms are formal and impersonal, but Giddens
(1990) maintains that this impersonal character of late modernity stands in a
dialectical relationship with intimacy. The findings presented in this dissertation
show that the style in NWFs changed from an intimate, personal
conversationalist to a scientist and advertiser, which made the style of
communication less personal and intimate over time. At the same time, the
information presented developed to include more locations and even common
vacation destinations which can make the NWFs more useful for planning social
encounters. The findings from this dissertation on journalistic COVID-19 DVs
show that they were mainly experienced as impersonal representations of the
state of the pandemic that convey a sense of control and authority. This affirms
key findings in the literature on mathematization as a social process, which
suggests that mathematical content in journalism can be used to create a sense of
objectivity, rationality, power and impersonality (Chassapis & Giannakopoulou,
2015; Giannakopoulou, 2021; Jablonka & Bergsten, 2021). More generally, it
affirms the claim that mathematics and statistics can act as instruments of control
in democratic societies (Rose, 1991). At the same time, it was also observed that
Abe and Bea interpreted the DVs in their personal lives and gave them a sense of
urgency, and that it was personal needs and interests that prompted Bea to
regularly visit the page during the pandemic. Hence, the impersonal aspects of
DVs and statistics go hand in hand with the personal and intimate. As Didier
(2024) concludes, data “frequently perceived as objective evidence, can at the
same time be a source of emotional engagement” (p. 1). Further, Abe’s
difficulties with interpreting DVs and his self-exclusion from DVs in journalism
indicate that the growing use of DVs in journalism contributes to create social
barriers between people with a well-developed VNL and those without. This
99
agrees well with Gingras (2001), who found that social exclusion driven by
unequal access to mathematical knowledge was one of the main consequences of
the mathematization of physics. At the same time, it shows that the impersonal
aspects of DVs in particular and mathematics in general are intertwined with
intimacy, which aligns with Giddens’ (1990) idea of the transformation of
intimacy.
6.4.3 Mathematization in the context of expertise and reappropriation
When processes of mathematization create social barriers, these processes also
create and reinforce particular divisions of labor and demands for knowledge.
Hence, the social exclusion produced by mathematization has a direct connection
to Giddens’ (1990) dialectic of expertise and reappropriation as it implies
changes for the people who are experts in mathematized practices, and the
knowledge necessary to reappropriate to productively interact with mathematized
practices. In Study A on NWFs, this was evidenced by the changing demands on
readers’ mathematical literacy. As weather prediction was mathematized, the role
of the reader changed from an interpreter of verbal text to an organizer of
geospatial data presented in tables and maps. Further, the signs used in weather
forecasts became more iconic and therefore more readily interpreted. There is
evidence suggesting that lay people have difficulties interpreting weather
forecasts but not that they have produced social exclusion (Masson & Es, 2020;
Sivle et al., 2014). Considering that weather forecasts, in newspapers and other
platforms, are intended for lay audiences, this is not surprising. Rather than using
a mathematical language that excludes large groups of the intended audience,
weather forecasters have developed ways of making the forecasts accessible and
appealing. Hence, the developments in the representation of data in NWFs
towards more iconic symbols and more extensive use of tables and maps can be
understood as developments intended to facilitate reappropriation.
In Study B, it was observed that journalistic COVID-19 DVs demanded a
complex form of mathematical literacy that was called VNL. Paper IV described
how young adults had unequal access to these DVs. Thus, the literacy necessary
for reading and making sense of COVID-19 DVs was not always reappropriated,
which could create social exclusion. The reasons and consequences for the
unequal reappropriation of VNL are still poorly understood. Some likely
mechanisms include the role of emotions and people’s background in school
100
mathematics (Kennedy & Hill, 2018), as well as interests, needs, and social
identity.
Another theme in the literature on mathematization as a social process is the
role of trust in mathematized technology. Jablonka and Gellert (2007) connect
issues of trust with opacification and dissolution of responsibility. When a
mathematized technology becomes opaque, the role of users is merely to know
when and how to use it, and to trust that the technology does what it is supposed
to do. When users have no insight into the inner workings of the machine, they
are not responsible if it turns out to be flawed. According to Davey Smith (2019),
the mathematization of epidemiology, that is, the transition in the discipline when
mathematical and statistical approaches became the norm, was the hallmark that
made the field trustworthy to the scientific community at large. In Study A, the
main finding related to trust was that meteorological explanations to weather
predictions were commonly included in the weather forecasts until 1970, at
which point this practice suddenly stopped. The end of this practice coincides
with the implementation of computerized mathematical methods in weather
prediction. In light of Giddens (1990), I interpret this to mean that the
mathematization of weather prediction dissolved the need to routinely re-
establish the trustworthiness of their methods. In Study B, it was observed that
the mathematical and statistical methods were semi-transparent as methods for
data collections, data manipulation and modelling as well as an error list were
made available to readers. Thus, the findings from this dissertation paints a more
nuanced picture than what the literature on the implications of mathematization
suggests. In some situations, mathematized expert systems do indeed induce trust
among non-experts, and this trust is vested in the mathematical models which are
inaccessible and therefore cannot be critically scrutinized. In other situations, the
need emerges to establish trust through critical scrutiny of mathematical models,
and consequently, critical discussions of the validity and reliability of
mathematical models surface in the public, as happened during COVID-19 (Gal
& Geiger, 2022). Based on Giddens (1984, 1990) perspective, I suggest that this
need for public critical scrutiny of epidemiological mathematical models is an
expression of a need for greater trustworthiness due to a high-risk global disaster
that asked for a sudden change in everyday routines. As this kind of critical
scrutiny invites reflection literacy, this has implications for the kind of
mathematical literacy that is expected of lay audiences and that compulsory
education should prepare students for.
101
The final theme that I will discuss in relation to the dialectic of expertise and
reappropriation is the role of schools and formal education. As Study B on Abe
and Bea and other studies have shown (e.g., Engebretsen, 2020; Jackson et al.,
2018; Kennedy & Hill, 2018), it is possible to develop a functional everyday
mathematical literacy in adult life. However, the success varies. A key variable
determining success seems to be participation. People who participate in
everyday situations involving DVs have access to learning opportunities where
they can develop their VNL. Therefore, a key aim of school mathematics should
be to enable students to participate in everyday mathematical practices. Because
the inventory of DVs in the public sphere is rapidly evolving, it is not feasible to
teach students how to make sense of every DV format that they are likely to
encounter. Instead, it should be a goal to provide students with sufficient
experiences to participate in out-of-school society at the levels of recognition,
action and reflection literacy. Societal participation can then be a vehicle for
lifelong learning. Determining the exact content of school mathematics falls
beyond the scope of this dissertation. However, since DVs and mathematical
models play a vital role in everyday mathematics, some familiarity with the wide
array of DVs and of mathematical modelling and meta-knowledge of
mathematical models should be part of the curriculum. It seems that one of the
main obstacles for people to participate in everyday mathematics is negative
emotions such as fear, anxiety and poor confidence towards mathematics (Heyd-
Metzuyanim et al., 2021; Kennedy & Hill, 2018). Finding ways of teaching and
learning mathematics that overcomes this obstacle is one of the main challenges
to contemporary school mathematics. Some promising strategies toward this goal
include making the students aware of the importance and relevance of
mathematics (Brown et al., 2008) and finding teaching methods that provide
learning opportunities to all students (Andersson et al., 2015).
6.4.4 Mathematization in the context of privatism and engagement
The presence of high-consequence risks and opaque expert systems in late
modern societies puts individuals in dilemmas between pragmatically accepting
issues that are out of their reach, and activism for collective solutions to these
problems (Giddens, 1990). This dialectic was pertinent to the COVID-19
pandemic. This was a global crisis where any one individual had negligible
influence on the situation. They could either accept the situation and the policies
used to manage it or call for different approaches to manage the spread and
102
severity of the pandemic. The latter alternative requires reflection literacy. This
dialectic was observed in Study B as the participants mainly used politically
mandated guidelines and legislation to manage their behavior in relation to the
pandemic, and not the DVs. Even Bea, who had spent considerable time and
effort studying the DVs and had keen insights into the statistics, relied mainly on
current policies to manage her own behavior towards the pandemic. The
participants spotted tensions between the information presented in the DVs and
public mitigation measures, but these tensions were not always interrogated.
Further, and related to this, Bea remarked that the sheer amount of data was
intimidating and conveyed a sense of power and control. Thus, despite critical
reflections, the participants nevertheless pragmatically accepted the implemented
policies. This supports earlier research that suggests that mathematics can create
a persuasive truth discourse (Chassapis & Giannakopoulou, 2015; Jablonka,
2017; Jablonka & Bergsten, 2021). So, in terms of privatism and engagement, the
participants readily accepted current policies and did not call for change. While
this may suggest pragmatic acceptance of the situation, it may also indicate an
overall satisfaction with the political management of the pandemic.
6.4.5 Mathematization and demathematization
According to the literature, mathematization and demathematization pertain to
multiple issues, including knowledge and literacy, trust, technology, everyday
life, emotions, epistemology and ontology (Chevallard, 1989; Davis & Hersh,
1986; Jablonka, 2017; Jablonka & Gellert, 2007; Keitel, 1989). These issues have
complex interrelations and pertain to societal structures. Therefore, I decided to
approach this as a sociological problem through a multi-faceted framework from
Giddens (1990). The purpose of this section is to revisit the concept of
demathematization.
According to the researchers who first used this concept, the engine of
demathematization is the ‘implicitation’ of mathematics into routines, objects
and technology. Mathematics is always transformed from an explicit, visible
form to an implicit, invisible form. As this happens, mathematics becomes
trivialized, devalued and taken for granted (Chevallard, 1989; Davis & Hersh,
1986; Keitel, 1989). The findings presented in this dissertation suggest that there
are exceptions to this rule. In situations of high-consequence risks where
mathematical models are crucial for managing policy, mathematical models can
reappear in public discourse to be scrutinized.
103
The first issue I will discuss in light of the concepts of mathematization and
demathematization is whether journalistic DVs have become demathematized. In
the NWFs that were analyzed in Study A, it was observed that the NWFs
changed from narrative text and rather technical maps with information about air
pressure and wind streams, to maps and tables where the technical information
was stripped away and the reader was only presented with temperature data
(typically represented as numbers with color codes to indicate negative and
positive temperatures), wind (represented as arrows with a number inside, where
the direction of the arrow represents the direction of the wind and the number
represents wind speed), and icons that showed the predicted cloud coverage and
precipitation. This development constitutes a demathematization in two ways,
both in that the increasingly mathematical science of meteorology becomes
invisible as the technical information is stripped away, and in that numerical
information (e.g., the percentage of the sky that is covered with clouds) is
replaced by icons. Thus, it is warranted to conclude that as meteorology was
mathematized, weather forecasting was demathematized in certain ways. At the
same time, it was also observed in Study A that the NWFs studied became more
mathematical in the sense that the use of maps and tables increased. Thus, there
were traces of both mathematization and demathematization in NWFs. Similar
examples of demathematization were found in the journalistic COVID-19 DVs
analyzed in Study B, where numerical values were frequently replaced by visual
elements, such as colors and bars. For example, high infection rates were
represented as red shades on choropleth maps, indicating danger. These visuals
hid several layers of mathematization such as COVID-19 statistics and
epidemiological mathematical models. Thus, they condensed advanced
mathematics into simple and relatively accessible formats, a demathematization
of advanced mathematics. A potential explanation of the dynamics between
mathematization and demathematization in these situations is found in the
dialectic of expertise and reappropriation (Giddens, 1990). The expert systems of
meteorology and epidemiology were inaccessible to most lay users. However,
many lay agents relied on and interacted with these systems. In this everyday
interaction, NWFs and journalistic COVID-19 DVs acted as interfaces between
the expert systems and lay agents. Hence, journalistic media became a site of
reappropriation of the rudimentary skills needed to productively make sense of
these expert systems. The journalistic work of producing these DVs was a
balancing act of presenting the information in ways that were accessible, useful
104
and trustworthy, where journalism was an expert system in its own right
(Engebretsen et al., 2018). The transformation of numbers and mathematical
models into DVs and journalistic storytelling enhances readability, applicability
and “clickability”, and even enables emotional engagement (Didier, 2024). Lay
agents could reappropriate some rudimentary knowledge of the expert systems
through journalistic storytelling. In these stories, numbers continued to be salient
and therefore contributed to mathematize people’s understanding of their
environment, while much of the underlying mathematics was invisible.
The second issue I will discuss in relation to mathematization and
demathematization is the role of trust and accountability. According to Jablonka
and Gellert (2007), demathematization of mathematical technology implies that
the user must “trust the black box” and “know when and how to use it” (p. 8).
When faults occur, the technology is accountable. Does this logic hold with the
NWFs and journalistic COVID-19 DVs analyzed in this dissertation? Only
partly. In the case of NWFs, mathematical models were used to forecast natural
phenomena, and the models were uncertain. Thus, flawed forecasts with
detrimental effects were bound to occur (Compton, 2018), which means that the
forecasts could not be trusted blindly. Mathematical models used to predict the
spread of COVID-19 were also uncertain, and as they were used to inform
policy, they were also subject to a variant of Giddens’ (1984) double
hermeneutic. When mathematical models predict a severe scenario of the
pandemic, extra mitigation policies can be implemented to avoid this scenario,
which in turn leads to a different and hopefully less severe outcome. Thus, the
logic of mathematized epidemiology and the logic of policy interact to produce a
new outcome. Therefore, predictions made by mathematical models can be
regarded as “what-if”-scenarios and not actual predictions. Mitigation policies
often had upsetting implications for peoples’ lives, and semi-transparency of
epidemiological mathematical models used to inform policy was likely helpful to
explain to lay people why the policies were needed, and to make the policies
trustworthy. So, in both weather forecasting and COVID-19, mathematized
technologies were not merely black boxes to be trusted blindly, but uncertain
instruments used to inform decision making.
Lensing (2017) called for research on why mathematics in use takes the two-
sided form of mathematization and demathematization. I think that the two
concepts of mathematization and demathematization alone are not sufficient for
understanding the implications of mathematization in late modern society, but the
105
application of Giddens (1990) in this dissertation has made progress in this
regard.
6.5 The connection between mathematization as a social process,
everyday mathematics and mathematical literacy
As stated in the introduction, a core concern in this dissertation is the non-trivial
connection between processes of mathematization and the mathematical literacy
expected of citizens. In this section, I will delineate a tentative local integration
of these concepts, bridged by the concept of everyday mathematics. This
integration is illustrated in Figure 9.
Processes of mathematization change human practices and technology so that
they become more mathematical. Such processes have occurred in many
institutions, including the sciences of meteorology, epidemiology and journalism.
As these institutions change, so do the everyday practices of the people who
interact with these institutions. This is true of workers within these institutions –
meteorologists, weather forecasters, epidemiologists, health policy makers,
journalists and so forth – and of consumers – people watching the news, people
reading weather forecasts, people affected by health policy, and so forth.
Therefore, processes of mathematization affect everyday mathematics. It leads to
more mathematics in the public sphere. However, it does not always translate
into more or more complex mathematics in everyday life. Here, the distinction
between the three aspects of literacy is relevant. In terms of recognition and
action literacy, the change can go either way. For the NWFs, the development is
not straight forward. The early weather forecasts were based on written texts with
ready interpretations. When weather maps were used, the signs gradually became
more iconic. The theoretical meteorological concepts that constitute the basis of
the mathematical models of the atmosphere (barometric pressure, humidity, etc.)
disappeared, while the amount of meteorological data and the use of maps and
tables grew. Thus, the sign system faced by consumers became more
mathematical in the sense that they became more dense with numerical
information and relied more on maps and tables. However, the underlying,
mathematized science was largely shielded from the consumer. In journalistic
COVID-19 DVs, on the other hand, the sign system was quite mathematical,
involving several mathematical concepts and models (e.g., logarithms, cartesian
coordinates, relative numbers, R-number). In this case, even though the full
complexity of the mathematical models were not available, the mathematization
106
of epidemiology yielded a certain mathematical sign system that readers were
expected to understand and use. For reflection literacy, some rudimentary
knowledge of mathematical modelling was useful. While this kind of knowledge
was not overt in post-1970 NWFs, it was relatively salient in the journalistic
COVID-19 DVs, which indicates that reflection literacy was invited. I have
pointed at the role of routine and trust to explain this difference regarding the
transparency of mathematical models.
Thus, processes of mathematization change the recognition and action
literacy expected of citizens, in working life and elsewhere, but this literacy is
not necessarily more mathematical. For reflection literacy, it is necessary to have
some insight regarding the inner workings of systems – who is in charge, how are
decisions taken and whose interests are at stake. Thus, some understanding of the
mathematics involved can be instrumental.
A final remark regarding this model is that the three perspectives are mutually
dependent. On the one hand, processes of mathematization are key to
determining the form of everyday mathematics and demands for mathematical
literacy. On the other hand, people’s actual mathematical literacy and actual
engagement with mathematics in everyday life shapes the cultural-historical
processes of mathematization.
107
Figure 9: Illustration of the connection between mathematical literacy, everyday mathematics,
and mathematization as a social process.
Mathematization
Everyday
mathematics
Mathematical
literacy
108
109
7 Conclusions and recommendations
The aim of this research project was to explore the uses of data visualizations
(DVs) in journalism, both historically through a study on newspaper weather
forecasts (NWFs), and in the present through a study on journalistic COVID-19
DVs. The uses of journalistic DVs, and what this use implies for readers, was
researched through the perspectives of mathematical literacy, everyday
mathematics, and mathematization as a social process. The overarching research
question was
What do DVs in journalistic media imply for readers, from perspectives of
mathematical literacy, everyday mathematics and mathematization as social
process?
The research questions for Paper I were how has the use of semiotic resources in
newspaper weather forecasts changed from 1945 to 2015? and how do the
changes in newspaper weather forecasts change the readers’ role, and what do the
changes indicate for everyday mathematics?
For Paper II, the research question was how have the identities, relations and
discursive roles in newspaper weather forecasts changed since 1945?
For Paper III, the research question was what characterizes the visual-numeric
literacy (VNL) expected of readers of COVID-19 DVs in online news media?,
and for Paper IV the research question was what characterizes adults’ VNL when
reading and making sense of journalistic COVID-19-related DVs?
The main findings are summarized in Table 7 below.
This Conclusion chapter follows the same structure as the Introduction, with
Section 7.1 devoted to empirical conclusions and Section 7.2 devoted to
theoretical conclusions. Empirical and theoretical conclusions cannot be
separated entirely, so this distinction is not always clear-cut. In Section 7.3
reflections on the theories and methodologies used are presented, and in Section
7.4 recommendations for practice and further research are offered.
110
Finding 1
In NWFs, between 1945 and 2000, there was a gradual shift from
verbally telling readers the weather forecast to visually showing
the weather forecast through tables and maps. After 2000, the
NWFs became increasingly visually appealing and dense with
tabulated information (Paper I).
Finding 2
Between 1945 and 2015, the role of the reader of NWFs changed
from a receiver of verbally given information to an organizer and
interpreter of visual and tabulated information (Paper I). The
identity of the sender changed from a friend, via a scientist, to a
hybrid of a scientist and an advertiser (Paper II).
Finding 3
Changes in the NWFs between 1945 and 2015 can be related to
processes of mathematization in meteorology and journalism
(Paper II).
Finding 4
In 2020, journalistic DVs were used to tell the story of how the
COVID-19 pandemic spread (through maps, Sankey diagrams,
etc.), changed (through lines graphs, colors, etc.) and impacted
peoples’ health (through graphs of deaths, recoveries,
hospitalizations, etc.). DVs can be used as a storytelling device
by visually showing a lot of numeric information. They tell
stories about where, when, who, and how through a time variable
(Paper III).
Finding 5
The concept of VNL consists of three aspects (recognition, action
and reflection) and two contexts (the context of the sign system
and the social context) that together capture the capacity of
reading and making sense of DVs. The three aspects and two
contexts mutually reinforce one another and offered an
illuminating and novel way of analyzing how people read and
make sense of DVs (Paper IV).
Finding 6
The use of journalistic COVID-19 DVs in the media expected
from readers a particular form of VNL. Readers were expected to
decode a wide range of DV types (including line graphs, bar
charts, choropleth maps), mathematical concepts (including
logarithmic scales and relative numbers) and codes (including
red=bad). Further, readers were expected to connect the
visualized information to events in several ways and make
111
judgments, and to use interactive features to explore the DVs.
Also, readers were invited to critically reflect on the validity and
reliability of the data visualized, the methods used to collect these
data, and the processes of managing and visualizing the data.
Adults have different approaches to dealing with such DVs in
everyday life, such as avoiding or auto-didactically developing
sophisticated reading strategies (Paper III & IV).
Finding 7
Journalistic DVs are cultural artifacts that can be both local and
global – for example, showing weather predictions for a
particular country or COVID-19 statistics for the whole globe.
Likewise, DV formats are frequently exchanged across countries
and VNL is applicable in diverse contexts and countries. Hence,
DVs and VNL are both displaced from any particular situation
and are at the same time reembedded in situations of many kinds
(this dissertation, Chapter 6).
Finding 8
While journalistic DVs are often perceived as impersonal,
objective and abstracting the phenomena they represent, they can
also provide socially relevant information that helps people
understand and appreciate other peoples’ situations and can aid
people to manage intimate relationships. Hence, intimacy and
impersonality are intertwined in journalistic DVs (This
dissertation, Chapter 6).
Finding 9
Journalistic DVs are interfaces between expert systems and lay
users, mediated by journalists who design the DVs to be relevant
and accessible. The lay users do not need to fully understand the
expert systems involved but need some rudimentary knowledge
to make sense of the DVs. In situations where trust is at stake,
information about the underlying data infrastructure can
strengthen the trustworthiness of the data. Further, journalistic
DVs can provide insights into expert systems and thereby offer
lay users opportunities to learn about, for example, the use of
mathematical models in epidemiology (This dissertation, Chapter
6).
Finding 10
Journalistic DVs often carry information relevant to social and
political issues and can be used to critique, explain or justify
112
policies. Therefore, they can put individuals in dilemmas between
passively accepting the status quo or calling for change. The
importance and urgency of some situations, such as the COVID-
19 pandemic, can prompt people to take interest in journalistic
DVs in order to stay updated on the situation. Journalistic DVs
can elicit emotions, engagement and reflection (This dissertation,
Chapter 6).
Table 7: Summary of main findings.
7.1 Empirical conclusions and implications thereof
The empirical aims were to study how the use of journalistic DVs have changed
over time, and to study opportunities and demands associated with journalistic
DVs in current practices. Each paper yielded conclusions towards these aims
(Findings 1–4 and 6 in Table 7). Paper I showed that the representation of data in
NWFs changed between 1945 and 2020. The change can be summarized as a
shift from verbal representations (written words and numbers) towards increased
use of multimodal, visual-numeric representations (DVs, tables). This implies
that a change in the role of the reader also occurred, from an interpreter of written
messages to an organizer and interpreter of visualized and tabulated data. Paper
II, which was also based on analyses of NWFs, showed that the sender, too,
changed in this period. The changes were summarized through the lens of sender
identity. In the early years of the period, the sender mostly took the identity of a
scientist, that is, a top-down expert who uses accurate and scientific terminology.
Sometimes, a different identity appeared, the conversationalist who translated the
weather forecast into a more vernacular and accessible language. However, the
conversationalist disappeared around 1970, and the scientist again dominated.
Around 2000, the sender identity shifted to a blend of a scientist and an
advertiser. This identity is someone who retains a mastery of scientific
terminology, but at the same time ensured that the NWFs were visually appealing
and offered information that catered for needs and desires of the reader, such as
weather predictions for popular vacation destinations. These identity changes
coincided with milestones in the mathematization of weather prediction. This
indicated that the changes could have a relation to processes of mathematization
within weather prediction. However, other processes, such as innovative media
technologies, commercialization and rising levels of mathematical literacy in the
general population, can also have contributed to these developments. Therefore,
113
it seems most plausible to consider the identity changes as a consequence of
multiple, parallel societal processes where processes of mathematization (of
weather forecasting, of journalistic media, of people’s mathematical literacy,
etc.) are some among many contributing factors.
Paper III, which described characteristics of journalistic COVID-19 DVs
from the perspective of VNL, showed that the use of DVs in the media coverage
of the pandemic demanded a complex form of mathematical literacy. This
entailed that readers must (1) be familiar with a broad range of mathematical and
statistical concepts and symbols, (2) have a capacity to interact with digital media
effectively, (3) connect the statistical data with the social context they describe
and (4) use these to inform actions and judgements. I also found that the
journalistic COVID-19 DVs offered readers opportunities for reflection and
critique regarding data quality, validity and trustworthiness of the data and the
way they were represented. Paper IV described an investigation into two young
adults’ actual engagement with journalistic COVID-19 DVs. Even though the
participants had similar backgrounds and experiences with school mathematics,
their approaches to journalistic DVs were very different. One participant, Abe,
would generally avoid DVs in his everyday media life, whereas the other, Bea,
was a regular consumer of journalistic COVID-19 DVs during the pandemic.
Abe’s VNL was poorly developed, while Bea had autodidactically developed a
rich repertoire of knowledge, habits and strategies for reading and making sense
of DVs. Consequently, they had unequal access to journalistic DVs, the
information therein, the associated discourse, and political decision-making.
Nevertheless, both encountered hurdles in their reading process and both used
their knowledge of the social context to support their interpretations.
Beyond the four papers, the sociological synthesis of the findings in Chapter
6 also yielded findings (Findings 7–10 in Table 7). It showed that journalistic
DVs were aspects of processes of mathematization that were connected to
globalization through the dialectic of displacement and reembedding. DVs, DV
formats and VNL were exchanged and adapted across the globe and were in
constant development. DVs were both impersonal, as they showed seemingly
objective information, ‘hard facts’, but could also give insight into other peoples’
lives (e.g., the state of the pandemic or the expected weather in a friends’ city)
and assist decision making regarding social relationships (e.g., when should I go
and visit my family?). DVs acted as interfaces between expert systems (e.g.,
meteorology and epidemiology) and lay people, mediated by journalism which
114
acted as an expert system in its own right. Thus, DVs offered rudimentary insight
into socially relevant expert systems, sometimes including background
information about data infrastructure, so that lay people could appropriate the
knowledge needed to make this information useful and beneficial. Lastly, as
carriers of information with social and political relevance, DVs could sway
people between pragmatic acceptance of the state of the situation (e.g., accepting
that current policies are sufficient for managing the spread of COVID-19) or
engagement (e.g., calling for more equitable mitigation policies for the
management of COVID-19). Hence, DVs in particular and mathematics in
general play a role in many aspects of society. The ways DVs are designed and
used to communicate, and how they are read and made sense of by readers, have
implications for how people live, what knowledge is accessible and what
decisions and judgements people make.
7.2 Theoretical conclusions, implications and reflections
This research aimed to develop theoretical perspectives that shed light on the
roles of journalistic DVs. This aim involved developing theoretical frameworks
for analyzing and guiding the synthesis of the findings. In this section, I will
present the main conclusions regarding these theoretical aims. I will present
conclusions pertaining to the concept of VNL, social semiotics, the application of
Giddens’ (1990) framework of phenomenology in late modernity, and the
connections between mathematical literacy, everyday mathematics and
mathematization as a social process.
7.2.1 Theoretical conclusions, implications and reflections regarding the
VNL frameworks
VNL, as coined by Tønnessen (2020) and grounded in the work of Hasan (1996,
2003) on literacy, is the most salient concept in this dissertation. As it is regarded
as the special form of mathematical literacy that is needed to read and make
sense of DVs, it is the concept that connects this dissertation to the literature on
mathematical literacy. For this research, it was necessary to adapt the concept for
two different analyses. First, it was adapted to analyze the VNL that was
expected and invited of readers of journalistic COVID-19 DVs. This was realized
by applying relevant concepts to each of the three aspects of VNL (recognition,
action and reflection), and was used for the textual analysis of a corpus of
COVID-19 DVs presented in Paper III. It was adapted again for the analysis in
115
Paper IV, where the purpose was to analyze young adults’ VNL as they read and
make sense of journalistic COVID-19 DVs in a digital and interactive media, a
process captured by audio and screen recordings. In this case, the adaptation of
VNL largely consisted of studying the original text by Hasan (1996) and creating
open categories of recognition, action and reflection literacy that could be
identified in dialogue and screen activity. Further, Hasan’s (1996) distinction of
the context of the sign system and the social context proved useful and was
applied to all the three aspects of VNL. By adapting these two analytic
frameworks and applying them to data analysis, some new insights about VNL
were learned. First, the three aspects of literacy were a versatile and illuminating
approach to studying VNL that proved useful both for analyzing DVs and
processes of reading and making sense of DVs. Furthermore, I expect that these
three aspects are useful for analyzing design processes, for example among
professional DV designers in newsrooms. In the case of VNL, the three aspects
are connected to meaning-making grounded in the sign system of DVs. If the
three aspects are applied to other mathematical signs systems, Hasan’s (1996,
2003) concept of literacy can also be applicable to other forms of mathematical
literacy. This may apply to studies in which researchers study how people
decode, use and reflect on mathematics in various social activities.
Second, through the analysis of young adults’ VNL, it was observed that the
three aspects of VNL (recognition, action and reflection) and the two contexts
(the context of the sign system and the social context) were mutually reinforcing
(Finding 5 in Table 7). This observation runs contrary to a statement of Hasan
(1996) which is also reiterated by Tønnessen (2020): “So these three facets of
literacy are related by logical inclusion: reflection literacy includes a well-
informed version of action literacy […], and the latter includes recognition
literacy; the reverse is, however, not true” (Hasan, 1996, p. 417). Hasan here
describes the relationship between the aspects as a hierarchy. I do not critique
Hasan’s (1996) general theory of literacy as this is beyond the scope of the
insights gained from this research. Rather, I will conclude that in the context of
informal everyday reading of DVs, the three aspects and two contexts of VNL
are mutually reinforcing and not hierarchical.
Although the adaptation of Hasan’s (1996, 2003) framework resulted in a
novel framework that deviated from the original ideas in certain ways, the
resulting frameworks for VNL proved useful and illuminating. Because the new
frameworks for VNL integrates several key aspects of literacy (recognition,
116
action, reflection, situatedness) into one coherent framework, it is a promising
analytical tool for further research.
7.2.2 Theoretical conclusions, implications and reflections regarding social
semiotics
The theoretical perspective of social semiotics is present in all the four papers.
The frameworks used for data analysis in Paper I and II are developed from
social semiotic concepts such as genre, semiotic resources and style (van
Leeuwen, 2005). In Paper III and IV, social semiotics is present through Hasan’s
(1996, 2003) literacy framework, which forms the backbone of VNL and is
grounded in social semiotics and various sociological, socio-linguistic and
pedagogical theories. Social semiotics has been a very useful theoretical
perspective to use for several reasons. First, social semiotics offer a set of
concepts on how people communicate that are intended to be networked with
other social theories (van Leeuwen, 2005). This openness towards theory
networking gave many opportunities to use theoretical frameworks and concepts
from other, relevant theories. Second, social semiotics has a rich and well-
developed inventory of frameworks and concepts (e.g., Kress & van Leeuwen,
1996; van Leeuwen, 2005). This provided me with ample opportunity to find
relevant and useful frameworks and concepts that answered my needs. The
frameworks could be adapted to the needs of my project by using relevant
concepts from other sources. For example, the framework of style used in Paper
II was adapted to the analysis of NWFs by combining it with ideas from science
and technology studies which yielded a novel and useful framework. In
summary, social semiotics was a useful theoretical perspective because it offered
a sound theoretical foundation for research, it offered relevant and useful analytic
concepts, and it was readily networked with other theories.
7.2.3 Theoretical conclusions, implications and reflections regarding
mathematization as a social process
An overarching motivation for this research was to explore the idea that society
is becoming more mathematical. This idea, mathematization as a social process,
also suggests that the growing presence and importance of mathematics have far-
reaching consequences for people’s lives (e.g., Keitel, 1989). Much of the
research on mathematization as a social process that originates from the
mathematics education literature uses the two-sided framework of
117
mathematization and demathematization (Chevallard, 1989; Jablonka, 2017;
Jablonka & Gellert, 2007; Keitel, 1989; Lensing, 2017; Skovsmose, 2020;
Straehler-Pohl, 2017). In my research, I deviated from this strand by instead
using Giddens’ (1990) sociological framework of four dialectics (displacement
and reembedding, intimacy and impersonality, expertise and reappropriation, and
privatism and engagement) to study mathematization as a social process and its
implications. I chose this framework because it connects themes such as
technology, trust, emotions, knowledge and political engagement to the
institutions and structures that characterize late modern society. This framework
proved useful for structuring the discussion on the roles and consequences of
mathematization as a social process. It also provided explanatory power for
phenomena such as the tension between critical awareness of issues associated
with mathematization and the lack of critical action towards such issues
(Straehler-Pohl, 2017). Hence, it offered a novel perspective on the roles of
mathematics in contemporary society and yielded answers to questions that the
mathematization-demathematization framework had left unanswered.
A key aim of this research was to explore the non-trivial connections between
mathematization as a social process and the mathematical literacy expected of
citizens. After the findings from this dissertation were discussed in light of the
perspectives of mathematical literacy, everyday mathematics and
mathematization as a social process, it was possible to outline this connection. As
different expert systems (e.g., weather prediction, epidemiology, journalism) in
society are mathematized, the kind of mathematical literacy expected of
practitioners and users of these systems change, but these changes depend on the
circumstances and can be very different in the recognition, action and reflection
aspect of literacy. In the case of NWFs, the reader is expected to recognize a
relatively stable inventory of signs (maps, tables, short texts, numerals, some
special meteorological symbols) and know how to use this to inform action and
judgements. Further, some meta-knowledge of weather prediction, such as the
fact that not all storms are accurately predicted or that an expected rainfall may
not occur (Compton, 2018; Masson & Es, 2020), can inform reflections, while
not requiring insights into the particularities of numerical weather prediction. The
circumstances around journalistic COVID-19 DVs were very different, and the
mathematical literacy they expected of citizens was very different too. The
pandemic disrupted peoples’ everyday routines and posed risks to people’s well-
being. The DVs expected citizens to be able to decode a wide range of different
118
DVs, many of which broke DV conventions and challenged readers to explore
the data themselves. Furthermore, journalists invited readers to scrutinize the
presented data infrastructure and models, which may have conveyed a sense of
openness and trustworthiness. These DVs raised relatively high expectations on
readers’ ability to recognize and decode mathematical signs and to use the DVs
to inform action and judgements. Furthermore, the invitations to learn about the
underlying data infrastructures and models yielded opportunities to reflect on the
validity and reliability of the data, as well as the validity of using mathematical
models in the first place.
To sum up, the use of Giddens’ (1990) framework was successful in that it
offered a perspective on mathematization as a social process that complemented
the literature that uses the concepts of mathematization and demathematization,
and it aided me in connecting mathematization as a social process to everyday
mathematics and VNL.
To sum up the theoretical conclusions, in this research, several theories have
been applied and adapted. Some of the theories are new to mathematics
education. For example, Hasan’s (1996, 2003) literacy framework and Giddens’
(1990) framework for late modernity have not, to my knowledge, been used in
mathematics education research before. The application of novel theoretical
frameworks has provided new ways of understanding mathematical literacy and
mathematization as a social process and given new insights. Social semiotics has
worked well as a theoretical and analytical foundation that readily connected
with the other theories used.
7.3 Strengths, limitations and reflections on methodology
In this section, I will discuss some of the strengths and weaknesses of this
research, and some overarching reflections.
In Study A, which includes Paper I and II, a weakness was that I only
researched one channel of weather forecasting, and did not consider channels
such as radio, television and the internet. The different channels of weather
forecasting may have influenced one another. For example, readers may have
learned about weather maps on television and used this knowledge to make sense
of NWFs. I may have missed developments relevant to literacy requirements and
the implications of mathematization as a social process that happened in other
channels. Further, the popularity of printed newspapers has fallen in recent years,
meaning that NWFs reach a smaller, and likely a different, audience now than
119
before. Also, I only considered two newspapers in one country. Hence,
similarities or differences between countries were not identified. However, the
relatively small sample (n=44 in Paper I, n=46 in Paper II) of weather forecasts
of a relatively similar format that spanned a long time enabled me to explore
long-term trends.
In Study A, I only looked at printed NWFs from archives, and not the people
who read, used and relied on them. Therefore, I could not observe how people
read, reacted to and used these forecasts in their everyday lives.
In Study B, which was described through Paper III and IV, the focus was on
journalistic COVID-19 DVs. Both these papers used Verdens Gang’s (VG)
online collection of COVID-19 DVs, but Paper III presented a textual analysis of
these DVs, while Paper IV presented an analysis of interviews of young adults
reading and making sense of these DVs. The combination of textual analysis and
interviews is a strength, as it offers two perspectives on the same phenomenon.
The interviews reported in Paper IV had only two participants. While the low
number is a weakness, it also enabled an in-depth analysis of each case, which
revealed connections that would otherwise go unnoticed such as the crucial role
of lived experience in making sense of DVs and the many interconnections
between the three aspects of VNL. The focus on a compilation page hosted by a
popular news site complemented other relevant studies that look at the
mathematical aspects of the COVID-19 media coverage through news stories and
government reports (Aguilar & Castaneda, 2021; da Silva et al., 2021; Gal &
Geiger, 2022; Jablonka & Bergsten, 2021; Kwon et al., 2021).
In Section 6.3 I synthesized the research guided by Giddens’ (1990)
framework. This synthesis achieved two important things. First, it offered a novel
theoretical perspective on the notion of mathematization as a social process,
which itself yielded new insights (Findings 7–10 in Table 7). Second, it guided
the synthesis of the findings from this dissertation into the literature on
mathematization as a social process, which helped gaps in the empirical research
in this area. Nevertheless, as this dissertation is focused only on journalistic DVs,
more research is needed to further substantiate this theoretical perspective with
diverse empirical studies.
The adaptations and use of Hasan’s (1996, 2003) and Tønnessen’s (2020)
ideas into the frameworks for VNL that I have used warrant some reflections.
The main ideas that I have used are the three aspects of literacy (recognition,
action and reflection), and the two contexts, the social context and the context of
120
the sign system. In my adaptation, the core idea of each aspect and context was
preserved but it was adapted to the reading of DVs rather than the very general
“variable semiotic process” unrestricted to reading and writing, which Hasan
(1996, p. 384) originally wrote about. Hasan developed these concepts in a larger
socio-critical and semiotic framework on the role of pedagogical institutions in
the learning of language. I have not made use of the entire framework. For
example, she distinguishes between the quotidian line of literacy development
and a specialized line of development which is prescribed by official pedagogical
institutions (Hasan, 1996), a distinction I have not used. The omission of certain
concepts changes the overall framework. Therefore, my use of Hasan’s ideas was
both an adaptation and a recontextualization of her original work.
Learning does not have a central place in this research. Nevertheless, aspects
of literacy development are described through the application of Hasan’s (1996,
2003) framework, and the distribution of mathematical knowledge is sketched
out through the application of Giddens’ (1990) dialectic of expertise and
reappropriation. Hence, learning is a peripheral theme which is described
indirectly. I encourage further research aiming to describe mathematical learning
processes in everyday situations.
7.4 Recommendations
This research has yielded findings that can offer recommendations for further
research and implications for practice and policy.
The field of everyday mathematics still needs more research. I identify a need
for more research on everyday mathematical activities in diverse contexts,
including what context-specific forms of mathematics are involved and what
characterizes people’s mathematical literacy in these activities. Further questions
related to everyday mathematics concern the people who are included and
excluded from participation, for what reasons, and what the consequences of
unequal participation are for the individual and for society as a whole. Also, I
recommend research on people’s mathematical learning trajectories in everyday
activities and how everyday mathematical knowledge is connected (or not
connected) to school mathematics.
As my research and other studies have shown (Heyd-Metzuyanim et al.,
2021; Kennedy & Hill, 2018), school mathematics does not always succeed in
preparing students for participation in everyday mathematical activities.
Therefore, I encourage further research on design and implementation of school
121
interventions that target development of mathematical skills that are useful
outside of the school setting and that foster positive emotions and dispositions
towards mathematics. Such interventions can strengthen the connection between
school mathematics and everyday mathematics so that they can better support
one another, and so that students are better prepared to recognize everyday
mathematics activities as mathematical. Further, long-term studies that trace
peoples’ participation in everyday mathematical activities from school to
adulthood can provide insight into the trajectories that lead to unequal
participation.
This research, as well as Heilmann (2020) and OECD (2019), have shown
that unequal access to everyday mathematical practices is a significant issue.
Therefore, another area for further research can be on how meteorological,
medical, administrative and other institutions can invite people to productively
engage with their quantitative information, and what challenges and
opportunities people encounter in these processes. Relevant efforts include
designing legends and user guides with information about underlying
mathematical models and data infrastructures. Such efforts may support readers
to recognize the meaning of relevant mathematical signs, be able to act on
mathematical information, and reflect on issues of validity and trustworthiness.
The institutions most closely related to this research are newsrooms. As sites
of non-formal mathematical practices that foster non-formal mathematical
learning, newsrooms offer interesting opportunities for future research. Future
research can target the workplace mathematics of journalists, including their
approaches to problem solving, their use of (digital) tools, division of labor, data
collection, data handling and mathematical modelling. Further, their views on
their role as non-formal mathematics educators and their efforts to make their
journalistic mathematical content attractive and accessible are relevant themes
for research.
In terms of recommendations for use and development of theory, I
recommend further work on applying Hasan’s (1996, 2003) literacy framework
to other mathematical sign systems to develop novel frameworks for particular
forms of mathematical literacy in particular social contexts. The works of
Giddens’ (1984, 1990, 1991) on late modernity were used to shed light on the
roles of mathematics in contemporary society. This research can be extended, for
example by incorporating other forms of mathematical practices and further
theorize the roles of schools and other institutions.
122
Everyday mathematics is a complex and changing phenomenon, and it
continually challenges mathematics educators to find new ways to prepare people
for participation in a complex world. Different people need different forms of
mathematical literacy, and the needs are changing. This research cannot give a
comprehensive diagnosis of the challenges in this area. Nevertheless, some key
challenges are that everyday mathematics is often not recognized as
‘mathematical’ because it is qualitatively different from school mathematics, and
that negative emotions towards school mathematics hold people back from
participating in mathematics-rich everyday activities. Therefore, I encourage
efforts to show that mathematics can be useful, relevant, accessible, interesting,
and even fun, to students and non-students alike.
123
List of references
Abtahi, Y., Graven, M., & Lerman, S. (2017). Conceptualising the more
knowledgeable other within a multi-directional ZPD. Educational Studies
in Mathematics, 96(3), 275–287. https://doi.org/10.1007/s10649-017-
9768-1
Aguilar, M. S., & Castaneda, A. (2021). What mathematical competencies does a
citizen need to interpret Mexico’s official information about the COVID-
19 pandemic? Educational Studies in Mathematics, 108(1), 227–248.
https://doi.org/10.1007/s10649-021-10082-9
Aiello, G. (2020). 3. Inventorizing, situating, transforming: Social semiotics and
data visualization. In M. Engebretsen & H. Kennedy (Eds.), Data
visualization in society (pp. 49–62). Amsterdam University Press.
https://doi.org/10.2307/j.ctvzgb8c7.9
Allen, W. L. (2018). Visual brokerage: Communicating data and research
through visualisation. Public Understanding of Science, 27(8), 906–922.
https://doi.org/10.1177/0963662518756853
Andersson, A., Valero, P., & Meaney, T. (2015). “I am [not always] a maths
hater”: Shifting students’ identity narratives in context. Educational
Studies in Mathematics, 90(2), 143–161. https://doi.org/10.1007/s10649-
015-9617-z
Bauer, P., Thorpe, A., & Brunet, G. (2015). The quiet revolution of numerical
weather prediction. Nature, 525(7567), 47–55.
https://doi.org/10.1038/nature14956
Beck, U., Giddens, A., & Lash, S. (1994). Reflexive modernization: Politics,
tradition and aesthetics in the modern social order. Stanford University
Press.
Benjamin, S. G., Brown, J. M., Brunet, G., Lynch, P., Saito, K., & Schlatter, T.
W. (2019). 100 years of progress in forecasting and NWP applications.
Meteorological Monographs, 59, 13.11–13.67.
https://doi.org/10.1175/amsmonographs-d-18-0020.1
Bezemer, J., & Kress, G. (2020). Semiotic work in the science classroom.
Cultural Studies of Science Education, 15(1), 71–74.
https://doi.org/10.1007/s11422-019-09957-4
Bikner-Ahsbahs, A., & Prediger, S. (2010). Networking of theories—An
approach for exploiting the diversity of theoretical approaches. In B.
Sriraman & L. English (Eds.), Theories of mathematics education:
Seeking new frontiers (pp. 483–506). Springer.
https://doi.org/10.1007/978-3-642-00742-2_46
Bolstad, O. H. (2020). Secondary teachers’ operationalisation of mathematical
literacy. European Journal of Science and Mathematics Education, 8(3),
115–135. https://doi.org/10.30935/scimath/9551
Bonk, C. J., & Kim, K. A. (1998). Extending sociocultural theory to adult
learning. In M. Cecil Smith & T. Pourchot (Eds.), Adult learning and
development: Perspectives from educational psychology (pp. 67–88).
124
Bonotto, C. (2005). How informal out-of-school mathematics can help students
make sense of formal in-school mathematics: The case of multiplying by
decimal numbers. Mathematical Thinking and Learning, 7(4), 313–344.
https://doi.org/10.1207/s15327833mtl0704_3
Bourdieu, P. (2004). Science of science and reflexivity. Polity.
Brown, M., Brown, P., & Bibby, T. (2008). “I would rather die”: Reasons given
by 16-year-olds for not continuing their study of mathematics. Research in
Mathematics Education, 10(1), 3–18.
https://doi.org/10.1080/14794800801915814
Bryant, C. G., & Jary, D. (2003). Anthony Giddens. In G. Ritzer (Ed.), The
blackwell companion to major contemporary social theorists (pp. 247–
273). https://doi.org/10.1002/9780470999912.ch11
Bryman, A. (2016). Social research methods. Oxford university press.
Börner, K., Bueckle, A., & Ginda, M. (2019). Data visualization literacy:
Definitions, conceptual frameworks, exercises, and assessments.
Proceedings of the National Academy of Sciences, 116(6), 1857–1864.
https://doi.org/10.1073/pnas.1807180116
Carmi, E., Yates, S. J., Lockley, E., & Pawluczuk, A. (2020). Data citizenship:
Rethinking data literacy in the age of disinformation, misinformation, and
malinformation. Internet Policy Review, 9(2), 1–22.
https://doi.org/10.14763/2020.2.1481
Carraher, D. W., & Schliemann, A. D. (2002). Chapter 8: Is everyday
mathematics truly relevant to mathematics education? Journal for
Research in Mathematics Education. Monograph, 131–153.
https://doi.org/10.2307/749968
Cazden, C., Cope, B., Fairclough, N., Gee, J., Kalantzis, M., Kress, G., Luke, A.,
Luke, C., Michaels, S., & Nakata, M. (1996). A pedagogy of
multiliteracies: Designing social futures. Harvard Educational Review,
66(1), 60–92. https://doi.org/10.17763/haer.66.1.17370n67v22j160u
Chassapis, D., & Giannakopoulou, E. (2015). Numbers on the front page:
Mathematics in the news. In U. Gellert, J. Giménez Rodríguez, C. Hahn,
& S. Kafoussi (Eds.), Educational Paths to Mathematics: A C.I.E.A.E.M.
Sourcebook (pp. 247–262). Springer International Publishing.
https://doi.org/10.1007/978-3-319-15410-7_16
Chevallard, Y. (1989). Implicit mathematics: Their impact on societal needs and
demands. In J. Malone, H. Burkhardt, & C. Keitel (Eds.), The
mathematics curriculum: Towards the year 2000: Content, technology,
teachers, dynamics (pp. 49–57). Curtin University of Technology.
Civil, M. (2002). Chapter 4: Everyday mathematics, mathematicians’
mathematics, and school mathematics: Can we bring them together?
Journal for Research in Mathematics Education. Monograph, 11, 40–62.
https://doi.org/10.2307/749964
Coddington, M. (2015). Clarifying journalism’s quantitative turn: A typology for
evaluating data journalism, computational journalism, and computer-
assisted reporting. Digital Journalism, 3(3), 331–348.
https://doi.org/10.1080/21670811.2014.976400
125
Compton, J. (2018). When weather forecasters are wrong: Image repair and
public rhetoric after severe weather. Science Communication, 40(6), 778–
788. https://doi.org/10.1177/1075547018799101
Cushion, S., Lewis, J., & Callaghan, R. (2017). Data journalism, impartiality and
statistical claims. Journalism Practice, 11(10), 1198–1215.
https://doi.org/10.1080/17512786.2016.1256789
d’Ambrósio, U. (2006). Ethnomathematics: Link between traditions and
modernity. Brill.
da Silva, A. S., Barbosa, M. T. S., de Souza Velasque, L., da Silveira Barroso
Alves, D., & Magalhães, M. N. (2021). The COVID-19 epidemic in
Brazil: How statistics education may contribute to unravel the reality
behind the charts. Educational Studies in Mathematics, 108(1), 269–289.
https://doi.org/10.1007/s10649-021-10112-6
Davey Smith, G. (2019). Post–modern epidemiology: When methods meet
matter. American Journal of Epidemiology, 188(8), 1410–1419.
https://doi.org/10.1093/aje/kwz064
Davis, P. J., & Hersh, R. (1986). Descartes’ dream: The world according to
mathematics. Harcourt Brace Jovanovich.
Debreu, G. (1991). The mathematization of economic theory. The American
Economic Review, 81(1), 1–7. https://www.jstor.org/stable/2006785
Deejay, A., & Henne, K. (2024). Creating a new normal? Technosocial relations,
mundane governance and pandemic-related disruption in everyday life.
Sociology, 58(3), 699–716. https://doi.org/10.1177/00380385231205135
Didier, E. (2024). Numbers and emotions in the governance of the Covid-19
datademic. Social Studies of Science, Ahead of print.
https://doi.org/10.1177/03063127241262457
Engebretsen, M. (2013). Visuelle samtaler: Anvendelser av fotografi og grafikk i
nye digitale kontekster. Fagbokforlaget.
Engebretsen, M. (2020). From decoding a graph to processing a multimodal
message: Interacting with data visualisation in the news media. Nordicom
Review, 41(1), 33–50. https://doi.org/10.2478/nor-2020-0004
Engebretsen, M., & Kennedy, H. (2020). Data visualization in society.
Amsterdam University Press. https://doi.org/10.2307/j.ctvzgb8c7
Engebretsen, M., Kennedy, H., & Weber, W. (2018). Data visualization in
scandinavian newsrooms: Emerging trends in journalistic visualization
practices. Nordicom Review, 39(2), 3–18. https://doi.org/10.2478/nor-
2018-0007
Engebretsen, M., & Weber, W. (2017). Graphic modes: The visual representation
of data. In C. Cotter & D. Perrin (Eds.), The Routledge handbook of
language and media (pp. 277–295). Routledge.
Ferreira, C. T., & Silva, C. C. (2020). The roles of mathematics in the history of
science: The mathematization thesis. Transversal: International Journal
for the Historiography of Science, 8, 6–25. https://doi.org/10.24117/2526-
2270.2020.i8.03
Few, S. (2012). Show me the numbers. Analytics Press.
126
FitzSimons, G. E., & Björklund Boistrup, L. (2017). In the workplace
mathematics does not announce itself: Towards overcoming the hiatus
between mathematics education and work. Educational Studies in
Mathematics, 95(3), 329–349. https://doi.org/10.1007/s10649-017-9752-9
Frejd, P., & Bergsten, C. (2016). Mathematical modelling as a professional task.
Educational Studies in Mathematics, 91(1), 11–35.
https://doi.org/10.1007/s10649-015-9654-7
Friel, S. N., Curcio, F. R., & Bright, G. W. (2001). Making sense of graphs:
Critical factors influencing comprehension and instructional implications.
Journal for Research in Mathematics Education, 32(2), 124–158.
https://doi.org/10.2307/749671
Frisinger, H. H. (1972). Aristotle and his “Meteorologica”. Bulletin of the
American Meteorological Society, 53(7), 634–638.
https://doi.org/10.1175/1520-0477(1972)053<0634:Aah>2.0.Co;2
Gainsburg, J. (2006). The mathematical modeling of structural engineers.
Mathematical Thinking and Learning, 8(1), 3–36.
https://doi.org/10.1207/s15327833mtl0801_2
Gainsburg, J. (2008). Real-world connections in secondary mathematics
teaching. Journal of Mathematics Teacher Education, 11(3), 199–219.
https://doi.org/10.1007/s10857-007-9070-8
Gal, I., & Geiger, V. (2022). Welcome to the era of vague news: A study of the
demands of statistical and mathematical products in the COVID-19
pandemic media. Educational Studies in Mathematics, 111(1), 5–28.
https://doi.org/10.1007/s10649-022-10151-7
Geiger, V., Gal, I., & Graven, M. (2023). The connections between citizenship
education and mathematics education. ZDM – Mathematics Education,
55(5), 923–940. https://doi.org/10.1007/s11858-023-01521-3
Giannakopoulou, E. (2021). Adults watching numbers: Numerical information
about COVID-19 presented in Greek TV news. In D. Kollosche (Ed.),
Exploring new ways to connect: Proceedings of the Eleventh International
Mathematics Education and Society Conference (Vol. 2, pp. 467–476).
Tredition. https://doi.org/10.5281/zenodo.5414074
Giddens, A. (1984). The constitution of society: Outline of the theory of
structuration. Polity.
Giddens, A. (1990). The consequences of modernity. Polity.
Giddens, A. (1991). Modernity and self-identity: Self and society in the late
modern age. Polity.
Gingras, Y. (2001). What did mathematics do to physics? History of Science,
39(4), 383–416. https://doi.org/10.1177/007327530103900401
Giroux, H. A. (2004). Cultural studies, public pedagogy, and the responsibility of
intellectuals. Communication and Critical/Cultural Studies, 1(1), 59–79.
https://doi.org/10.1080/1479142042000180926
Glazer, N. (2011). Challenges with graph interpretation: A review of the
literature. Studies in Science Education, 47(2), 183–210.
https://doi.org/10.1080/03057267.2011.605307
127
Gray, J., Chambers, L., & Bounegru, L. (2012). The data journalism handbook:
How journalists can use data to improve the news. O’Reilly Media, Inc.
Gray, J., Gerlitz, C., & Bounegru, L. (2018). Data infrastructure literacy. Big
Data & Society, 5(2), 1–13. https://doi.org/10.1177/2053951718786316
Greiffenhagen, C., & Sharrock, W. (2008). School mathematics and its everyday
other? Revisiting Lave’s ‘Cognition in Practice’. Educational Studies in
Mathematics, 69(1), 1–21. https://doi.org/10.1007/s10649-008-9115-7
Halevy, A., Norvig, P., & Pereira, F. (2009). The unreasonable effectiveness of
data. IEEE intelligent systems, 24(2), 8–12.
https://doi.org/10.1109/MIS.2009.36
Halliday, M. A. K. (1978). Language as social semiotic: The social
interpretation of language and meaning. Hodder Education.
Harper, K. C. (2008). Weather by the numbers: The genesis of modern
meteorology. MIT Press.
Hasan, R. (1996). Literacy, everyday talk and society. In R. Hasan & G.
Williams (Eds.), Literacy in society (pp. 377–424).
Hasan, R. (2003). Globalization, literacy and ideology. World Englishes, 22(4),
433–448. https://doi.org/10.1111/j.1467-971X.2003.00311.x
Heilmann, L. (2020). Health and numeracy: the role of numeracy skills in health
satisfaction and health-related behaviour. ZDM, 52(3), 407–418.
https://doi.org/10.1007/s11858-019-01106-z
Heyd-Metzuyanim, E., Sharon, A. J., & Baram-Tsabari, A. (2021). Mathematical
media literacy in the COVID-19 pandemic and its relation to school
mathematics education. Educational Studies in Mathematics, 108(1), 201–
225. https://doi.org/10.1007/s10649-021-10075-8
Illouz, E. (2007). Cold intimacies: The making of emotional capitalism. Polity.
Jablonka, E. (2003). Mathematical literacy. In A. J. Bishop, M. A. Clements, C.
Keitel, J. Kilpatrick, & F. K. S. Leung (Eds.), Second international
handbook of mathematics education (pp. 75–102). Springer Netherlands.
https://doi.org/10.1007/978-94-010-0273-8_4
Jablonka, E. (2017). Mathematisation in environments of Big Data–“implicit
mathematics” revisited. In B. di Paola & U. Gellert (Eds.), Proceedings of
CIEAEM69. Mathematisation: Social process & didactic principle (pp.
43–51).
Jablonka, E., & Bergsten, C. (2021). Numbers don’t speak for themselves:
Strategies of using numbers in public policy discourse. Educational
Studies in Mathematics, 108(3), 579–596. https://doi.org/10.1007/s10649-
021-10059-8
Jablonka, E., & Gellert, U. (2007). Mathematisation–demathematisation. In U.
Gellert & E. Jablonka (Eds.), Mathematisation and demathematisation:
Social, philosophical and educational ramifications (pp. 1–18). Brill
Sense. https://doi.org/10.1163/9789460911439_002
Jewitt, C., Bezemer, J., & O’Halloran, K. (2016). Introducing multimodality.
Routledge.
Jorgensen Zevenbergen, R. (2011). Young workers and their dispositions
towards mathematics: Tensions of a mathematical habitus in the retail
128
industry. Educational Studies in Mathematics, 76(1), 87–100.
https://doi.org/10.1007/s10649-010-9267-0
Jornet, A., & Damşa, C. (2021). Unit of analysis from an ecological perspective:
Beyond the individual/social dichotomy. Learning, Culture and Social
Interaction, 31, 100329, 1–10. https://doi.org/10.1016/j.lcsi.2019.100329
Jungwirth, H. (2010). On networking strategies and theories’ compatibility:
Learning from an effective combination of theories in a research project.
In B. Sriraman & L. English (Eds.), Theories of mathematics education:
Seeking new frontiers (pp. 519–535). Springer.
https://doi.org/10.1007/978-3-642-00742-2_49
Keitel, C. (1989). Mathematics education and technology. For the Learning of
Mathematics, 9(1), 7–13. https://www.jstor.org/stable/40247939
Kennedy, H., & Hill, R. L. (2018). The feeling of numbers: Emotions in
everyday engagements with data and their visualisation. Sociology, 52(4),
830–848. https://doi.org/10.1177/0038038516674675
Kirk, A. (2019). Data visualisation: A handbook for data driven design (2nd ed.).
Sage.
Kress, G. (2003). Literacy in the new media age. Routledge.
Kress, G., & Van Leeuwen, T. (2020). Reading images: The grammar of visual
design. Routledge.
Kristiansen, T. A. (2017). Meteorologi på reise: Veivalg og impulser i Arnt
Eliassen og Ragnar Fjørtofts forskerkarrierer. [Doctoral dissertation,
University of Bergen]. https://hdl.handle.net/1956/15587
Kucharski, A. J., Russell, T. W., Diamond, C., Liu, Y., Edmunds, J., Funk, S.,
Eggo, R. M., Sun, F., Jit, M., Munday, J. D., Davies, N., Gimma, A., van
Zandvoort, K., Gibbs, H., Hellewell, J., Jarvis, C. I., Clifford, S., Quilty,
B. J., Bosse, N. I., . . . Flasche, S. (2020). Early dynamics of transmission
and control of COVID-19: a mathematical modelling study. The Lancet
Infectious Diseases, 20(5), 553–558. https://doi.org/10.1016/S1473-
3099(20)30144-4
Kwon, O. N., Han, C., Lee, C., Lee, K., Kim, K., Jo, G., & Yoon, G. (2021).
Graphs in the COVID-19 news: A mathematics audit of newspapers in
Korea. Educational Studies in Mathematics, 108(1), 183–200.
https://doi.org/10.1007/s10649-021-10029-0
Lave, J. (1988). Cognition in practice: Mind, mathematics and culture in
everyday life. Cambridge University Press.
Lefebvre, H. (1991). The critique of everyday life. Verso. (Original work
published 1947)
Lensing, F. (2017). The dialectics of mathematization and demathematization. In
B. di Paola & U. Gellert (Eds.), Proceedings of CIEAEM69.
Mathematisation: Social process & didactic principle (pp. 199–202).
Li, N., & Molder, A. L. (2021). Can scientists use simple infographics to
convince? Effects of the “flatten the curve” charts on perceptions of and
behavioral intentions toward social distancing measures during the
COVID-19 pandemic. Public Understanding of Science, 30(7), 898–912.
https://doi.org/10.1177/09636625211038719
129
Lipsitch, M., Swerdlow, D. L., & Finelli, L. (2020). Defining the epidemiology
of Covid-19 — Studies needed. New England Journal of Medicine,
382(13), 1194–1196. https://doi.org/10.1056/NEJMp2002125
Lowrey, W., Broussard, R., & Sherrill, L. A. (2019). Data journalism and black-
boxed data sets. Newspaper Research Journal, 40(1), 69–82.
https://doi.org/10.1177/0739532918814451
Lüders, M., Prøitz, L., & Rasmussen, T. (2010). Emerging personal media
genres. New Media & Society, 12(6), 947–963.
https://doi.org/10.1177/1461444809352203
Mahroum, S. (2016). StormGeo in Bergen does it again and reinvents weather. In
Black swan start-ups: Understanding the rise of successful technology
business in unlikely places (pp. 161–176). Palgrave Macmillan UK.
https://doi.org/10.1057/978-1-137-57727-6_10
Maier, S. R. (2002). Numbers in the news: A mathematics audit of a daily
newspaper. Journalism Studies, 3(4), 507–519.
https://doi.org/10.1080/1461670022000019191
Maier, S. R. (2003). Numeracy in the newsroom: A case study of mathematical
competence and confidence. Journalism & Mass Communication
Quarterly, 80(4), 921–936. https://doi.org/10.1177/107769900308000411
Masson, E., & Es, K. v. (2020). 5. Rain on your radar : Engaging with weather
data visualizations as part of everyday routines. In M. Engebretsen & H.
Kennedy (Eds.), Data Visualization in Society (pp. 77–94). Amsterdam
University Press. https://doi.org/10.1515/9789048543137-009
Mau, S. (2019). The metric society: On the quantification of the social. Polity.
Mayer-Schönberger, V., & Cukier, K. (2013). Big data: A revolution that will
transform how we live, work, and think. Houghton Mifflin Harcourt.
Mediekatalogen. (2022). Rapporter og statistikk. Mediekatalogen. Retrieved 29
June 2022 from http://www.mediekatalogen.no/jsf/report/index.jsf
Mertens, D. M. (2016). Research and evaluation in education and psychology:
Integrating diversity with quantitative, qualitative, and mixed methods (4th
ed.). Sage publications.
Meyer, J. F. C. A., & Lima, M. (2022). Relevant mathematical modelling efforts
for understanding COVID-19 dynamics: An educational challenge. ZDM
– Mathematics Education, 55(1), 49–63. https://doi.org/10.1007/s11858-
022-01447-2
Miller, C. R. (1984). Genre as social action. Quarterly Journal of Speech, 70(2),
151–167. https://doi.org/10.1080/00335638409383686
Nilsen, Y., & Vollset, M. (2016). Vinden dreier: Meteorologiens historie i
Norge. Scandinavian Academic Press.
Niss, M., & Jablonka, E. (2014). Mathematical literacy. In S. Lerman (Ed.),
Encyclopedia of mathematics education (pp. 391-396). Springer
Netherlands. https://doi.org/10.1007/978-94-007-4978-8_100
Nunes, T., Carraher, T. N., Schliemann, A. D., & Carraher, D. W. (1993). Street
mathematics and school mathematics. Cambridge University Press.
O’Neil, C. (2016). Weapons of math destruction: How big data increases
inequality and threatens democracy. Broadway Books.
130
OECD. (2019). Skills matter: Additional results from the suirvey of adult skills.
https://doi.org/10.1787/1f029d8f-en
Parkin, B., & Harper, H. (2020). The application of three exotropic theories in
education. Mind, Culture, and Activity, 27(3), 233–248.
https://doi.org/10.1080/10749039.2020.1794008
Presmeg, N. (2007). The role of culture in teaching and learning mathematics. In
F. K. Lester (Ed.), Second handbook of research on mathematics teaching
and learning (Vol. 1, pp. 435–458).
Reimann, D., Struwe, M., Ram, N., & Gaschler, R. (2022). Typicality effect in
data graphs. Visual Communication, Ahead of print.
https://doi.org/10.1177/14703572221130445
Riggs, E. E., Shulman, H. C., & Lopez, R. (2022). Using infographics to reduce
the negative effects of jargon on intentions to vaccinate against COVID-
19. Public Understanding of Science, 31(6),751–765.
https://doi.org/10.1177/09636625221077385
Rose, N. (1991). Governing by numbers: Figuring out democracy. Accounting,
Organizations and Society, 16(7), 673–692. https://doi.org/10.1016/0361-
3682(91)90019-B
Roth, W.-M. (2003). Towards an anthropology of graphing: Semiotic and
activity-theoretic perspectives. Springer. https://doi.org/10.1007/978-94-
010-0223-3
Roth, W.-M. (2014). Rules of bending, bending the rules: The geometry of
electrical conduit bending in college and workplace. Educational Studies
in Mathematics, 86(2), 177–192. https://doi.org/10.1007/s10649-011-
9376-4
Rubel, L. H., Nicol, C., & Chronaki, A. (2021). A critical mathematics
perspective on reading data visualizations: Reimagining through
reformatting, reframing, and renarrating. Educational Studies in
Mathematics, 108(1), 249–268. https://doi.org/10.1007/s10649-021-
10087-4
Shah, P., & Hoeffner, J. (2002). Review of graph comprehension research:
Implications for instruction. Educational Psychology Review, 14(1), 47–
69. https://doi.org/10.1023/A:1013180410169
SIKT. (n.d.). Information for participants in research projects. Retrieved 05.
december 2023 from https://sikt.no/en/fylle-ut-meldeskjema-
personopplysninger/information-participants-research-projects
Sivle, A. D., Kolstø, S. D., Kirkeby Hansen, P. J., & Kristiansen, J. (2014). How
do laypeople evaluate the degree of certainty in a weather Report? A case
study of the use of the web service yr.no. Weather, Climate, and Society,
6(3), 399–412. https://doi.org/10.1175/wcas-d-12-00054.1
Skinner, J. (2020). Intimacy, Zoom tango and the COVID-19 pandemic.
Anthropology in Action, 27(2), 87–92.
https://doi.org/10.3167/aia.2020.270214
Skovsmose, O. (1994). Towards a critical mathematics education. Educational
Studies in Mathematics, 27(1), 35–57.
https://doi.org/10.1007/BF01284527
131
Skovsmose, O. (1998). Linking mathematics education and democracy:
Citizenship, mathematical archaeology, mathemacy and deliberative
interaction. Zentralblatt für Didaktik der Mathematik, 30(6), 195–203.
https://doi.org/10.1007/s11858-998-0010-6
Skovsmose, O. (2020). Mathematization as social process. In S. Lerman (Ed.),
Encyclopedia of mathematics education (pp. 605–608). Springer.
https://doi.org/10.1007/978-94-007-4978-8_112
Skovsmose, O. (2021). Mathematics and crises. Educational Studies in
Mathematics, 108(1), 369–383. https://doi.org/10.1007/s10649-021-
10037-0
Slettebø, T. (2021). Participant validation: Exploring a contested tool in
qualitative research. Qualitative Social Work, 20(5), 1223–1238.
https://doi.org/10.1177/1473325020968189
Snaprud, M., & Velazquez, A. (2020). 7. Accessibility of data visualizations: An
overview of European statistics institutes. In M. Engebretsen & H.
Kennedy (Eds.), Data Visualization in Society (pp. 111–126). Amsterdam
University Press. https://doi.org/10.1515/9789048543137-011
Straehler-Pohl, H. (2017). De| mathematisation and ideology in times of
capitalism: Recovering critical distance. In H. Straehler-Pohl, N.
Bohlmann, & A. Pais (Eds.), The disorder of mathematics education (pp.
35–52). Springer. https://doi.org/10.1007/978-3-319-34006-7_3
Tufte, E. R. (2001). The visual display of quantitative information (2nd ed.).
Graphics Press.
Tønnessen, E. S. (2020). 12. What is visual-numeric literacy, and how does it
work? In M. Engebretsen & H. Kennedy (Eds.), Data Visualization in
Society (pp. 189–206). Amsterdam University Press.
https://doi.org/10.2307/j.ctvzgb8c7.18
van Geenen, D., & Wieringa, M. (2020). 9. Approaching data visualizations as
interfaces: An empirical demonstration of how data are imag(in)ed. In M.
Engebretsen & H. Kennedy (Eds.), Data Visualization in Society (pp.
141–156). Amsterdam University Press.
https://doi.org/10.2307/j.ctvzgb8c7.15
Van Leeuwen, T. (2005). Introducing social semiotics. Psychology Press.
Vos, P., & Frejd, P. (2020). The object-tool duality in mathematical modelling. A
framework to analyze students’ appropriation of Sankey diagrams to
model dynamic processes. Avances de Investigación en Educación
Matemática(17), 52–66. https://doi.org/10.35763/aiem.v0i17.305
Vygotsky, L. S. (1978). Mind in society: Development of higher psychological
processes. Harvard university press.
Wager, A. A. (2012). Incorporating out-of-school mathematics: From cultural
context to embedded practice. Journal of Mathematics Teacher Education,
15(1), 9–23. https://doi.org/10.1007/s10857-011-9199-3
Wake, G. (2014). Making sense of and with mathematics: The interface between
academic mathematics and mathematics in practice. Educational Studies
in Mathematics, 86(2), 271–290. https://doi.org/10.1007/s10649-014-
9540-8
132
Weber, W., Engebretsen, M., & Kennedy, H. (2018). Data stories. Rethinking
journalistic storytelling in the context of data journalism. Studies in
Communication Sciences, 18(1), 191–206.
https://doi.org/10.24434/j.scoms.2018.01.013
Wedege, T. (1999). To know or not to know–mathematics, that is a question of
context. Educational Studies in Mathematics, 39(1), 205–227.
https://doi.org/10.1023/A:1003871930181
Wedege, T. (2010). Ethnomathematics and mathematical literacy: People
knowing mathematics in society. In C. Bergsten, E. Jablonka & T.
Wedege (Eds.), Mathematics and mathematics education: Cultural and
social dimensions. Proceedings of MADIF7, the seventh mathematics
education research seminar (pp. 31–46). SMDF.
Weiland, T. (2017). Problematizing statistical literacy: An intersection of critical
and statistical literacies. Educational Studies in Mathematics, 96(1), 33–
47. https://doi.org/10.1007/s10649-017-9764-5
Wellington, J. (2015). Educational research: Contemporary issues and practical
approaches. Bloomsbury Publishing.
Wells, G. (1994). The complementary contributions of Halliday and Vygotsky to
a “language-based theory of learning”. Linguistics and Education, 6(1),
41–90. https://doi.org/10.1016/0898-5898(94)90021-3
Wigner, E. P. (1960). The unreasonable effectiveness of mathematics in the
natural sciences. Richard courant lecture in mathematical sciences
delivered at New York University, May 11, 1959. Communications on
Pure and Applied Mathematics, 13(1), 1–14.
https://doi.org/10.1002/cpa.3160130102
Williams, J., & Wake, G. (2007). Black boxes in workplace mathematics.
Educational Studies in Mathematics, 64(3), 317–343.
https://doi.org/10.1007/s10649-006-9039-z
Wilson, K. (2008). Television weathercasters as potentially prominent science
communicators. Public Understanding of Science, 17(1), 73–87.
https://doi.org/10.1177/0963662506065557
Yates, S., Carmi, E., Lockley, E., Wessels, B., & Pawluczuk, A. (2021). Me and
My Big Data: Understanding Citizens Data Literacies Research Report.
https://.doi.org/10.17638/03180002
Åndsverkloven. (2018). Lov om opphavsrett til åndsverk mv. (LOV-2018-06-15-
40). Lovdata. https://lovdata.no/dokument/NL/lov/2018-06-15-40
133
Appendix
The appendix contains the four papers:
Paper I: Wiik, A. (2022). Trends in everyday mathematics: The case of
newspaper weather forecasts. In G. Nortvedt, N. Buchholtz, J. Fauskanger, M.
Hähkiöniemi, B. E. Jessen, H. K. Nilsen, M. Naalsund, G. Pàlsdòttir, P.
Portaankorva-Koivisto, J. Radišic, J. Ö. Sigurjónsson, O. L. Viirman & A.
Wernberg (Eds.), Bringing Nordic mathematics education into the future.
Proceedings of Norma 20, the ninth Nordic Conference on Mathematics
Education (pp. 257–264). SMDF.
Paper II: Wiik, A. (2022). From reader to friend to advertiser: Norwegian
newspaper weathercasters’ identities, roles and reader relations 1945-2020.
[Manuscript submitted for publication].
Paper III: Wiik, A., Vos, P., & Engebretsen, M. (2022). Visual-numeric
literacy: The case of COVID-19 data visualizations for news media and their
expectations on readers. [Manuscript submitted for publication].
Paper IV: Wiik, A., & Vos, P. (2024). Making sense of journalistic COVID-19
data visualizations: An in-depth study of two young adults’ visual-numeric
literacy. Adults Learning Mathematics – An International Journal, 18(1), 7–27.
Paper I
Wiik, A. (2022). Trends in everyday mathematics: The case of newspaper weather
forecasts. In G. Nortvedt, N. Buchholtz, J. Fauskanger, M. Hähkiöniemi, B. E. Jessen,
H. K. Nilsen, M. Naalsund, G. Pàlsdòttir, P. Portaankorva-Koivisto, J. Radišic, J. Ö.
Sigurjónsson, O. L. Viirman & A. Wernberg (Eds.), Bringing Nordic mathematics
education into the future. Proceedings of Norma 20, the ninth Nordic Conference on
Mathematics Education (pp. 257–264). SMDF.
Proceedings of NORMA 20
257
Tren ds in e veryday math ematics :
the case of newspaper weather forecasts
Anders Wiik
University of Agder, Faculty of Science and Technology, Norway; anders.wiik@uia.no
Making sense of weather forecasts in newspapers is a form of everyday mathematics with which many
people engage. In this paper, I describe how newspaper weather forecasts have changed between
1945 and 2015, thereby indicating trends in everyday mathematics. Aided by social semiotic theory,
a corpus of weather forecasts from two major Norwegian newspapers are analyzed. The findings
indicate that newspaper weather forecasts have shifted towards more non-verbal forms of
communication (maps, graphs, tables). This shift also changed the readers’ role, from an interpreter
of text to an organizer of information. I argue that students need to be better prepared for
participating in an increasingly quantified public discourse. I suggest that more interdisciplinary
schoolwork between mathematics and other school subjects where students use mathematical literacy
skills to explore socially relevant issues is needed.
Keywords: Everyday mathematics, weather forecasts, social semiotics
Introduction and literature review
Data are more available than ever before and are also more often communicated in the shape of –
sometimes innovative – data visualizations in news media (Engebretsen, 2017). The use of visual
representations of statistical data in the media is a domain of everyday mathematics in rapid
development. Further, the growing presence of quantitative data in the public means that
mathematical literacy is getting more important for citizens’ participation in public discourses.
Everyday mathematics, that is, the forms of mathematics that are developed and wanted in everyday
life (Wedege, 2010), has been a topic of interest in mathematics education research for a long time.
So, what does everyday mathematics look like in practice, and how does it change over time?
According to Zevenbergen (2011), mathematical practices and dispositions vary across generations.
For example, younger generations are more comfortable using technology and more inclined to
estimation and problem-solving, whereas older generations value mental, accurate calculations more
highly (Zevenbergen, 2011). Hence, mathematical practices should be regarded as evolving entities,
and the nature of these practices says something about the world for which school mathematics is
preparing them. In Goos’ model of mathematical literacy for living in the 21st century, elements of
mathematical literacy include the capacity to use mathematics in diverse contexts and using
representative tools such as symbol systems, graphs, and maps (Goos et al., 2014). In this paper, such
tools are called semiotic resources, a term from the discipline called social semiotics that I will
elaborate on later. The objective of this paper is to explore trends in everyday mathematics through
the context of newspaper weather forecasts (NWFs), focusing on the use of semiotic resources and
demands on the reader. The research questions are:
How has the use of semiotic resources in newspaper weather forecasts changed from 1945 to 2015?
How do the changes in newspaper weather forecasts change the readers’ role, and what do the
changes indicate for everyday mathematics?
Proceedings of NORMA 20
258
Regarding people’s abilities to read weather forecasts, Sivle and Aamodt (2019) found that many
non-experts struggle to understand verbal meteorological jargon in online weather forecasts. Instead,
readers prefer visual weather icons because they are more readily understood (Sivle & Aamodt,
2019). The respondents in Masson and van Es’ (2020) study used internet weather forecasts to
organize their plans, but they were nevertheless skeptical of the information provided by the service.
The skepticism was mostly due to lay knowledge concerning the difficulties involved with weather
prediction. However, it was also related to the nature of data visualizations: the informants regarded
the information displayed as incomplete and selective, and the methods used to compile the data as
obscured (Masson & van Es, 2020). Thus, at least some readers take a critical reader role.
To shed light on the research questions, I use data from the two leading newspapers in Norway,
Aftenposten and Verden Gang. The data covers the period from 1945 to 2015. I understand newspaper
weather forecasts as representations of meteorological data adapted to an audience of general news
readers and presented in newspapers printed on paper. I analyzed the NWFs using an analysis tool
inspired by social semiotic theories of communication (van Leeuwen, 2005), where the semiotic
resources used were categorized according to type (verbal-numeric text, maps, graphs, tables) and
degrees of visual salience (high, middle, low).
Semiotic resources, genre, and affordances
In my research, I use social semiotics to study communication. Semiotics is the study of meaning-
making, and social semiotics emphasizes how humans use various means of expression to act in the
social and cultural world (van Leeuwen, 2005). The things used to communicate (e.g., pictures,
spoken and written words) are called semiotic resources. According to van Leeuwen (2005), semiotic
resources are material, such as sound waves. The semiotic resources in NWFs are made by ink on
paper. I take NWFs as a genre, that is, a class of expressions that is recognizable based on shared
characteristics of content, form, and function (Miller, 1984). Engebretsen (2006) provides a model of
genre evolution based on the concept of affordances. As Gibson (1979) defines, ”[t]he affordances
of the environment are what it offers the animal, what it provides or furnishes, whether for good or
ill” (p. 127, emphasis in original). Hence, the affordances of the environment constitute a space of
possibilities and constraints for the evolution of the genre. Engebretsen differentiates between social
and instrumental affordances. Examples of social affordances are the needs, expectations, and
literacies of the readers. Textual innovations will not affect the genre over time unless they match
these social affordances. Printing technology is an example of instrumental affordances. Current
printing technology affords a much greater range of semiotic resources than precursors. According to
Engebretsen (2006), genre evolution reflects an interplay of social and instrumental affordances.
Methods
The object of study for this paper is the NWF genre. The research design is inspired by van Leeuwen
(2005). I chose to focus on the period starting in 1945 because it captures the period when
computerized modeling and simulations gradually became the dominant weather prediction method
(Bauer et al., 2015). I chose to focus on the two leading Norwegian newspapers in this period, Verdens
Gang (VG) and Aftenposten (Ap). Until 2012, Ap came in two daily issues, Ap morgen and Ap aften.
Note that Ap aften was a regional paper that was distributed in the Norwegian capital region. Because
Ap aften show some interesting deviations from the overall pattern, I decided to include it in this
Proceedings of NORMA 20
259
study. I opted to sample for 1 March every five years or the closest available date (1945, 1950, …,
2015) from all three issues. This sampling strategy yielded a corpus of n=44.
Next, the corpus was analyzed by (a) categorizing the types of semiotic resources and (b) assessing
the degree of salience for each semiotic resource. The categories are verbal-numeric text, maps,
graphs, and tables. I draw on Few (2012) to define the categories:
• Verbal-numeric text is text based on written words and numbers that is a separate
compositional element in the NWF and not embedded in a table or graph.
• Maps are geospatial representations of (a part of) the world.
• Graphs are defined according to three characteristics, ”[v]alues are displayed within an area
delineated by one or more axes”, ”[v]alues are encoded as visual objects positioned in relation
to the axes” and ”[a]xes provide scales (quantitative and categorial) that are used to label and
assign values to the visual objects” (p. 45). This includes line graphs and histograms.
• Tables are defined by two characteristics, ”[i]nformation is arranged in columns and rows”
and ”[i]nformation is encoded as text (including words and numbers)” (p. 43), colors, or icons.
Two issues arose in the categorization that I will briefly elaborate. First, there were a few cases were
verbal-numeric text was arranged in a table-like manner (e.g., Figure 2, left, ”været i morgen”). These
cases were classed as verbal-numeric text. Second, some semiotic resources use a table-like structure
but encode information with colors or icons. These cases were categorized as tables. Hence, I extend
Few’s definition of tables by including colors and icons as possible means of encoding information.
Another aspect of the semiotic resources that I analyzed is visual salience. Salience refers to ”the
degree to which [a semiotic resource] attract the viewer’s attention” (van Leeuwen, 2005, p. 284),
that is, how a semiotic resource stands out or is obscured. Salience is an outcome of an interaction of
several factors, including size, color, contrast, placement, and more (van Leeuwen, 2005). Because
the degree of salience is the product of a complex semiotic interplay, it does not lend itself to stringent
analysis. For this reason, I have opted to use three heuristic degrees of salience. High salience means
an item clearly stands out, for example, by being placed partly over other components, conspicuous
colors, large size, and so forth. Semiotic resources that are neither standing out nor being significantly
downplayed are categorized as middle salience. Low salience is used for components that are much
downplayed, for example, by being small, blending into the background color, and so forth. I use the
following symbols to encode high, middle, and low salience: ●◐○.
The final step in the analysis concerns situating the semiotic resources in their social, cultural, and
technological context (van Leeuwen, 2005). For this, I focus on the affordances of the four semiotic
resources outlined above (Few, 2012; Kress, 2003) in accordance with Engebretsen (2006). Verbal-
numeric text primarily affords narration (Kress, 2003). For displaying larger amounts of data, tables,
graphs, and maps are better suited (Few, 2012). The main affordance of graphs and maps is that they
can be used to display relationships among and between data, and maps have the additional affordance
of showing the geospatial location of the data (Few, 2012). The main affordance of tables is that they
make it easy for the reader to look up individual values (Few, 2012). The different semiotic resources
assign different roles to the reader (Kress, 2003), which I will elaborate in the conclusion.
Proceedings of NORMA 20
260
Results and analysis
Figure 1. Examples of NWFs. Left: Ap morgen 2 June 1945. Right: Ap aften 1 March 1975.
Reprinted with permission
Figure 2. Examples of NWFs. Left: VG 1 March 1990. Right: VG 1 March 2015.
Reprinted with permission
Proceedings of NORMA 20
261
Verdens Gang
Aftenposten morgen
Aftenposten aften
Year
Text
Maps
Tables
Text
Maps
Graphs
Tables
Text
Maps
Graphs
Tables
45
●
●
●
50
●
●
○
●
55
●
●
○
●
60
◐
●
○
●
65
●
◐◐
●!
◐
○
●
!
◐
70
●
●
!
◐
○
●
75
●
●
○
●
80
●
◐
●
◐
◐
!
◐
85
◐
●
●
◐◐◐
!
◐
◐
◐
!
◐
90
◐
●
◐◐○○
!
◐
◐
!
◐◐
95
◐
●
◐
◐◐
!
◐◐○○○
◐
○
00
◐◐◐
●◐
◐○
◐◐
●◐◐
!
◐◐
○○○○
◐
○
◐!
◐◐
05
◐○!
●◐
◐◐○○
◐◐○○
◐◐◐
◐!
◐◐◐◐◐
○○○
○○
●
◐!
◐◐
10
●◐
◐◐◐
○○○
◐○
◐◐◐
●◐◐
○○○○
◐○
◐◐◐
●◐◐
○○○○
15
◐
●
◐◐◐○
◐
◐◐◐
●◐◐
○○○
Discontinued
Table 1. Overview of the data. Legend: ●: High salience,
◐
: Middle salience, ○: Low salience
The results (Table 1) show that the four different semiotic resources have different distributions
across the period. While verbal-numeric text was the dominant resource in VG and Ap morgen for
most of the period until 1980, it was gradually phased out after 1985. Maps dominate in Ap aften
from 1950 until 1975, and from 1985 they dominate in VG. Starting ca. 2000, tables are by far the
largest category and continue to increase in frequency and salience towards 2015. NWFs from 2000
onward includes many tables, a few maps, and little supporting text. Graphs appear occasionally in
Aftenposten, but never in VG. Over the period, the amount of data and the level of detail has
significantly increased, and the information was gradually distributed across more semiotic resources.
Some possible explanations: social and instrumental affordances
The early NWFs, with verbal-numeric text only, may have been influenced by radio weather
forecasts. In December 1960, weather forecasts became a regular feature in Norwegian television
newscasts (Nilsen & Vollset, 2016), introducing maps, weather icons, and voiced narration to the
presentation of weather forecasts. This development afforded TV watchers opportunities to learn to
read weather maps. Around 1980, maps started to become a regular feature in the NWFs in VG and
Ap morgen. Assuming that the readers of NWFs were also TV watchers, it is likely that the demands
of reading weather maps were, at least in part, learned through watching TV.
Throughout the period this study covers, meteorological science underwent a mathematization and
computerization that led to increasingly accurate meteorological data that covered longer time spans
(Bauer et al., 2015). Hence, the meteorologists gradually produced more meteorological data, and the
data they produced became more accurate. Since maps, graphs, and tables better afford to visualize
Proceedings of NORMA 20
262
larger amounts of quantitative data (Few, 2012), meteorological developments likely contributed to
the shift away from verbal-numeric text and towards increased use of maps and tables. Further,
printing technology has undergone major changes, making images, graphs, maps, and tables cheaper
to print (Kress, 2003). Thus, printing technology also afforded the use of a broader repertoire of
semiotic resources.
Conclusions
This paper aims to shed light on two research questions. The first research question asked how the
semiotic resources used in NWFs changed between 1945 and 2015. The early NWFs were heavily
reliant on verbal-numeric text. The use gradually transitioned towards using multiple forms of data
representations where maps and tables typically were salient elements, and text, if any, only plays a
supporting role. Graphs were used occasionally. Maps, graphs, and tables are semiotic resources that
can efficiently express meaning by degree (Lemke, 2003). According to Lemke, meanings by degree
are the starting point for the language of mathematics: mathematical language enables one to move
smoothly between meaning by degree and meaning by category. The shift towards maps, tables, and
graphs foregrounds quantitative data (e.g., millimeters of precipitation, meters per second of wind)
and backgrounds the practical interpretation and evaluation of the data. This means that contemporary
NWFs present more quantitative data and use more semiotic resources that are characteristic of
mathematical language. Thus, NWFs have become more mathematical. This may be indicative of
broader developments in everyday mathematics, but further research is needed. This brings me closer
to the second research question, which asked for how the semiotic changes changed the role of the
reader and what these developments suggest for everyday mathematics. To do this, I will borrow
ideas from Kress (2003). According to him, verbal-numeric text on the one hand and maps, tables,
and graphs on the other affords somewhat different ways of communicating. He expresses this
difference with the metaphors telling and showing. Verbal-numeric text, which dominates the early
NWFs, mainly affords telling how the world is, where the message is more or less explicitly
formulated by the sender. The other semiotic resources used (maps, tables, graphs) affords showing
the world to a greater extent, leaving the interpretation of the information more open. Kress argues
that, around the turn of the millennium, the landscape of communication was characterized by a shift
from telling to showing the world. Thus, NWFs fall into this broader trend. This shift implies a change
in the role of the reader. Kress expresses this change by the metaphors ”reading as interpreting” (what
has been told) to ”reading as ordering” (what has been shown). Before this shift, readers were told
how to understand the data. However, after this shift, readers were shown the meteorological data to
extract, order, and interpret it themselves. Thus, the role of the reader changed from an interpreter to
an organizer. By taking the reading of NWFs as an example of an everyday mathematical practice,
they provide an example where everyday mathematics has become more complex, active, and
individualized.
Discussion: implications and significance for mathematics education
Being mathematically literate is not only about knowing procedures, conventions, facts, and concepts.
It is also about having the capacity to draw out useful information from semiotic artifacts like NWFs
and use this information to reflectively inform action in the world (Tønnessen, 2020). This study’s
significance is that it provides an example where everyday mathematics has become more complex,
requiring a complex mathematical literacy from the reader. However, in Kennedy and Hill’s (2018)
Proceedings of NORMA 20
263
work on people’s interactions with data visualizations, they found that many of their participants
experienced lacking confidence in their mathematical literacy, which hindered their reading of data
visualizations. Kennedy and Hill attribute their participants’ poor confidence to the way they learned
to relate to data, which was through formal mathematics education. Due to the growing importance
of quantitative data and visual representations of data in public discourses, here exemplified by
NWFs, teaching for mathematical literacy ought to prepare students to engage effectively with such
resources. NWFs, and data journalism more generally, are rapidly evolving genres. Hence, education
should equip students with strategies for organizing, sense-making, and critical questioning of
quantitative information and using this information to inform reflection and action in an evolving
semiotic landscape.
The accountability of preparing students to make sense of mathematical artifacts in everyday settings,
such as reading NWFs, is not clear cut. People learn and use mathematics in formal, institutionalized
settings like schools and universities and in informal settings like everyday interactions and
workplace activities. Kennedy and Hill’s (2018) findings suggest that the mathematical literacy that
students need outside of school should not be learned in the mathematics classroom alone. Kennedy
and Hill even claim that formal mathematics education can be counter-productive in preparing
students for engaging with data by making them anxious and unconfident when dealing with data. I
suggest that students can benefit from more interdisciplinary work between mathematics and other
school subjects, where students can use and develop their mathematical literacy to explore socially
relevant issues. Vos and Frejd (2020) provide an example of such an interdisciplinary intervention,
where students used Sankey diagrams, a mathematical concept, as a tool to explore issues related to
waste management. The grade 8 students in Vos and Frejd’s study readily appropriated Sankey
diagrams as tools despite little prior knowledge.
The research presented in this paper has some limitations. While the use of four categories of semiotic
resources does not provide a full description of NWFs, it is well suited to provide an overview of
historical changes. Also, the analysis of the reader’s role guided by affordances is limited because I
do not rely on user data. It remains an open question if other forms of everyday mathematics have
seen a similar development. More research is needed on the nature of everyday mathematics, the
mathematical obstacles that people experience in everyday contexts, and how students and adults can
be better prepared for the mathematical demands of living in the 21st century.
Acknowledgments
I am equally grateful to Martin Engebretsen and Pauline Vos for their contributions in the preparation
of this paper. I also wish to thank the anonymous reviewers for their insightful comments.
References
Bauer, P., Thorpe, A. & Brunet, G. (2015). The quiet revolution of numerical weather prediction.
Nature, 525(7567), 47–55.
Engebretsen, M. (2006). Shallow and static or deep and dynamic? Studying the state of online
journalism in Scandinavia. Nordicom Review, 27(1), 3–16.
Engebretsen, M. (2017). Levende diagrammer og zoombare kart: Datavisualisering som nyskapende
fortellerform i journalistikken. Norsk Medietidsskrift, 24(2), 1–27.
Proceedings of NORMA 20
264
Few, S. (2012). Show me the numbers! Designing tables and graphs to enlighten (2nd ed.). Analytics
press.
Gibson, J. J. (1979). The ecological approach to visual perception. Lawrence Erlbaum.
Goos, M., Geiger, V. & Dole, S. (2014). Transforming professional practice in numeracy teaching.
In Y. Li, E. A. Silver & S. Li (Eds.), Transforming mathematics instruction: Multiple approaches
and practices (pp. 81–102). Springer.
Kennedy, H. & Hill, R. L. (2018). The feeling of numbers. Emotions in everyday engagements with
data and their visualization. Sociology, 52(4), 830–848.
Kress, G. (2003). Literacy in the new media age. Routledge.
Lemke, J. L. (2003). Mathematics in the middle: Measure, picture, gesture, sign. In M. Anderson, A.
Saenz-Ludlow, S. Zellweger & V. V. Cifarelli (Eds.), Educational perspectives on mathematics
as semiosis: From thinking to interpreting to knowing (pp. 215–234). Legas.
Masson, E. & van Es, K. (2020). Rain on your radar: Engaging with weather data visualizations as
part of everyday routines. In M. Engebretsen & H. Kennedy (Eds.), Data visualization in society
(pp. 77–93). Amsterdam University Press.
Miller, C. R. (1984). Genre as social action. Quarterly Journal of Speech, 70(2), 151–167.
Nilsen, Y. & Vollset, M. (2016). Vinden dreier: Meteorologiens historie i Norge. Scandinavian
Academic Press.
Sivle, A. D. & Aamodt, T. (2019). A dialogue-based weather forecast: Adapting language to end-
users to improve communication. Weather, 74(12), 436–441.
Tønnessen, E. S. (2020). What is visual-numeric literacy, and how does it work? In M. Engebretsen
& H. Kennedy (Eds.), Data visualization in society (pp. 189–205). Amsterdam University Press.
van Leeuwen, T. (2005). Introducing social semiotics. Routledge.
Vos, P. & Frejd, P. (2020). The object-tool duality in mathematical modelling: A framework to
analyze students’ appropriation of Sankey diagrams to model dynamic processes. Avances de
Investigación en Educación Mathemática, 17, 52–66.
Wedege, T. (2010). Ethnomathematics and mathematical literacy: People knowing mathematics in
society. In C. Bergsten, E. Jablonka & T. Wedege (Eds.), Mathematics and mathematics
education: Cultural and social dimensions. Proceedings of MADIF 7. The seventh mathematics
education research seminar, Stockholm. (pp. 31–46). Linköping University.
Zevenbergen, R. J. (2011). Young workers and their dispositions towards mathematics: Tensions of
a mathematical habitus in the retail industry. Educational Studies in Mathematics, 76(1), 87–100.
Paper II
Wiik, A. (2022). From reader to friend to advertiser: Norwegian newspaper
weathercasters’ identities, roles and reader relations 1945-2020. [Manuscript
submitted for publication].
Title: From scientist to friend to advertiser: Norwegian newspaper weathercasters' identities, roles,
and reader relations 1945–2020
Author information:
Corresponding author:
Anders Wiik, Department of Mathematical Sciences, University of Agder, Universitetsveien 25, 4630
Kristiansand, Norway.
Email: anders.wiik@uia.no
Contents
From scientist to friend to advertiser: Norwegian newspaper weathercasters' identities, roles, and reader
relations 1945–2020 ...................................................................................................................................... 2
1. Introduction ....................................................................................................................................... 2
2. Shifts in public discourse ................................................................................................................... 3
3. Theoretical framework ...................................................................................................................... 4
4. Methods ............................................................................................................................................ 5
5. Findings and discussion ..................................................................................................................... 6
Style and identity ................................................................................................................................... 6
Discursive roles and relations .............................................................................................................. 10
Trust ..................................................................................................................................................... 11
Graphics and legends .......................................................................................................................... 12
6. Conclusions and discussion ............................................................................................................. 13
Answering the research question ........................................................................................................ 13
Revisiting style characteristics ............................................................................................................. 13
The role of the mathematization and computerization of meteorology ............................................ 15
References ............................................................................................................................................... 16
From scientist to friend to advertiser: Norwegian newspaper
weathercasters' identities, roles, and reader relations 1945–2020
Abstract
Weather forecasting is an important arena in which socially relevant scientific information is
disseminated to lay audiences. This paper focuses on historical changes in interpersonal aspects of
weather forecasts in two popular Norwegian newspapers in the period 1945–2020. This was a period of
major changes in meteorological science, public communication, and science communication. The
weather forecasts are analyzed with a social semiotic approach, focusing on the identity and discursive
roles of the sender, and the relationship between sender and reader. The results of the study indicate
that newspaper weather forecasting has changed in many ways, including sender identity and strategies
for establishing trust. This article also makes contributions to the literature on characteristics of
communication styles. Finally, it discusses how the mathematization and computerization of
meteorology may explain some of the observed changes.
Keywords
Weather forecasting, science communication, social semiotics, interpersonal communication,
mathematization
1. Introduction
Advances in meteorology and weather prediction have had profound effects on people's lives and
society at large (e.g., Dash, 2015; Hamilton et al., 2016; Randalls, 2010). However, the arenas in which
weathercasters interact with their audiences – weather forecasts on television, newspapers, radio,
internet browsers, and mobile applications – are relatively overlooked from the perspective of the public
understanding of science (Wilson, 2008. See Compton, 2018; Vannini & McCright, 2007, and Wilson,
2008, for some exceptions). Wilson suggested that television weathercasters are "perhaps the most
visible and least understood science communicators in our culture" (p. 85). Weathercasters are faced
with the difficult task of conveying uncertain meteorological information to broad and diverse audiences.
This communication practice should meet the information needs of the audience while being
trustworthy and understandable. According to Wilson (2008), many television weathercasters also use
their position as highly visible scientists to educate their audiences on science topics, such as weather
data collection and climate change. Such practices of providing insights into the underlying science is an
efficient method for creating trust in the presented data (Goodwin & Dahlstrom, 2014). Significant
interpersonal issues in the communication between scientists and lay audiences are sender identity,
roles, and relations with the audience – who is behind the forecast, what discursive roles does the
sender take, and what kind of relationship does the sender have with the audience (Bucchi, 2013;
Compton, 2018; Gerhards & Schäfer, 2009; Goodwin & Dahlstrom, 2014; Molek-Kozakowska, 2017)?
These interpersonal aspects of science communication remain largely untouched in studies of weather
forecasting and are addressed in this paper.
Underlying this paper is an expectation that changes in meteorological research will influence how
weather predictions are communicated. First, the path from weather prediction to weather forecasting is
very short, often involving the same people (Nilsen and Vollset, 2016; Wilson, 2008). Second, as
elaborated later in this paragraph, the research underlying weather prediction has changed significantly
since the second world war turning to an increasingly numerical discipline. Bauer et al. (2015: 53),
described these changes as 'the quiet revolution of numerical weather forecasting'. One of the earliest
traces of this 'revolution' occurred in 1904 when Vilhelm Bjerknes published some foundational concepts
of numerical weather forecasting (Kristiansen, 2017). This led him to formulate a set of equations, the
so-called primitive equations, that enabled him to study the dynamics of the atmosphere by applying
differential equations to empirical measurements (Kristiansen, 2017). These equations are still used in
weather prediction models. However, applying these equations requires tremendous computational
power. It was several decades later that the necessary technology and competence were available.
During the second world war, a research agenda to operationalize air mass analysis to computer-assisted
numerical weather predicting was initiated (Harper, 2008). In the 1950s, the first computerized
atmosphere simulations were run (Harper, 2008). Norwegian researchers were active participants in the
pioneering work on computerized numerical weather prediction in the US in the 1940s and 1950s
(Harper, 2008; Kristiansen, 2017). By 1970, computers were implemented in the day-to-day activity of
weather prediction in Norway, and the discipline of weather prediction gradually developed into a data-
intensive scientific practice relying on high-end technology (Kristiansen, 2017). Following the
computerization and mathematization of operational meteorology, improvements in theoretical
understanding, and use of global satellite observations, the accuracy and time span of weather
prediction gradually improved and continues to improve (Bauer et al., 2015; Benjamin et al., 2019).
Hence, the period from second world war to the present captures 'the quiet revolution'.
This paper presents a study on weather forecasts in the two leading newspapers in Norway since the
second world war, analyzing the interpersonal dimension of communication, that is, the semiotic choices
contributing to construing, negotiating, and enacting interpersonal relations (Halliday, 1978). Specifically,
I have investigated how identities, relations, and discursive roles are construed and how they evolve. The
developments are contrasted with broader trends in public discourse and explore how developments in
weather prediction can explain some findings. I use the term weather prediction to talk about the
scientific activity of predicting the weather, and weather forecasting to talk about the activity of
disseminating weather predictions.
2. Shifts in public discourse
Since 1945, communication patterns have evolved in many ways (e.g., Bucchi, 2013). Therefore, weather
forecasting discourse must be understood within the broader landscape of public communication. Two
such trends, informalization and marketization, are presented here. According to Wouters (1986, 2011),
around the end of the nineteenth century, a process of social informalization started. Each successive
generation increasingly became freer in their expression of emotions, social strata were redefined, and
conditions for material, physical, and medical safety improved (Wouters, 1986). This laid the groundwork
for processes of democratization and mixing of lifestyles (Fairclough, 1992). The informalization
processes accelerated dramatically between the 1950s and 1980s (Wouters, 2011). A semiotic
expression of informalization is identified as conversationalization, that is, discourses come to resemble
the style of everyday conversations between peers. Conversational style makes discourses more
accessible and makes participants appear equal (Fairclough, 1992; Van Leeuwen, 2005). However,
conversational style might divert attention from critical issues about validity and trust (Molek-
Kozakowska, 2017).
A different shift in public discourse is captured by Fairclough (1993) as 'the marketization of discourse',
specifying that promotional culture has colonized more and more discourses. This involved that public
discourses adopted an advertising style. However, advertising style does not necessarily imply that there
is a product and a potential customer: advertising style also functions to "model the identities and values
of consumer society" (Van Leeuwen, 2005: 149). Conversational style and advertising style are
elaborated in the theoretical framework below.
3. Theoretical framework
According to Halliday (1978), all meaningful communication fulfills an interpersonal metafunction, that
is, it enacts and construes social relations. Fairclough (1993) identified two sub-functions to the
interpersonal metafunction, the identity function and the relational function. The identity function refers
to the personal and social identities of social actors. The relational function refers to the ways that
relationships between social actors are construed. The participants have discursive roles, an attribute
closely associated with their identity (Halliday and Mathiessen, 2014; Van Leeuwen, 2005). In this paper,
discursive role is treated as an analytic category comprising what actions the participants are allowed
and expected to take. The participants' discursive role says something about social structures
(Fairclough, 2003). The research question guiding this paper is:
How have the identities, relations and discursive roles in newspaper weather forecasts changed since
1945?
An analytic construct that foregrounds identity is style (Van Leeuwen, 2005: 139–159). Style refers to
"[t]he manner in which a semiotic artifact is produced or a semiotic event is performed" (Van Leeuwen,
2005: 287). He has identified several types of styles and characteristics associated with each. Styles are
not mutually exclusive but may coexist within the same text. Three of his styles are found particularly
relevant to the present study. These are expert style (Van Leeuwen calls it "the style of the expert"),
conversational style, and advertising style.
Expert style is characterized by formal and technical vocabulary, preference for abstract nouns, few
verbs, and referring to social actors in the third person. Van Leeuwen suggests that expert style is
authoritative and creates a top-down relationship with the reader, especially when it is not mixed with
other styles. Scientific style may also be interpreted as respectful because by using it, the sender
assumes that the reader can understand it. Still following Van Leeuwen (2005), conversational style is
characterized by incomplete sentences, slang, informal vocabulary and spelling, evaluative statements,
and more. Conversational style gives a sense of equality to the discourse, even if only in appearance
(Fairclough, 1992; Van Leeuwen, 2005), giving the impression that readers have a chance to participate
on equal terms. Advertising style is characterized by a direct address to the reader, poetic devices such
as alliteration and rhymes, and use of positively laden adjectives (Fairclough, 1993; Van Leeuwen, 2005).
Importantly, in the context of advertising style, adjectives can apply to both the product and the
consumer – e.g., by wearing dramatic and passionate clothing, the consumer construes a similar identity
for him/herself (Van Leeuwen, 2005). Advertising style creates a bottom-up relation with the audience
by focusing on their needs and desires. The sender has certain discursive roles, such as providing
relevant meteorological information and establishing the trustworthiness of this information. Van
Leeuwen's work is not based on science communication: what he calls an expert is in the newspaper
weather forecasts is a meteorologist who offers information obtained through scientific methods.
Therefore, I adapt the styles from Van Leeuwen into scientific style, conversational style, and advertising
style. Below, I will review some relevant literature to specialize the style framework to weather
forecasts.
In Gilbert and Mulkay's (1984) work on scientists' discourse, they identified two markedly different,
coexistent discourses: 'the empiricist repertoire' and 'the contingent repertoire'. These are used for
different purposes and exhibit different styles. The styles correspond closely to scientific style and
conversational style, respectively. Characteristics for the empiricist repertoire are impersonality, minimal
reference to social actors, and minimal reference to beliefs. The empiricist repertoire produces an
appearance of distance between the scientists and his/her research, making the research appear
objective and independent of the researcher's beliefs and speculations. On the other hand, the
contingent repertoire is characterized by stressing the researcher's intuition and feel for research; social
actors and their beliefs and values are included, often in first and second person; and the significance of
past experiences and social ties are stressed. Interestingly, Gilbert and Mulkay associate the two
'repertoires' with different discursive roles. The two forms of discourse are used in contexts in which the
participants act under different social structures. When researchers use the empiricist repertoire, the
physical world should be presented so as "to speak, and sometimes to act, for itself" (p. 56). Hence, any
action dictated by anything else than ostensibly objective empirical data and stringent methods must be
suppressed and excluded from the account. Scientists used the contingent repertoire in situations in
which it was accepted that their actions were contingent on a range of social and cultural circumstances.
Plough and Krimsky (1987) present a related distinction between two forms of discourse. They
distinguish between 'technical rationality' and 'cultural rationality' as two modes of communicating risk.
Technical rationality, similar to scientific style, is characterized by an emphasis on scientific methods,
appeals to authority, depersonalized descriptions of risk, and more. On the other hand, cultural
rationality weighs democratic processes, folk wisdom, personalizes risk, emphasizes how it impacts
people, and more. Cultural rationality comes close to conversational style. Coleman (1995) takes up
Plough and Krimsky's framework for empirical research on the social impact of the different rationalities
in journalistic coverage of risks associated with a mine. She found that actors who used technical
rationality (she calls it 'scientific rationality') and thereby ostensibly distanced themselves from cultural
and moral concerns, were more often used to frame news stories. Further, their claims were accepted
without question. Other actors – environmentalists and native inhabitants – were portrayed as opposing
the established order and were disregarded as irrational and unscientific. Consequently, technical
rationality established authority and delegitimized opposing views.
4. Methods
Data were obtained by sampling weather forecasts from the two leading Norwegian newspapers in the
period 1945-2020. This period is chosen because the Norwegian Meteorological Institute resumed
producing weather forecasts in 1945 after a hiatus during the second world war, this period captures
'the quiet revolution' of weather prediction (Bauer et al., 2015) and it is a period when science coverage
in newspapers underwent significant changes (Bauer et al., 2006). The newspapers are Verdens Gang
(VG) and Aftenposten (Ap). Until 2012, Ap came in two different daily issues, the national morning issue
Aftenposten Morgen (ApM) distributed nationally, and the evening issue Aftenposten Aften (ApA) for the
capital region. The data consist of the weather forecasts published on 1 March every five years (1945,
1950, …, 2020) or the closest available date. The size of this stratified sample is 46.
The analysis consisted of 3 stages. First, I read and interpreted the material multiple times to get a 'feel'
for the material. Second, I systematically analyzed each forecast for the style characteristics they exhibit
as reviewed in the theoretical framework above. The results of this analysis are summarized in Table 1.
Because social actors are represented in characteristic ways in each style, Van Leeuwen's (2008)
frameworks for the linguistic and pictorial representation of social actors was used. Third, I investigated
other semiotic features related to identity, discursive roles, and relations, including how trust is
established and changes in the design of graphic elements. All translations were done by the author.
5. Findings and discussion
Style and identity
This section first presents two analyses of contrasting examples (Figure 1 and Figure 2). Then the
outcomes of the style analysis are presented (Table 1).
The first example (Figure 1) is typical for the early period in the sample. All the information is presented
as text. It has two sections, an introduction and a part with regional weather predictions. In the
introduction, which is the main bulk of the long vertical column to the left, we read statements like
"sparkling winter weather" and "blistering cold degrees", informal ways of evaluating the weather.
Further, we read, "… it's heading towards significantly better temperatures, reported by our
meteorological friends at Blindern [the location of the meteorologists' offices], mostly because a high-
pressure zone over South Norway is weakened and a west-north-western air stream is starting to have
an impact". This indicates that the sender is not a meteorologist but a journalist who is paraphrasing a
forecast and a meteorological scientific explanation to the forecast. While the reported temperatures
are fairly accurate with two and three significant figures (“-15,5 degrees Celsius at 8 AM” and “-17
degrees Celsius at the coldest”), the forecast is overall rough and imprecise and only reports a general
expectation that temperatures will rise. Two social actors are included – "us", first-person plural, and
"our meteorological friends", third-person plural. The use of informal, evaluative statements, absence of
scientific jargon, lack of focus on accuracy and uncertainty, and inclusion of social actors in first and third
person means that this piece of text is classified as conversational style. The paraphrasing of the
message from the meteorological scientists strengthens this conclusion because it suggests that the
sender's identity is a journalist, clearly distinguished from the scientist.
After the introduction, there is a quotation from a meteorologist, saying, "We cannot promise mild
weather – but, as mentioned, better temperatures". Next, there is a sentence on the lowest recorded
temperatures the past few days. This serves as a bridge to the regional weather predictions for the rest
of the day. These predictions are given with scientific jargon with accurate meanings. For example,
predicted wind is presented with intervals of Beaufort numbers (e.g., “5–3”) and Beaufort descriptions
(e.g., breeze). Cloud coverage is given with standardized terms (e.g., “clear sky”), and uncertainty is
explicitly mentioned when presenting expected precipitation (“probably a little snow”). No social actors
are included in this part of the forecast. The use of scientific jargon, regard for accuracy and uncertainty,
and exclusion of social actors, means that this piece of text is categorized as scientific style. Thus, this
weather forecast exhibits two styles. These styles are not hybridized but occur in separate parts of the
same forecast.
In the second example (Figure 2), the information is mostly presented in maps and tables. The maps and
tables are used to convey accurate weather data for specific locations, making it easier to personalize
the weather forecast. Four social actors are included. At the top of the forecast, "Pent.no" is written. This
serves both as a vignette to the weather forecast, as a URL to the newspaper's online weather service,
and suggests that the service pent (the Norwegian meteorological term for 'clear sky') is the sender
behind the forecast. Next to the map, StormGeo is represented with the company's logo and indicates
that this commercial meteorological enterprise provided the weather data. At the top right, there is a
black box containing an advertisement to Pent's mobile phone application. This application compares the
weather predictions from yr.no, the governmental weather service, and Storm, a commercial weather
service. Thus, these two weather services are referred to in the third person, and VG is directly
addressing readers, notifying them of a service that enables comparing the data from the two competing
services. The graphic design is refined and attractive, using colors, shading, iconic weather symbols, and
recurring elements such as the red arrows used as vignettes to the map and tables. At the bottom of the
forecast, there is a brief text explaining scientifically how low-pressure zones over Eurasia typically
behave in winter, accompanied by a photograph of a foggy terrain. The personalization and direct
address, the appealing design, and the juxtaposition of competing services are all characteristic of
advertising style. However, by presenting accurate data and a scientific elaboration of a meteorological
phenomenon, the forecast also exhibits characteristics of scientific style. Thus, this forecast has a hybrid
style.
Insert Figure 1 and Figure 2 ca. here.
Figure 1. Weather forecast from Verdens Gang, 1 March 1965. This forecast exemplifies the
conversational and scientific styles. Reproduced with permission from VG, 2020, VG, 1 March 1965.
Figure 2. Weather forecast from Verdens Gang, 1 March 2015. This weather forecast exemplifies the
expert/advertiser hybrid style. Prominent characteristics include accurate and abundant data, a scientific
exposition, visually attractive graphics, and links to additional commercial services. Reproduced with
permission from VG, 2020, VG, 1 March 2015.
Table 1 gives an overview of the analysis of styles used in the weather forecasts, showing how discursive
styles developed between 1945 and 2015. Before the year 2000, scientific style dominates, but there are
many instances of conversational style. From around 2000 onwards, every forecast exhibits a hybrid of
scientific and advertising style. In the next section I will describe the different sender identities.
V
erdens Gang
A
ftenposten Morgen
A
ftenposten Aften
19
45
S
S
C
19
50
S
S
S
1955 C S S
19
60
C
S
S
1965 S/C S S
1970 S S S
19
75
S
S
S
19
80
S
S
S
/C
19
85
S
S
S
1990 S S/C S
19
95
S
S
/A
S
2000 S/A S/A S/A
20
05
S
/A
S
/A
S
/A
20
10
S
/A
S
/A
S
/A
20
15
S
/A
S
/A
2020
S
/A
S
/A
Table 1. Summary of analysis of style. Legend: S: Scientific, C: Conversational, A: Advertising. Slashes
indicate that the NWF exhibits a hybrid or multiple styles.
Until 1995: Scientist and conversationalist identities
Between 1945 and 1995, the corpus is dominated by scientific style, interspersed by conversational style
(Table 1). Scientific style construes the sender identity as an authority in weather forecasting – someone
who knows technical jargon, scientific methods, and can give accurate descriptions of weather
predictions. Sometimes, this also means that the sender will communicate with the readers through
technical maps. The scientist is brief, factual and uses accurate wording. It is reasonable to assume that
the scientific style forecasts are indeed the words of practicing meteorologists with minimal editorial
filtering.
In the handful of forecasts with conversational style, this style is sometimes used consistently
throughout the forecasts (VG 1955, 1960, ApA 1945), and other times it coexists with scientific style (VG
1965, ApM 1990, ApA 1980). When using conversational style, the sender appears as a friend having a
casual conversation about the weather, hence a very different identity from the scientist. Two illustrative
examples follow:
"It's the first of March, and if the days develop as normal, we should see a lot of beautiful nylon
on the main street of Oslo in about fourteen days." (VG 1955)
"it will be grey and gloomy" (VG 1955)
Rather than focusing on accurate and technical words, the conversationalist uses vernacular words,
focuses on sensory experience, and evaluates the weather. When scientific style and conversational style
co-occur, the identities are not blended but appear in separate sections of the forecast such as in VG
1965 (Figure 1).
From 2000: Scientist and advertiser identities
A style shift occurs around the turn of the millennium. This time, the change is consistent in all three
newspapers and constitutes a turn towards advertising style. It never becomes pure advertising style
because characteristics of scientific style remain. Instead, the style is a hybrid of scientific style and
advertising style in which each style's characteristics are interwoven. Key features of advertising style
include that graphic elements become more visually attractive, words and symbols are used creatively,
and there are often references to commercial weather services. At the same time, there are
characteristics of scientific style such as using terminology and symbols from the meteorological register
(e.g., the Beaufort scale, pressure zones, isobars) and generally avoiding evaluating the weather forecast.
In this hybrid style, the sender identity is less coherent: s/he is both a scientist delivering scientifically
grounded weather predictions, and s/he is a salesperson, striving to make the weather forecast appear
attractive and marketing additional services.
In two instances in VG in my corpus, in 2000 and 2005, the sender identity is personalized (Van Leeuwen,
2008) with name, image, and personal contact information. In the image representation, the face of the
female presenter is directed to the reader at a frontal angle, she is at eye level with the reader, and her
gaze is directed towards the reader. According to Van Leeuwen (2008), all these choices carry meaning.
The frontal angle suggests involvement: the presenter is figuratively inviting the readers into her world.
Meeting the reader at eye-level indicates equality – although she represents science, she positions
herself at level with the reader. Her gaze, which directly addresses the reader, invites – or even demands
– social interaction. The social interaction is also captured in verbal text, in which she sometimes
answers weather-related questions from readers.
In the remaining instances, the sender is not personalized. However, multiple social actors are often
represented – actors such as weather data providers, pollution data provider, and graphic designer; and
additional services that the reader can access such as SMS weather services and internet weather
services. Because the weather forecast embodies the voices of multiple social actors, the sender identity
appears as a composite identity.
Discursive roles and relations
The discursive roles of the participants in an act of communication is what they are expected and
permitted to do. This section gives an overview of the discursive roles and reader relations of the
different sender identities. Because creating trust emerged as an important discursive role in the
analysis, a separate section is devoted to this.
The discursive role of the scientist is mainly to convey weather predictions with a high focus on accuracy
and less focus on elaborations of the meanings, practical implications, and evaluation thereof. For
example, the scientist will describe the weather using terms from meteorological science such as the
Beaufort wind scale and will avoid evaluating the predicted weather as "good" or "bad". The scientist
never mentions social actors. The use of technical, standardized, and impersonal jargon distances the
scientist from the reader and construes a top-down relationship with the reader (Van Leeuwen, 2005).
The conversationalist primarily fulfills a discursive role of interpretation, paraphrasing the
meteorologists' predictions with vernacular and evaluative terms, typical of conversation between
friends. Further, the conversationalist never uses tables, graphs, or maps. By mimicking peer interaction,
the conversationalist reduces the gap between sender and reader and construes a relationship of
equality (Van Leeuwen, 2005).
In the three cases in which scientific and conversational style coexist (VG 1965, ApM 1990, ApA 1980),
the two identities appear as two separate senders with an internal division of roles. All these three cases
start with the conversationalist providing some basic information paraphrased from the meteorologists
such as past weather statistics, general nation-wide trends in predicted weather, and a brief
meteorological explanation to the prediction. After the conversational introduction, the scientist takes
over and presents more detailed and accurate weather predictions in the characteristic scientific style.
Thus, the conversationalist establishes contact and relates to the readers as friends, while the scientist
gives authority to what the conversationalist says as well as providing further details.
In the scientist/advertiser hybrid style, the sender conveys weather data, but it is not always clear if the
sender is a meteorologist. Further, the sender acts as an advertiser in multiple ways. Foremost, the
forecasts of this period have a visually attractive design. This is achieved using colored symbols, maps
with effects such as shadowing and colors, and often a creative design. Further, the sender often
provides links to additional services such as pay-by-minute telephone weather services, mobile
applications, and internet services. Also, in this period, most forecasts have information about the source
of the data. This includes the weather data provider and providers of pollution data. In two cases, even a
graphic design company is credited. Another role aspect is that the scientist/advertiser is using less
technical jargon, often replacing it with numerical information or highly iconic symbols. This way, the
data is still represented accurately and demands less understanding of meteorological terminology from
the readers. The scientist/advertiser also meeting the user's demands by using abundant tables. Tables
make it possible for readers to find accurate weather data for their location (Few, 2012). The
characteristic 'advertising' way of presenting the forecast, with emphasis on aesthetic attractiveness,
personalization, and links to additional services, gives priority to the needs and desires of the reader and
hence creates a bottom-up relationship with the reader (Van Leeuwen, 2005: 149–152).
Trust
One important aspect of the discursive role of the sender is to establish trust in the weather predictions.
To some extent, trust is established by the reputation of the medium (Bucchi, 2013), in this case the
newspaper. Recent research suggests that trust in science is fluid and diverse (Coen et al., 2020).
Because weather prediction is inherently ridden with uncertainty (Bauer et al., 2015), and because the
highly complex methods that lead to the predictions are difficult to present in a brief and
comprehensible way, establishing trust in weather forecasts is a potentially difficult task. According to
Giddens (1990), trust in scientific research as used to make weather predictions depends in part on faith
because a full comprehension of the science is rarely an option. Here I will present the strategies that the
different sender identities use to establish trust. The success or failure in establishing trust is not subject
to the present paper.
In scientific style, I have identified two ways of establishing trust. The first way is to present the data in
an objective, non-evaluative way with technical jargon, like the empiricist repertoire that scientists use in
academic publications and public interactions to appear trustworthy and objective (Gilbert and Mulkay,
1984). Especially in older forecasts, it was common to include information with minimal relevance to the
reader, such as isobars and high-altitude air streams. The inclusion of such information, which is difficult
to operationalize even for scientists, helps establish the senders' identity as a researcher in meteorology
who can make sense of such advanced details. A second method for establishing trust, which often
coexists with the first, is to provide a scientific explanation to some of the predicted phenomena. For
example, in Aftenposten Morgen 1955, we read, "a south-western air stream over Great Britain and the
North Sea turns to the eastern mountains in southern Norway". This statement implicitly explains why
the weather is expected to become warmer. Such explanations, seeking to make the predictions
trustworthy, frequently appear until 1965. From 1970, there are no such explanations. It is also likely
that the reputation of the medium contributed to creating trust (Bucchi, 2013).
When conversational style is used, trust is established in slightly different ways. The identified strategies
for establishing trust in conversational style are the reputation of the medium, appeals to authority, and
secondhand meteorological explanations of the predictions made by said authority.
Advertising style always appears together with scientific style in the corpus. It seems that there is an
interpersonal division of roles between the two styles – advertising style conveys attractiveness and
scientific style conveys trust. For example, it is common to include a technical map with isobars in the
scientist/advertising hybrid style – a piece of information with minimal usefulness to the reader – which
therefore appears to mainly function to establish trust by reference to the scientific expertise of the
meteorologists. In the scientist/advertiser hybrid, there are always references to data sources, typically
in the form of a logo or a statement like "weather data are provided by The Norwegian Meteorological
Institute". Hence, trust is maintained by the reputation of the medium, maintaining aspects of scientific
style, and appeals to authority.
Graphics and legends
Across the corpus, there is a steadily growing use of non-verbal ways of communication, i.e., increased
use of maps, tables, and graphs, which changes the role of the reader from interpreting what has been
told to ordering what has been shown (see Wiik, 2020, for further details). As verbal text decrease and
graphic semiotic resources become more prominent, the signs used to represent meteorological
phenomena also change. In maps and tables, information is typically encoded without words but with
other signs. According to MacEachren (1995), signs vary by their degree of iconicity, that is, the extent to
which they resemble the phenomenon they represent. At the lower end of the scale are geometric
symbols (e.g., circles, squares), and at the higher end are pictorial symbols (e.g., clouds, raindrops). In
the first decades after the second world war, many of the symbols are best described as intermediate
between geometric and pictorial. These symbols include an asterisk for snow, a zig-zag line for lightning,
and arrows for air streams. Other symbols from this era are better described as geometric, for example,
the okta used to index cloud coverage. The early inventory of signs closely resembles the signs used in
the meteorologists' community, which still today remains largely unchanged (see World Meteorological
Association, 1992, 2017). Over time, many of the signs in weather forecasts evolve to become more
pictorial. From 1985, VG uses contour drawings of clouds, sun, raindrops, and snowflakes to indicate
cloud coverage and precipitation, and Aftenposten follows the same pattern from 1990 onwards. The
signs used for cloud coverage and precipitation gradually become even more pictorial, gradually starting
to use shading and colors. In most cases, the temperature is represented with numerals, in the newer
forecasts also color-coded to demarcate below and above freezing temperatures. This development
indicates that the collection of symbols and conventions used in weather forecasts have gradually
diverged from the collection of symbols used within the field of meteorology.
In the earlier forecasts, in which the graphics are relatively low on iconicity, legends are always included.
The legends are not always exhaustive. For example, even though maps in early weather forecasts
always included isobars, isobars are sometimes omitted from the legends. As the weather symbols
become more iconic, it also becomes less common to have legends. Nevertheless, legends still appear,
though not in any clear pattern. The decrease in the use of legends can be explained by the following
quote: "pictorial symbols should communicate without the necessity for a legend" (Robinson et al., cited
in MacEachren, 1995: 257). Hence, when signs became more pictorial, the need for legends diminished.
In the data selected for this study, graphics such as maps, tables, and graphs are never found in instances
holding a conversational style. It does not necessarily imply that graphic elements and conversational
style are incompatible.
6. Conclusions and discussion
Answering the research question
As shown above, there have been profound changes in the identities, discursive roles, and reader
relations in newspaper weather forecasts since the second world war. Because weather forecasts are an
arena in which weather prediction science achieves social relevance, such changes are significant for the
public understanding of science. There are two main findings in this study. First, conversational style
never got a hold for longer periods of time. This happens despite general trends towards informalization
in public discourse, especially between 1950–1980 (Wouters, 1986, 2011). Second, from 1995/2000
onwards, the style is a hybrid of scientific style and advertising style. The onset of advertising style
appears parallel to the establishment of the first competing commercial weather service in Norway,
Storm Weather Center (Kalvig, 2009; Mahroum, 2016). Thus, the sender's identity remains a weather
prediction scientist throughout most of the period with only brief interruptions by conversationalist
identity and hybridization with advertiser identity. All the sender identities (scientist, conversationalist,
and advertiser) have a discursive role of presenting trustworthy weather predictions, but they do this in
different ways. The scientist conveys weather predictions in an efficient, technical, scientific, and non-
evaluative way, which is likely effective in building trust (Gilbert and Mulkay, 1984; Plough and Krimsky,
1987). Before 1970, the scientist also uses meteorological theories to create trust. The scientist
construes a top-down relationship with the reader. The conversationalist's discursive role is mainly to
mediate the weather forecast by presenting it in an accessible language, interpreting it, and evaluating it.
The conversationalists' means for establishing trust are appeals to authority, secondhand meteorological
explanations, and the reputation of the newspaper. The conversationalist construes an equal
relationship with the reader. The scientist/advertiser conveys weather predictions and advertises for
data providers and additional services, responding to the needs and desires of the readers as consumers.
Trust is established partly by retaining some scientific characteristics, partly by appeals to authority, and
partly by the reputation of the medium. The scientist/advertiser construes a bottom-up relationship with
the readers by focusing on their needs and desires.
Revisiting style characteristics
A key construct used in this study is style. It was operationalized through a framework developed
primarily from Van Leeuwen (2005) in which three styles, scientific style, conversational style, and
advertising style, are characterized. Aided by related literature, this framework was specialized to
weather forecasts. Below I summarize the framework that I have developed and expand it based on
findings from this study.
A significant part of the semiotic resources used in newspaper weather forecasts is non-verbal, i.e.,
maps, graphs, and tables coded with colors and weather symbols. A challenge for the analysis was to
determine how these resources fit into the style categories. Based on the analyses, I have reached the
following conclusions:
Maps and graphs are commonly used in scientific style and the scientific/advertising hybrid style.
Scientific style maps can include meteorological information that is not directly relevant to the
reader (e.g., isobars, high-altitude air streams, high-pressure zones). In these cases, the
information functions to establish trust in the predictions and reinforce the sender's scientific
identity.
In scientific style, the symbols used can be more or less iconic.
In my corpus, advertising style is always combined with scientific style. In these style hybrids,
information not directly relevant to the reader is occasionally included. The symbols are always
highly iconic.
Maps, graphs, tables, and symbols used in advertising/scientific hybrid style are more 'polished',
that is, they have design features such as shading and colors that create a visually attractive look.
In the advertising/scientific hybrid style, it is common to use weather symbols in creative ways,
e.g., by putting headlines inside wind arrows and having sunset/sunrise tables on a sun icon.
Data visualizations other than maps (e.g., line graphs, histograms) are rarely used. However,
when used, it is in scientific style or in the scientific/advertising hybrid style.
Weather forecasts appear to be pioneering the use of data visualizations in newspapers. This is likely
because data visualizations, especially maps, are an integral part of the meteorologists' register, because
early weather forecasts relied heavily on the meteorologists' register, and because maps enable efficient
displays of geospatial data (Few, 2012).
The findings suggests that the visual symbols in weather forecasts diverged from the meteorologists'
register of meaningful signs. Thus, a new register developed, one which served to mediate between
meteorologists and laypeople. The emerging register, which we can term 'the weather forecasting
register', is born from the meteorologists' register. Over time, opportunities provided by printing
technology and reader demands for understandability and appeal have shaped it into a distinct system
for meaning-making. Characteristic of the weather forecasting register are (1) that the weather symbols
are more iconic (e.g., cloud coverage is represented by pictorial representations of clouds instead of
oktas), (2) visual semiotic resources are often aesthetically enhanced with remedies like color and
shading, and (3) visual elements like wind arrows are sometimes used in the composition of the weather
forecast in creative ways such as framing headlines, thereby creating a sense of playfulness and visual
coherence in the forecast. I interpret the divergence away from the meteorologists' register (both its
verbal and graphic parts) as a process in which the reader's demands and needs gained more
momentum in the design of weather forecasts. By using more iconic symbols and adding visual effects,
the forecast becomes more polished and attractive, conforming to advertising style.
This study contributes to the literature on style in science communication. First, I have specialized Van
Leeuwen's (2005) framework to science communication. Second, based on my empirical findings and
existing literature, I have expanded this framework with characteristics pertaining to trust and graphics
(Table 1). The data for this study does not include examples of graphics (maps, graphs, and tables) in
forecasts holding a conversational style. Therefore, the entry for graphics in conversational style is
denoted “informal graphics”. A category that appears to align with the description “informal graphics” is
“humanized data visualizations” (Alamalhodaei et al., 2020) which includes comic-style drawing,
concrete examples and narrative text embedded in the graphics, and labels that explain the meaning of
unfamiliar symbols.
Scientific style
Related terms:
"The style of the expert" (Van
Leeuwen
,
2005);
Conversational style
Related terms:
“conversational style” (Van
Leeuwen
,
2005); “contingent
Advertising style
Related terms:
“advertising style” (Van
Leeuwen
,
2005);
"
empiricist repertoire
"
(Gilbert
and Mulkay, 1984);
"scientific rationality" (Coleman,
1995);
"technical rationality" (Plough and
Krimsky, 1987)
repertoire” (Gilbert
and
Mulkay
,
1984);
"cultural rationality" (Coleman,
1995; Plough and Krimsky, 1987);
"informalization" (Wouters, 1986,
2011)
"
marketization of discourse
"
(Fairclough, 1993)
Social actors Describes social actors in third
person or passive voice
Includes social actors, often in first
and second person.
Direct address from sender
to reader
Data focus Is concerned with accuracy and
uncertainty
Is concerned with the bodily
experience of the weather
Is concerned with consumer
interests
Data
appearance
Appears objective Appears subjective Appears attractive
Evaluation
No evaluative statements
Uses evaluative statements
Uses positive
adjectives
Trust
Reputation of the medium; the
scientific identity; use of scientific
explanations
Reputation of the medium; appeal
to authority; mediated scientific
explanations
Reputation of the medium;
appeal to authority
Verbal
registers
Uses technical language from the
meteorologists' register
Uses vernacular language Uses poetic language
Graphics
Technical graphics
Informal
graphics
Polished graphics
Relation
ships
Top down
Equality
Bottom up
Table 2 Summary of style characteristics
The role of the mathematization and computerization of meteorology
This final section is dedicated to discussing how the findings can be explained. While it is difficult to
assert causal relations, it is possible to reflect and hypothesize about potential links. The focus here is on
how the "quiet revolution" (Bauer et al., 2015: 53; Benjamin et al., 2019; Harper, 2008) of numerical
weather prediction may have affected weather forecasting.
When juxtaposed with the history of weather prediction, the two main findings outlined above appear in
parallel with significant historical events. First, the year 1970 emerges as important in the data material.
By this year, conversational style vanishes from VG despite a general trend of informalization that
accelerated between the 1950s and 1980s (Wouters, 1986, 2011). Similarly, conversational style never
gets a hold in Ap in this period. Further, it is in 1970 that the sender stops providing meteorological
explanations of the predictions despite a period of growing distrust in technocratic expertise (Heymann,
2017: 7). So, what happened that made weather forecasts develop this way? While there may be several
reasons for this, it stands out that the Norwegian Meteorological Institute started using computers in
their day-to-day weather predictions in 1970 (Kristiansen, 2017), thereby implementing computerized
numerical techniques. Therefore, it seems that the mathematization and computerization of
meteorology pushed the development of weather forecasts to a more formalistic style. The change in
strategies for establishing trust suggests that when operational weather prediction was mathematized
and computerized, trust was vested in algorithms and mathematical models as opposed to the skill of
human meteorologists. By omitting meteorological explanations, the science of weather predictions also
becomes opaquer for laypeople. Scientific elaborations of meteorological phenomena still occur after
1970 (e.g., in VG 2015, see Fig 1), but they no longer explain the predictions.
Second, the onset of advertising style around the year 2000 coincides with the establishment of Storm
Weather Center in 1998, the first commercial weather service in Norway and a competitor to the
government-funded meteorological institute (Kalvig, 2009). The introduction of a competing service
appears to have prompted advertising style. According to the company’s founder, Kalvig (2009, see also
Mahroum, 2016), the company's establishment was due to a serendipitous combination of events. Aided
by the available competence in Norway, Storm started to develop numerical modelling techniques early
on (Mahroum, 2016). According to Mahroum, this was a crucial component of their success because it
made them able to deliver weather predictions independently of the governmental services. Thus, Storm
would likely not have succeeded without the advanced state of numerical weather prediction at the
time, and so the mathematization of meteorology seems to have played a role in the advent of
advertising style.
While this study provides some answers, it also opens new questions. For example, are the findings
generalizable across countries, media outlets and other media channels? How have processes of
mathematization and computerization affected other scientific endeavors (see Gingras, 2001)? More
research is needed to better understand how interactions between scientists and laypeople are affected
by scientific developments.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this
article.
References
Alamalhodaei A Alberda AP and Feigenbaum A (2020) Humanizing data through ‘data comics’: An introduction to graphic
medicine and graphic social science. In Engebretsen M and Kennedy H (eds) Data Visualization in Society. Amsterdam:
Amsterdam University Press, pp.347–365. https://doi.org/10.5117/9789463722902_CH21
Bauer MW Petkova K Boyadjieva P and Gornev G (2006) Long-term trends in the public representation of science across the 'iron
curtain': 1946–1995. Social Studies of Science 36(1): 99–131. https://doi.org/10.1177/0306312705053349
Bauer P Thorpe A and Brunet G (2015) The quiet revolution of numerical weather prediction. Nature 525: 47–55.
https://doi.org/10.1038/nature14956
Benjamin SG Brown JM Brunet G Lynch P Saito K and Schlatter TW (2019) 100 years of progress in forecasting and NWP
applications. Meteorological Monographs 59: 13.1–13.67. https://doi.org/10.1175/AMSMONOGRAPHS-D-18-0020.1
Bucchi M (2013) Style in science communication. Public Understanding of Science 22(8): 904–915.
https://doi.org/10.1177/0963662513498202
Coen S Meredith J Woods R and Fernandez A (2020) Talk like an expert: The construction of expertise in news comments
concerning climate change. Public Understanding of Science. Epub ahead of print 30 December.
https://doi.org/10.1177/0963662520981729
Coleman CL (1995) Science, technology and risk coverage of a community conflict. Media, Culture & Society 17(1): 65–79.
https://doi.org/10.1177/016344395017001005
Compton J (2018) When weather forecasters are wrong: Image repair and public rhetoric after severe weather. Science
Communication 40(6): 778–788. https://doi.org/10.1177/1075547018799101
Dash B (2015) Public understanding of cyclone warning in India: Can wind be predicted? Public Understanding of Science 24(8):
970–987. https://doi.org/10.1177/0963662514553203
Fairclough N (1992) Discourse and Social Change. Cambridge: Polity Press.
Fairclough N (1993) Critical discourse analysis and the marketization of discourse: The universities. Discourse & Society 4(2):
133–168. https://www.jstor.org/stable/42888773
Fairclough N (2003) Analyzing Discourse: Textual Analysis for Social Research. London: Routledge.
Few S (2012) Show me the Numbers! Designing Graphs and Tables to Enlighten. El Dorado Hills: Analytics Press.
Gerhards J and Schäfer MS (2009) Two normative models of science in the public sphere: Human genome sequencing in German
and US mass media. Public Understanding of Science 18(4): 437–451. https://doi.org/10.1177/0963662507082891
Giddens A (1990) The Consequences of Modernity. Cambridge: Polity.
Gilbert N and Mulkay M (1984) Opening Pandora's Box: A Sociological Analysis of Scientists' Discourse. Cambridge: Cambridge
University Press.
Goodwin J and Dahlstrom MF (2014) Communication strategies for earning trust in climate change debates. WIREs Climate
Change 5(1): 151–160. https://doi.org/10.1002/wcc.262
Gingras Y (2001) What did mathematics do to physics? History of Science 39(4): 383–416.
https://doi.org/10.1177/007327530103900401
Hamilton LC, Wake, CP, Hartter J, Safford, TG and Puchlopek, AJ (2016) Flood realities, perceptions and the depth of divisions on
climate. Sociology 50(5): 913–933. https://doi.org/10.1177/0038038516648547
Harper KC (2008) Weather by the Numbers: The Genesis of Modern Meteorology. Cambridge: MIT press.
Halliday MAK (1978) Language as Social Semiotic. London: Arnold Edward.
Halliday MAK and Mathiessen CMIM (2014) Halliday's Introduction to Functional Grammar, 4th edn. London: Routledge.
Heymann M (2017) 1970's: Turn of an era in the history of science? Centaurus, 59(1–2): 1–9. https://doi.org/10.1111/1600-
0498.12146
Kalvig SM (2009) Fra lett bris til full storm [From light breeze to whole gale]. Bergen: NHH. Available from
https://openaccess.nhh.no/nhh-
xmlui/bitstream/handle/11250/2437163/Fra%20lett%20bris%20til%20full%20Storm.pdf?sequence=3
Kristiansen TA (2017) Meteorologi på Reise: Veivalg og Impulser i Arnt Eliassen og Ragnar Fjørtofts Forskerkarrierer
[Meteorology on Travel: Crossroads and Impulses in the Research Careers of Arnt Eliassen and Ragnar Fjørtoft]. Unpublished
doctoral dissertation. University of Bergen: University of Bergen Press.
MacEachren A (1995) How Maps Work: Representation, Visualization, and Design. New York: The Guilford Press.
Mahroum S (2016) Black Swan Start-ups: Understanding the Rise of Successful Technology Business in Unlikely Places. Cham:
Springer. https://doi.org/10.1057/978-1-137-57727-6
Molek-Kozakowska K (2017) Stylistic analysis of headlines in science journalism: A case study of New Scientist. Public
Understanding of Science 26(8): 894–907. https://doi.org/10.1177/0963662516637321
Nilsen Y and Vollset M (2016) Vinden Dreier: Meteorologiens Historie i Norge. [The Wind is Turning: The History of Meteorology
in Norway]. Oslo: Scandinavian Academic Press.
Plough A and Krimsky S (1987) The emergence of risk communication studies: Social and political context. Science, Technology,
& Human values 12(3–4): 4–10. https://www.jstor.org/stable/689375
Randalls S (2010) Weather profits: Weather derivatives and the commercialization of meteorology. Social Studies of Science
40(5): 705–730. https://doi.org/10.1177/0306312710378448
Van Leeuwen T (2005) Introducing Social Semiotics. London, England: Routledge.
Van Leeuwen T (2008) Discourse and Practice: New Tools for Critical Discourse Analysis. London, England: Routledge.
Vannini P and McCright AM (2007) Technologies of the sky: A socio-semiotic analysis of televised weather discourse. Critical
Discourse Studies 4(1): 49–73. https://doi.org/10.1080/17405900601149475
Wilson K (2008) Television weathercasters as science communicators. Public Understanding of Science 17(1): 73–87.
https://doi.org/10.1177/0963662506065557
Wiik A (2020) Trends in everyday mathematics: The case of newspaper weather forecasts. Manuscript submitted for publication.
World Meteorological Association (1992) Manual on the Global Data-Processing and Forecasting System. (Volume 1, supplement
11). Retrieved from: https://www.wmo.int/pages/prog/www/DPS/Publications/WMO_485_Vol_I.pdf
World Meteorological Association (2017) Manual on Codes: International Codes (Volume I.1., Annex II to the WHO Technical
Regulations, Part A). Retrieved from
https://www.wmo.int/pages/prog/www/WMOCodes/WMO306_vI1/Publications/2017update/WMO306_vI1_2011UP2017_
en.pdf
Wouters C (1986) Formalization and informalization: Changing tension balances in civilizing processes. Theory Culture & Society
3(2): 1–18. https://doi.org/10.1177/0263276486003002002
Wouters C (2011) How civilizing processes continued: Towards an informalization of manners and a third nature personality. The
Sociological Review 59(1): 140–159. https://doi.org/10.1111/j.1467-954X.2011.01982.x
Paper III
Wiik, A., Vos, P., & Engebretsen, M. (2022). Visual-numeric literacy: The case of
COVID-19 data visualizations for news media and their expectations on readers.
[Manuscript submitted for publication].
Anders Wiik (corresponding author) 1
Anders Wiik is PhD research fellow in mathematics education at 2
the University of Agder, Norway. His research interests include 3
data visualizations in public discourse, everyday mathematics and 4
the social role of mathematics. 5
anders.wiik@uia.no 6
Pauline Vos 7
Pauline Vos is professor of mathematics education at University of 8
Agder, Norway. She leads research that relates mathematics 9
education to mathematical models, visualization, inquiry-based and 10
project-based learning, relevance and authenticity. She is interested 11
in questions about the why, when and how of learning and using 12
mathematics (and what mathematics?) by which students or users. 13
pauline.vos@uia.no 14
Martin Engebretsen 15
Martin Engebretsen is professor of language and communication at 16
the University of Agder, Norway. His research interests include 17
multimodal and visual studies, rhetorics, discourse analysis and 18
journalism studies. He is particularly interested in the exploration 19
of genre development and literacy related to processes of 20
digitalization in the public media. 21
martin.engebretsen@uia.no 22
Abstract 23
This study aimed at characterizing what is expected of readers of 24
data visualizations (graphs, charts) in news media. Using a socio-25
semiotic perspective, we conceptualized the capacity to read and 26
make sense of such data visualizations as visual-numeric literacy 27
which is part of mathematical literacy. We developed an analytic 28
framework and applied it to a sample of COVID-19 data 29
visualizations from a popular online newspaper. Our findings 30
showed that visual-numeric literacy entails decoding labels, colors, 31
scales, relative numbers, etc., but also meaningful and critical 32
engagement as active and reflective participants in society. We 33
discuss implications for mathematics education. 34
Visual-numeric literacy: The case of COVID-19 data 35
visualizations for news media and their expectations on 36
readers 37
An important question in mathematics education is what kind of 38
mathematical literacy people need in their lives (Geiger et al., 39
2015). Part of mathematical literacy is the capacity to read and 40
make sense of data visualizations (DVs) such as graphs and charts. 41
DVs are visual representations of quantitative data and have a 42
growing presence in the public sphere (Engebretsen & Kennedy, 43
2020). DVs can compactly convey quantitative information and 44
make otherwise obscure patterns visible, but they also pose 45
challenges to readers and designers (Kirk, 2019). 46
In 2020, the COVID-19 pandemic profoundly affected the lives of 47
citizens across the globe. News media used DVs to report on 48
infection rates, mortality rates, geographic distribution, and so on 49
(Aguilar & Castaneda, 2021; Kwon et al., 2021). This raised the 50
question of what expectations the journalistic DVs put on readers 51
in the context of urgent social events. To gain insight into the 52
mathematical literacy needed for reading and making sense of DVs 53
in news media, we selected a journalistic webpage rich in 54
pandemic DVs and investigated the expectations this collection put 55
on readers. 56
Following Geiger et al. (2015), mathematical literacy includes 57
capabilities to use mathematics in a range of personal, social and 58
professional contexts and it can contribute to informed and critical 59
citizenship. Thus, the capacity to be active and reflective regarding 60
quantitative information offered visually is an important aspect 61
when looking at the expectations that DVs put on readers. 62
However, as we will explain in the literature review, many 63
frameworks in the mathematics education literature focus more on 64
the reading and designing of DVs, and less on reader actions and 65
reflections. We found one exception, which was the concept 66
visual-numeric literacy (VNL) (Tønnessen, 2020) that 67
accommodates these dimensions. In this paper, we develop this 68
concept further and use it for our analysis of expectations by DVs. 69
The article is structured as follows: in the literature review we 70
present research relevant to reading and meaning making of DVs 71
and highlight how the concept of VNL contributes to the field. 72
Next, we present the concept, its theoretical underpinnings, and an 73
analytic framework. The methods section gives an elaboration of 74
how data were collected and analyzed, followed by an exemplary 75
analysis of Figure 1, illustrating the application of the analytic
76
framework. Thereafter, we discuss and synthesize the expectations
77
that the DVs put on readers. Finally, we discuss theoretical and
78
educational implications. We ask:
79
RQ: What characterizes the visual-numeric literacy expected of
80
readers of COVID-19 data visualizations in online news media?
81
By expectations, we mean the capabilities that readers need to read
82
and make sense of the DVs. For example, when a data set is
83
presented to the readers in multiple formats, it is expected that the
84
readers can make sense of at least one of the formats. An analysis
85
of the mathematical literacy expected on citizens by news media
86
can inform a critical discussion of whether mathematics curricula
87
prepare students for a certain aspect of everyday life, namely to be
88
informed, reflective and critical about urgent social matters.
89
90
Figure 1. DV of registered COVID-19-infections (upper curve)
91
and mortality (lower curve) in Norway (Nye = new, endring av
92
testkriterier = change of testing criteria). Downloaded from vg.no
93
27.03.2020. Reprinted with permission.
94
95
Figure 2a. DV of the countries where Norwegians were infected
96
(Areal = Area; Søyle = Bar; Ikke avklart = not clarified).
97
Downloaded from vg.no 08.04.2020. Reprinted with permission.
98
99
Figure 2b. Percentual DV of the countries where Norwegians were
100
infected (Areal = Area; Søyle = Bar; Ikke avklart = not clarified).
101
Downloaded from vg.no 08.04.2020. Reprinted with permission.
102
Literature review
103
Mathematical literacy has a variety of definitions (Geiger et al.,
104
2015). An often-cited definition says:
105
Mathematical literacy is an individual's capacity to
106
formulate, employ and interpret mathematics in a variety of
107
contexts. It includes reasoning mathematically and using
108
mathematical concepts, procedures, facts, and tools to
109
describe, explain and predict phenomena. It assists 110
individuals to recognize the role that mathematics plays in 111
the world and to make well-founded judgments and 112
decisions needed by constructive, engaged and reflective 113
citizens. (OECD, 2016, p. 65) 114
The concept of mathematical literacy has helped researchers to 115
observe that mathematics teaching in schools focuses largely on 116
procedures, and hardly on other aspects of mathematical literacy, 117
despite these being prescribed by the curriculum (Bolstad, 2020). 118
Also, the concept points at a need for research on what 119
mathematics is needed across personal, social and work contexts, 120
and how this should be taught (Niss & Jablonka, 2020). For 121
instance, making quick but accurate estimations is an important 122
out-of-school skill, yet not addressed in curricula (Sunde et al., 123
2021). We contend that the same holds for the critical and 124
reflective reading of DVs in news media. 125
There is ample research on visualization in mathematics education, 126
in which we discern different streams. A first stream follows the 127
line of Presmeg (2006) in focusing on mental visualizations in 128
mathematics learning, that is on imagery within the mind. Another 129
stream studies the role of graphs in relation to the learning of 130
symbolic expressions, whereby the spatial visualization on paper 131
or screen enriches the ways mathematics can be represented 132
(O’Halloran, 2015). We follow a third stream of visualizations 133
research focusing exclusively on graphical representations, such as 134
Roth (2003), who found that expert scientists who were competent 135
in interpreting graphs in their field of expertise struggled to 136
interpret graphs from other fields and contexts. A concept used by 137
researchers in this stream is graph comprehension (Curcio, 1987), 138
which consists of capabilities to read the data (e.g., how many 139
infections at what date?), to read between the data (e.g., how much 140
did the infections increase between two dates?), and read beyond 141
the data (e.g., predict how the infections will develop). A related 142
concept is graphicacy (Olande, 2013; Åberg-Bengtsson & Ottoson, 143
2006), which extends graph comprehension with the capability to 144
criticize the design of a graph, for example when y-axes do not 145
start from zero in order to magnify differences between categories. 146
In this stream, most studies administer tasks on sector diagrams, 147
line graphs or bar charts with constructed contexts to young 148
students and find that the reading of such DVs is complex. Our 149
research aims to contribute to this stream regarding (1) reading and 150
meaning making in informal situations without prompts and 151
scaffolds from a task, and (2) the potential for critical reflection 152
regarding the quality, provenance and trustworthiness of the data 153
visualized. 154
Besides the concepts graphicacy and graph comprehension, 155
another related concept is statistical literacy, which concerns 156
capabilities to use statistical concepts, such as mean and 157
variability, in various contexts, and to design data collections and 158
present results thereof (Gal, 2002). Our study regarding 159
expectations put on readers by DVs can be perceived as being on 160
statistical literacy. However, rather than taking a disciplinary focus 161
on statistical concepts, we started from artefacts found in social 162
communication to convey journalistic information on urgent 163
matters based on data collected by health authorities. 164
Some recent studies have investigated journalistic DVs. Aguilar 165
and Castaneda (2021) and Kwon et al. (2021) analyzed DVs used 166
in the media coverage of the COVID-19 pandemic, finding that 167
these expected citizens to have an advanced mathematical level, in 168
some cases even higher than mandatory school mathematics. These 169
studies only focused on interpreting and did not consider critical 170
reflection and goal-oriented action. Tønnessen (2020) and Tvedt 171
(2020) investigated how students in the subject of Civics read and 172
made sense of a complex bubble plot about poverty showing multi-173
variate demographic data in a logarithmic plot. The participants in 174
Tønnessen’s (2020) study (17-18 years old) needed significant 175
teacher guidance to succeed in the action aspect of literacy and 176
showed few signs of reflection literacy. Tvedt (2020) used a 177
simplified version of the same DV, showing that a DV design 178
reducing cognitive load enabled her participants (15-16 years old) 179
to make sense of the logarithmic scale and critically reflect on the 180
trustworthiness of the information presented. Vos and Frejd (2020) 181
found that 8th grade students were able to make sense of and draw 182
Sankey diagrams. When asked to mention properties of a Sankey 183
diagram, the students emphasized what they can be used for and 184
not their mathematical properties. This showed that novices can 185
confidently reflect on DVs without focus on mathematical 186
properties. Mulligan (2015) explored how 21 highly able 1st grade 187
students designed DVs of data collected themselves. Over the 188
course of one year, the students showed impressive development in 189
creating own designs and strong critical reflection, for example, on 190
how to misrepresent their data by manipulating scales. 191
In the above-mentioned studies, researchers indicate that the 192
capability to read and create DVs involves familiarity with some 193
design conventions, such as scales increasing from left to right, and 194
the variable ‘time’ taken horizontally. Some sources also 195
emphasize capabilities to reflect, analyze and critique DVs, and 196
some mention their social roles. In the next section, we present our 197
conceptualization of the special literacy that is expected by DV’s 198
in the public sphere, which integrates reading, sense making, using 199
for informed action, and critical reflection. 200
Theoretical framework 201
Our perspective on reading and making sense of DVs does not start 202
from mathematical concepts, procedures, facts, and tools. We rather 203
take a perspective on reading and making sense of signs, in their 204
widest sense, by readers with a personal, social or work-related need 205
for doing so. We build on Hasan (1996), who critiqued school 206
learning of language for focusing mainly on recognizing and 207
reproducing signs and rules of grammar and decoding units of texts 208
with little meaning in the learners’ lives and neglecting their 209
capabilities to act and reflect. Hasan combined two theoretical 210
perspectives. The first was social semiotic theory of communication 211
(Kress, 2003; Halliday, 1978), which investigates meaning-making 212
in social social, cultural and historical contexts. The second was 213
socio-cultural theory of learning (Vygotsky, 1986), which describes 214
dialectics between mental activities and socio-cultural contexts. 215
Hasan’s (1996) resulting concept of literacy integrates aspects of 216
recognition, action and reflection. Tønnessen (2020) adapted 217
Hasan’s concept to the reading and sense making of DVs and coined 218
it visual-numeric literacy (VNL). This concept is named for the 219
visual and numeric modes, which are characteristic to DVs. Because 220
the numeric mode is mathematical, VNL is a form of mathematical 221
literacy. In this paper we aim to develop the concept further and 222
validate it with an analysis of COVID-19 DVs from news media. 223
Visual-numeric literacy is a concept that captures peoples’ capacity 224
to engage in communication involving DVs. It consists of three 225
sub-literacies. First, recognition literacy is the capacity to 226
recognize relevant semiotic resources and their meanings, such as 227
lines, bars and colors. Second, action literacy concerns people’s 228
ability “to do something with their language” (Hasan, 1996, p. 229
399) and “be active participants in society” (Tønnessen, 2020, p. 230
191) such as making sense of DVs in social contexts and using or 231
designing DVs to reach personal goals. Finally, reflection literacy 232
is about critically challenging existing DV conventions and their 233
provenance. Core elements are reflection, enquiry and critique, 234
such as assessing the trustworthiness of the data. The three sub-235
literacies of VNL are intertwined (Hasan, 1996; Tønnessen, 2020). 236
VNL is not a purely intellectual capacity. Recent studies have 237
shown how engaging with data is also an emotional and relational 238
activity. Emotional engagement can be both an obstacle and 239
advantage in working with numbers (Kennedy & Hill, 2018; Pinel 240
et al., 2020). 241
Following Hasan (1996), there are four tenets underlying VNL. The 242
first tenet is that a material element (cartesian graphs, bar charts, 243
etc.) becomes a sign when it can be interpreted, and vice versa: the 244
material elements that can be interpreted are signs. These elements 245
and their components (dots, lines, colors, etc.) are called semiotic 246
resources. The immaterial rules for the systematic connection 247
between semiotic resources and meaning are called codes. For 248
example, in Figure 1, colors are a semiotic resource and the code 249
that translates colors into meaning is elaborated by the legend above 250
the DV. Second, meaning making is both a formal and a social act. 251
To illustrate, consider the interpretation of a cartesian graph 252
displaying COVID-19 statistics as in Figure 1. The graph accrues 253
meaning in a formal sign system, that is, in the system of cartesian 254
coordinates and their conventions, which were established through 255
cultural-historical processes. Also, the graph accrues meaning 256
through participation in social events, such as the COVID-19 257
pandemic. Third, the interpretation of a sign differs between human 258
actors because they have different life stories and different personal 259
goals. Finally, different sign systems interact: DVs are typically 260
used together with verbal text, colors and other sign systems to 261
constitute a complex semiotic whole. 262
We have developed the framework so that it can be used for 263
analyzing DVs by integrating analytic concepts to each aspect. In 264
recognition literacy, the focus is on semiotic resources and codes. 265
For semiotic resources, we look for: 266
Lines 267
Bars 268
Dots 269
Colors 270
Written words 271
Written numbers 272
We were interested in the accessibility of the codes that give 273
meaning to these semiotic resources. For analyzing accessibility, we 274
used Bernstein’s (1981) concepts of elaborated and restricted 275
codes. Restricted codes are used with little or no explanation of the 276
intended meaning. Such codes are characteristic of contexts where 277
designers assume that readers understand the meanings – although 278
this assumption may be mistaken. Elaborated codes is unpacked by 279
the designer, for example through legends or short written texts. To 280
assess whether a code is more elaborated or more restricted, we 281
considered: 282
Elaborating texts (e.g., from mouseovers) 283
Legends 284
Adherence to conventions (e.g., red = bad, higher position = 285
higher number) 286
Next, in action literacy we were interested in the social and 287
communicative actions the page invites for. This pertains both to 288
actions within the page (e.g., scrolling, toggling) and the intended 289
social functions of the DVs (e.g., the social action that the DVs call 290
for). We used the following concepts from Kirk (2019): 291
Angle of analysis refers to the aspects of the data that a DV 292
foregrounds. For example, a choropleth map foregrounds the 293
geospatial distribution of a variable. 294
Framing refers to the selection of data included in a DV 295
Focus refers to elements that stand out, for example through 296
conspicuous colors 297
Explanatory DVs are designed to highlight a particular 298
message. In contrast, exploratory DVs are designed to 299
support readers to make their own interpretations. 300
These concepts enabled us to assess the intended messages of the 301
DVs and indicate the action literacy expectations that are put on 302
readers. 303
Third, in recognition literacy we were interested in the reflective and 304
critical engagement that the page invited fer. Here, we avoid the 305
term ‘expectations’ because readers can appreciate the content 306
without being reflective. Our focus was on issues of trust, and we 307
analyzed the page for cues pertaining to: 308
Are there missing data? 309
How and by whom were the data collected? 310
Have the data been modified (e.g., smoothening of curves)? 311
Are there errors in the data or their representation, and how 312
are errors handled? 313
Methods 314
Our methods were based on social semiotics (e.g., Aiello, 2020), 315
consisting of compiling a corpus of relevant DVs, iteratively 316
reading the corpus to get familiar with the content, and systematic 317
analysis based on the theoretical framework. 318
Context 319
The data used in this study was collected during the first wave of 320
the COVID-19 pandemic in March and April 2020. The first 321
confirmed case in Norway was in February 2020, and the country 322
entered lockdown in March. In Norway, the first wave of the 323
pandemic peaked in the transition from March to April 2020, at a 324
time when health authorities still had limited knowledge of the 325
virus. 326
Data collection 327
We decided to use journalistic COVID-19-related DVs from VG 328
(Verdens Gang) for our analysis because this is the most read 329
online newspaper in Norway (Mediebedriftene, 2020). From the 330
early days of the pandemic, they developed a web page where they 331
compiled COVID-19-related DVs. We regarded this web page as 332
the most developed Norwegian resource aimed at a general 333
audience. The page was launched in the beginning of March 2020. 334
We captured all content on the page on three occasions during the 335
first wave of the pandemic in 2020. We limited the analysis to the 336
DVs that were present on all three captures. This criterion left us 337
with a corpus of 24 DVs: 338
10 line graphs 339
3 bar charts 340
2 choropleth maps 341
1 histogram 342
1 sector diagram 343
1 Sankey diagram 344
3 DVs where readers can choose between line graph and 345
histogram 346
2 DVs where readers can choose between area graph and 347
histogram 348
1 DV where readers can choose between area graph, line 349
graph and histogram 350
Of these, we have selected to display four. Colors are not 351
reproduced here. Figure 1 is a line graph, the most common DV 352
format in the sample. Note the option menu at the top, where 353
readers can change between cumulative and new numbers; and 354
linear and logarithmic vertical axis. Figure 2 is an area graph, 355
shown as relative (Figure 2a) and absolute (Figure 2b) numbers. 356
Readers can also change it to a histogram. Figure 3 is one of the 357
two choropleth maps on the page. These maps were large and 358
colored and therefore salient. Figure 4 is a histogram, the second 359
most frequent DV format used on the page. It appears with 360
logarithmic vertical axis. The menu at the top includes options to 361
see cumulative or new numbers; bars (histogram) or line graph; 362
and logarithmic or linear vertical axis. 363
Data analysis 364
The data analysis started with all authors reading the corpus 365
multiple times. Next, the DVs were analyzed one by one for each 366
sub-literacy of VNL. We registered these and additionally noted 367
whether the page offered elaboration, for example in mouseovers. 368
Also, the sub-literacy reflection involves making connections 369
between DVs and other pieces of information on the page. Hence it 370
was necessary to take the whole corpus into consideration for this 371
analysis. 372
Exemplary analysis 373
Now we show how the analytic framework was applied to Figure 374
1. We start with recognition literacy. Readers are expected to 375
identify the semiotic resources and the restricted codes used 376
(Bernstein, 1981). The semiotic resources used in this line graph 377
are straight and curved graphical lines, numbers, words and colors. 378
These become meaningful by the way they are arranged in a 379
cartesian coordinate system. Thus, readers are expected to 380
understand the principles of the coordinate system and the relation 381
between the curves and the variables they represent. By changing 382
the vertical axis to logarithmic, new and less common codes are 383
introduced. In this DV, the logarithmic axis is not verbally 384
elaborated. The logarithmic display is not the default option, 385
suggesting that this alternative view targets readers who are 386
comfortable with this format and want to explore the data further. 387
Now we turn to action literacy. The readers are expected to 388
appreciate the situational meaning of the DV. Using Kirk’s (2019) 389
concepts, they need to understand the following: The angle of 390
analysis can be expressed as “how have infections and mortality 391
rates changed over time?” The frame here is the complete data set
392
for Norway, ranging from the first confirmed case in February
393
until the most recent date. The dotted vertical line creates a focal
394
point in the middle of the DV. With the linear vertical axis as
395
shown in Figure 1, infection rate is in focus because the mortality
396
curve merges with the horizontal axis. When the display is
397
switched to logarithmic, both curves are in focus. The end-point
398
labels draw focus to the most recent day. This DV presents the
399
readers with options concerning how to see the data and can use
400
mouseovers to see exact numbers. It does not present pre-
401
determined conclusions: is the infection rate out of control? Is the
402
mortality rate acceptable? Therefore, we consider this DV
403
exploratory, meaning that readers were expected to contextualize
404
and evaluate the data themselves.
405
Finally, we look at reflection literacy. The readers are invited to
406
reflect on the trustworthiness of the data and their representation.
407
The headline above Figure 1 on the page said “Registered infected
408
and dead”, there is a brief text explaining that there can be a large
409
number of missing data, and the vertical dotted line shows that the
410
testing criteria changed. These are all cues suggesting that there are
411
missing data. The brief text also explains that the data were
412
compiled by the Norwegian Institute of Public Health (NIPH), and
413
explains that the data were retrieved from reports from medical
414
doctors and laboratories. Thus, the sources and compilation
415
method were made transparent. There is a log of changes and
416
errors where some entries can relate to this DV. For example, a
417
death that was erroneously counted twice was corrected on 6 April.
418
419
Figure 3. Choropleth map of infection rates (Døde = Dead;
420
Smittet = Infected; Antall smittede per 100 000 innb. = Number
421
of infected per 100 000 inhab.). Downloaded from vg.no
422
08.04.2020. Reprinted with permission.
423
424
Figure 4. DV showing the total development for infections and
425
mortality. (Dag for dag: utviklingen i verden = Day by day: the
426
development in the world; Søyle = Bar; Linje = Line; Stor endring
427
her: Kina endret diagnose-kriterier = Major change here: China
428
changed the criteria for diagnosis). Downloaded from vg.no
429
21.042020. Reprinted with permission.
430
Results
431
Now we present findings emerging from the analysis.
432
Recognition literacy
433
Focusing on the recognition literacy expected from the readers of
434
these DVs, the most common DV type in our corpus is line graphs
435
in cartesian coordinate systems, with the horizontal axis showing
436
time and the vertical showing relative or absolute numbers of
437
people in some category such as infected, dead or hospitalized.
438
DVs of this type expect readers to recognize the basic semiotic
439
resources involved (lines, words, numbers, colors) and their
440
intended meanings (e.g., that the position of lines and points
441
relative to the axes signifies the value they represent, or that an
442
upward sloping curve means that the number is increasing). Some
443
of the graphs include additional semiotic resources that can be
444
decoded within the sign system of cartesian coordinates, such as 445
the vertical dotted line in Figure 1 labeled “change in test criteria”. 446
Some of the line graphs have multiple curves where each curve 447
represents a different category (e.g., infected and deaths, Figure 1). 448
The area graphs and histograms are always organized in cartesian 449
coordinate systems with time on the horizontal axis. In some area 450
graphs, the readers can switch to stacked histogram or line graph. 451
The histogram bars have uniform width and the readers can choose 452
between seeing absolute or relative numbers in three of the four 453
histograms. Absolute numbers can be retrieved through 454
mouseovers, with text such as “4 April. Registered infected: 455
5.493”. Hence, readers are expected to recognize bars, areas, words 456
and numbers, decode their meanings (from location in the 457
coordinate system, from mouseovers) and compare across bars and 458
areas (e.g., locate when the number is greater). 459
For bar charts, the readers were expected to decode the relevant 460
category from the labels on the horizontal axis, compare the 461
magnitudes of the categories through the height of the bars, and 462
find the numbers for each category by reading it off. 463
The two choropleth maps showed an outline of geographical 464
regions, Norway and the world respectively, where municipalities 465
and countries are color coded. By default, the color codes show 466
mortality rates per 100 000 inhabitants, and readers can change it 467
to infection rates. There are legends elaborating the meaning of 468
each color, and higher infection rates are represented by darker 469
shades of red. By hovering the cursor over a municipality or a 470
country, a mouseover with absolute and relative numbers appears. 471
Thus, these DVs expect the readers to locate geographical regions 472
in the map and compare infection rates and mortality rates by 473
comparing colors. To find numerical values, the readers must use 474
the mouseovers or match the color code with the legend. 475
The sector diagram shows gender distribution of infected people, 476
and the sectors are labeled with gender category and percentage. 477
This DV expected that the reader can read the labels, understand 478
the meaning of percent, and visually compare sectors. 479
The Sankey diagram connected countries to Norwegian counties 480
and shows where in the world people were infected and in what 481
county they live. Each country and each county were labeled with 482
the number of infected. This DV expects that readers understand 483
that the lines crossing from country to county represent a group of 484
infected people, and that the width of each line is proportional to 485
the number of infected people. 486
Some color codes are used in multiple DVs, making them easier to 487
compare and connect. For example, a bar chart, an area graph and 488
the Sankey diagram, all showing where people got infected, use the 489
same color codes. 490
Three of the DVs in the sample include an option to make the 491
vertical axis logarithmic. Below one of these DVs there is an 492
expandable text box titled “why logarithmic scales?”. The text box 493
elaborated that exponential growth appears as a straight line in log 494
plots, and therefore that deviations from exponential growth are 495
easier to uncover. The practical meaning of logarithmic scales can 496
be understood by studying the labels on the axis. Hence, the page 497
offers some elaboration to the mathematical code of logarithmic 498
scales. However, despite the elaboration, there are also 499
opportunities for misreading. For example, the DV in Figure 4 can 500
be interpreted as showing a flattening of infection and mortality 501
rates when there is a roughly linear increase. Because the log plots 502
are optional, we can not conclude that DV expects readers to 503
decode these. 504
Other mathematical concepts used on the page are absolute and 505
relative numbers (percent, per mille and per 100 000), daily and 506
cumulative numbers, and mathematical models based on the basic 507
reproduction number. These concepts are restricted, suggesting 508
that the readers are expected to understand them without further 509
elaboration. 510
To sum it up, there are four main types of information in our 511
corpus that readers are expected to recognize: (1) how much or 512
how many, (2) at what moment, (3) for what category, and (4) how 513
the quantities change over time. It is expected that the readers can 514
decode the restricted codes without further support, such as relative 515
numbers and coordinate systems. Other codes are partly 516
elaborated, such as logarithmic scales, indicating that readers are 517
expected to be capable of decoding these with some support. Some 518
codes are recurring, such as some of the color codes, which gives 519
the readers further support to decode across DVs. 520
Action literacy 521
Now we will focus on the action aspect of VNL, which involves 522
making sense of the DVs as situated social action. This pertains to 523
using the DVs to achieve personal goals; to find the desired 524
information on the page and use that information to inform some 525
sort of action. This process involves key skills such as identifying 526
the intended function and effect of the communication act, 527
identifying the origins of the DVs, and identifying the social actors 528
involved (senders, receivers, data sources, etc.). 529
The page under investigation is a commercial journalistic product. 530
The DVs on the page appear objective and scientific, similar to the 531
verbal elements, which are held in an objective, informative and 532
impersonal style. In this stage of the pandemic, VG played the role 533
of being the most updated national information provider and, at the 534
same time, acting as the people’s watchdog to the government. 535
Following Kirk (2019), to contextualize and understand the 536
intended meaning of the DVs, readers were expected to appreciate 537
the angle of analysis, the frame, the focus of the DVs and realize 538
whether they are explanatory or exploratory. 539
The DVs on the page provide many angles of analysis that shed 540
light on different aspects of the pandemic. The angles include 541
infection and mortality rates across geographic regions (e.g., 542
Figure 3); infection and mortality rates over time (Figures 1, 4); 543
age distribution over time; gender distribution over time; the 544
country where people were infected; and more. Some DVs include 545
options that enabled readers to change the angle. For example, in 546
the global choropleth map readers can choose between having 547
mortality or infection rates visualized. The amount and diversity of 548
angles suggest that readers were expected to make their own 549
assessment of what DVs were most relevant to them. 550
Now we turn to framing. The DVs that have an axis for time 551
always start at the first known case, or at the first day that the 552
displayed statistic made sense. Many of the categorial data sets 553
have a category for unknowns, for example, unknown age category 554
for recorded deaths or unknown country of infection. This suggests 555
that VG has included all the available data, including incomplete 556
data. In some of the DVs, readers are offered opportunities to 557
change the framing, for example by choosing between total 558
numbers or new numbers, that is, the number of new cases for each 559
day (e.g., Figure 1, 4) or by zooming in on a region in the maps 560
(Figure 3). The different framing choices enables readers to 561
personalize and compare different parts or aspects of the data, 562
given that they identify and understand the interactive options 563
presented. Hence, to contextualize the DVs, readers need to 564
understand that VG generally included all available data. 565
Some of the DVs in the corpus have design features that create 566
specific foci. The vertical dotted lines in Figure 1 and Figure 4 567
draw attention to specific spots in the DVs. The multiple line graph 568
showing hospitalizations, number of people on intensive care and 569
on respirator includes similar vertical dotted lines, indicating 570
changes in mitigation measures. The use of color also creates some 571
focal points in the DVs. Generally, VG used red and black for DVs 572
about infection and mortality rates and less conspicuous colors 573
(pale blue, pale violet, pale green, yellow, brown) for other 574
statistics such as age distribution and gender distribution. This 575
way, the DVs showing infection and mortality rates attract more 576
focus than other DVs. In the choropleth maps, regions with high 577
infection and mortality rates are color coded with dark shades of 578
red, and regions with low rates are colored white or pale orange. 579
This way, regions with high rates become focal points. In Figure 1, 580
the linear plot puts the infections in focus and the logarithmic plot 581
puts both curves in focus. By hovering the cursor over one of the 582
curves, the chosen curve is highlighted by reducing the color 583
saturation of the other curves, and a mouseover appears, showing 584
the date and numbers for that day. The same focusing technique is 585
used on several DVs. To understand the intended situational 586
meaning of the DVs, readers need to appreciate these foci. 587
Overall, the DVs in the corpus present readers with empirical facts, 588
offering no cues to the kinds of evaluations to make. it is up to the 589
readers to tell if the data means that Norway need harsher 590
regulations or not and to assess whether the regulations have been 591
effective Further, readers were often given interactive features that 592
enable them to explore and personalize the view. Therefore, we 593
consider the DVs to be exploratory. However, there are some 594
design choices which suggest that certain evaluations are more 595
valid than others. For example, the use of red to indicate high 596
infection and mortality rates, signals that these are negative 597
categories. Overall, we can conclude that readers are expected to 598
explore the DVs to find answers to questions they are curious 599
about and make their own assessments about the state of the 600
pandemic. 601
Reflection literacy 602
Reflection literacy pertains to critical enquiry concerning sources, 603
bias, trust and so forth. Here, we present content from the page that 604
gives cues to these aspects, and the skills readers need to identify 605
and reflect on them. 606
First, we look at cues that can inform readers about the 607
trustworthiness of the data. Critical and analytical readers may 608
notice that the number of tested persons is only a small proportion 609
of the population. Also, the number of tested and the number of 610
infected follow different patterns, indicating that there are 611
significant missing data. This possibility is acknowledged by VG 612
in writing: “Not everyone is tested, so there can be a large amount 613
of missing data” below the DV showing number of infected. VG 614
clearly informs about their sources, which are the NIPH, the 615
European Centre for Disease Prevention and Control, a selection of 616
international news organizations, and VGs own research. At the 617
bottom of the page, there is a log of updates and corrections. This 618
shows that the page is prone to errors, and that VG's journalists are 619
working to correct these and are willing to share their wrongs as 620
they are discovered. VG remarks that they do not show statistics 621
for recovered patients because there is no official register for such 622
data. They also noted that the numbers shown on their page may be 623
higher than numbers presented by others, because VG update their 624
data more frequently than the official sources do. In the projections 625
for number of hospitalized, the R-numbers that the respective 626
projections are based upon, are stated. 627
Hence, although VG’s staff do not give a full disclosure about the 628
mathematical models used, they do offer some details. These 629
pieces of information shed light on the processes of collecting and 630
handling data, uncertainty, design choices and choices of which 631
data to show. This shows that VG invited readers to be critical and 632
reflective about the page and its contents. 633
Discussion 634
The aim of this study was to characterize the VNL that is expected 635
of readers of COVID-19 DVs in online news media. We 636
approached this by developing an analytic framework tailored for 637
analysis of journalistic DVs in digital media and analyzing a 638
corpus of COVID-19-related DVs from the most read online 639
Norwegian newspaper, VG. A limitation is that we only 640
investigated one newspaper from one country. In the corpus, line 641
graphs, bar charts and histograms are most frequent DV types. 642
There were also area graphs, choropleth maps, a sector diagram 643
and a Sankey diagram. Further, the DVs touch upon the 644
mathematical concepts of logarithmic scales, relative vs absolute 645
numbers, and daily vs cumulative numbers. These DV types and 646
concepts are used with limited or no elaboration, suggesting that 647
VG expected the readers to be able to decode these DVs with little 648
support. 649
According to Aguilar and Castaneda (2021) and Kwon et al. 650
(2021), the media coverage of COVID-19 raised mathematical 651
expectations beyond mandatory schooling. The mathematics 652
curriculum in Norway does not specify what kinds of DVs students 653
should encounter in school (Utdanningsdirektoratet, 2019), making 654
it difficult to judge whether the school system prepares students for 655
the kinds of DVs analyzed in this paper. Because the use of DVs in 656
the news, at workplaces, in education and in governmental 657
campaigns is rapidly evolving (Engebretsen & Kennedy, 2020), it 658
is impossible to predict the VNL that will be expected of citizens 659
in the future. Therefore, we suggest that the role of mathematics 660
education towards VNL should be to enable students to participate 661
in discourses involving DVs as active and reflective readers with a 662
sufficiently broad recognition literacy. This way, mathematics 663
teachers can give students the opportunity to engage in lifelong 664
learning. The present study documents some of the expectations 665
raised in society today. 666
VG’s decision to use DVs to convey COVID-19 data implies that 667
they challenged their readers’ mathematical literacy. But we also 668
observed that VG elaborated some of the most challenging codes, 669
which made the contents accessible to a broader audience. These 670
elaborations can be considered a didactization, where readers are 671
guided to make sense of the DVs. Thus, the page offered readers 672
opportunities to learn about the pandemic while developing their 673
VNL. 674
Researchers have found that when children and adolescents collect 675
data themselves and the data is interesting and meaningful to them, 676
they can learn to model phenomena with DVs effectively, and 677
critically reflect on how DVs can be manipulated (Mulligan, 2015; 678
Vos & Frejd, 2020). However, other studies suggest that students 679
may need guidance to succeed in the action and reflection aspect of 680
literacy (Tvedt, 2020; Tønnessen, 2020). Because COVID-19 was 681
an unprecedented event that showcased how mathematical literacy 682
can be an important part of citizenship, COVID-19 DVs are likely 683
to be relevant and meaningful to many people. Thus, they represent 684
didactical opportunities. The DVs analyzed in this article can be 685
used to teach about various DV formats and mathematical models 686
as well as concepts such as exponential growth, logarithms, 687
relative numbers and cumulative numbers. Data pertaining to 688
COVID-19, or other socially relevant events, can also be used for 689
teaching DV design. We suggest that teaching for VNL should not 690
be limited to the recognition aspect, but also guide students to 691
explore the action and reflection aspect of literacy. 692
Concluding remarks 693
What have we learned about the VNL that is expected of readers of 694
COVID-19 DVs in news media? Our analysis suggests that readers 695
are expected to recognize a wide range of DVs that involve many 696
mathematical concepts. The recognition aspect mainly amounts to 697
decoding facts such as how much or how many, at what moment, 698
for what category and how magnitudes changed over time. 699
Concerning the action aspect, readers are expected to understand 700
the intended social functions of the DVs. We found that the DVs in 701
the corpus gave a range of angles, frames and foci to the data, and 702
they were, for the most part, exploratory, meaning that readers 703
were invited to make use interactive features to explore the data 704
and make their own interpretations. Finally, regarding the 705
reflection aspect, readers were provided with several cues about 706
data sources, data processing and data uncertainty, although many 707
details behind design choices and mathematical models remained 708
opaque. 709
References 710
Aiello, G. (2020). Inventorizing, situating, transforming: Social 711
semiotics and data visualization. In M. Engebretsen & H. 712
Kennedy (Eds.), Data visualization in society (pp. 49–62). 713
Amsterdam University Press. 714
Aguilar, M. S., & Castaneda, A. (2021). What mathematical 715
competencies does a citizen need to interpret Mexico’s 716
official information about the COVID-19 pandemic? 717
Educational Studies in Mathematics, 108 (1–2), 227–248. 718
Bernstein, B. (1981). Codes, modalities, and the process of cultural 719
reproduction: A model. Language in society, 19 (3), 327–720
363. 721
Bolstad, O. H. (2020). Secondary teachers' operationalisation of 722
mathematical literacy. European Journal of Science and 723
Mathematics Education, 8 (3), 115–135. 724
Curcio, F. R. (1987). Comprehension of mathematical 725
relationships expressed in graphs. Journal for Research in 726
Mathematics Education, 18 (5), 382–393. 727
Engebretsen, M., & Kennedy, H. (Eds.) (2020). Data visualization 728
in society. Amsterdam University Press. 729
Gal, I. (2002). Adults' statistical literacy: Meanings, components, 730
responsibilities. International Statistical Review, 70 (1), 1–731
25. 732
Geiger, V., Goos, M., & Forgasz, H. (2015). A rich interpretation 733
of numeracy for the 21st century: A survey of the state of the 734
field. ZDM Mathematics Education, 47 (4), 531–548. 735
Halliday, M. A. K. (1978). Language as social semiotic. Edward 736
Arnold. 737
Hasan, R. (1996). Literacy, everyday talk and society. In R. Hasan 738
& G. Williams (Eds.), Literacy in society (pp. 377–424). 739
Longman. 740
Kennedy, H., & Hill, L. R. (2018). The feeling of numbers: 741
Emotions in everyday engagements with data and their 742
visualization. Sociology, 52 (4), 830–848. 743
Kirk, A. (2019). Data visualization: A handbook for data driven 744
design (2nd ed.). Sage. 745
Kress, G. (2003). Literacy in the new media age. Routledge. 746
Kwon, O. N., Han, C., Lee, C., Kim, K., Jo, G., & Yoon, G. 747
(2021). Graphs in the COVID-19 news: A mathematics audit 748
of newspapers in Korea. Educational Studies in 749
Mathematics, 108 (1–2), 183–200. 750
Mediebedriftene (2020). Lesertall. Accessed 19 April 2021 at 751
https://www.mediebedriftene.no/tall-og-fakta/lesertall/ 752
Mulligan, J. (2015). Moving beyond basic numeracy: Data 753
modeling in the early years of schooling. ZDM Mathematics 754
Education, 47 (4), 653–663. 755
Niss, M., & Jablonka, E. (2020). Mathematical 756
literacy. Encyclopedia of mathematics education (pp. 548–757
553). Springer. 758
OECD. (2016). PISA 2015 Assessment and analytical framework: 759
Science, reading, mathematics and financial literacy. OECD 760
publishing. 761
O’Halloran, K. L. (2015). The language of learning mathematics: 762
A multimodal perspective. The Journal of Mathematical 763
Behavior, 40 (A), 63–74. 764
Olande, O. (2013). Making sense of a “misleading” graph. Nordic 765
Studies in Mathematics Education, 18 (1), 5–30. 766
Pinel, C., Prainsack, B., & McKevitt, C. (2020). Caring for data: 767
Value creation in a data-intensive research laboratory. Social 768
Studies of Science, 50 (2), 175–197. 769
Presmeg, N. C. (2006). Research on visualization in learning and 770
teaching mathematics: Emergence from psychology. In A. 771
Gutierrez & P. Boero (Eds.), Handbook of research on the 772
psychology of mathematics education (pp. 205–235). Sense. 773
Roth, W.-M. (2003). Towards an anthropology of graphing: 774
Semiotic and activity-theoretic perspectives. Springer. 775
Sunde, P. B., Petersson, J., Nosrati, M., Rosenqvist, E., & 776
Andrews, P. (2021). Estimation in the mathematics curricula 777
of Denmark, Norway and Sweden: Inadequate 778
conceptualisations of an essential competence. Scandinavian 779
Journal of Educational Research, advance online 780
publication. 781
Tvedt, E. E. (2020). «Svaret er 128 000, for verdien før er 64 000, 782
og verdien doblar seg kvar gong x-aksen er forlenga»: Ein 783
designstudie om eit tverrfagleg oppgåvesett i matematikk og 784
samfunnskunnskap som støttar elevar si tolking av 785
boblediagrammet til Gapminder (Master’s thesis, University 786
of Agder). 787
Tønnessen, E. S. (2020). What is visual-numeric literacy, and how 788
does it work? In M. Engebretsen & H. Kennedy (Eds.), Data 789
visualization in society (pp. 189–205). Amsterdam 790
University Press. 791
Utdanningsdirektoratet (2019). Læreplan i matematikk 1.-10- trinn. 792
Accessed 03 February 2022 at 793
https://www.udir.no/lk20/mat01-05 794
Vos, P., & Frejd, P. (2020). The object-tool duality in 795
mathematical modelling: A framework to analyze students' 796
appropriation of diagrams to model dynamic processes. 797
Avances de Investigación en Educación Mathemática, 17, 798
52–66. 799
Vygotsky, L. S. (1986). Thought and language. MIT Press. 800
Åberg‐Bengtsson, L., & Ottosson, T. (2006). What lies behind 801
graphicacy? Relating students' results on a test of graphically 802
represented quantitative information to formal academic 803
achievement. Journal of Research in Science Teaching, 43 804
(1), 43–62. 805
Paper IV
Wiik, A., & Vos, P. (2024). Making sense of journalistic COVID-19 data
visualizations: An in-depth study of two young adults’ visual-numeric literacy. Adults
Learning Mathematics – An International Journal, 18(1), 7–28.
Wiik, A. & Vos, P. (2024). Making sense of journalistic COVID-19 data visualizations: An in-depth
study of two adults visual-numeric literacy. Adults Learning Mathematics: An International Journal,
online first.
Adults Learning Mathematics – An International Journal
Making sense of journalistic COVID-19 data visualizations:
An in-depth study of two adults’ visual-numeric literacy
Anders Wiik
University of Agder, Kristiansand, Norway
<anders.wiik@uia.no>
Pauline Vos
Western Norway University of Applied Sciences, Bergen, Norway
<pauline.vos@hvl.no>
Abstract
In 2020, the COVID-19 pandemic urged authorities to share quantitative information such as
infection and death rates. One way of disseminating was through graphs, maps, and diagrams.
Such data visualizations communicate numeric data in compact ways, but also require a
particular mathematical literacy from readers. We conceptualized this particular mathematical
literacy as visual-numeric literacy. To study it, we interviewed two young adults with higher
education but low confidence in mathematics and asked them to make sense of COVID-19 data
visualizations from journalistic digital media. An in-depth analysis of their visual-numeric
literacy revealed that the two participants had developed various sense-making strategies. Their
lived experience in the pandemic assisted them to overcome obstacles in mathematical sense-
making, and gain insights from the data visualizations. We discuss out-of-school mathematics
learning and provide recommendations for improving adults’ visual-numeric literacy.
Keywords: data visualizations, visual-numeric literacy, out-of-school learning (of mathematics),
sensemaking, COVID-19 pandemic, mathematical literacy, numeracy.
Introduction and literature review
The COVID-19 pandemic affected people all over the world. Governments and journalistic
media tried to inform citizens of risks and the spread of the pandemic, among others through a
wide range of visual representations of public health statistics. These were based on quantitative
data and mathematical models. In this paper, we describe a study, carried out during the
COVID-19 pandemic, on how adults read these data visualizations (DVs).
DVs such as graphs, diagrams and maps have become important in public discourses
(Engebretsen & Kennedy, 2020), as illustrated by the many COVID-19 DVs. DVs are visual
representations of quantitative data. They are compact and useful to communicate messages, to
convince readers, to explore data, to see patterns that other formats render invisible, and to
explain connections between phenomena (Kirk, 2019; Li & Molder, 2021). Despite the
usefulness of DVs, there are also pitfalls. DVs can be misleading through irregular scaling and
incorrect proportions (Kwon et al., 2021) and some design conventions, such as logarithmic
scales, are less readily understood (Romano et al., 2020). Reading DVs is a complex skill that is
influenced by the design of the DVs, the readers’ knowledge of DVs, as well as the readers’
knowledge and expectations about the data (Shah & Hoeffner, 2002). Further, researchers have
identified several aspects of DV reading, including extracting information, finding relationships
Wiik, A. & Vos, P. Making sense of journalistic COVID-19 data visualizations.
Adults Learning Mathematics – An International Journal
in the data, and making judgements or extrapolations that go beyond the visualized data (Friel et
al., 2001). In the case of the COVID-19 pandemic, the mathematical literacy
1
demanded by the
media coverage was a complex blend of mathematical, statistical, linguistic, demographic, and
critical skills (Aguilar & Castaneda, 2021; Gal & Geiger, 2022). Due to the high stakes
involved, Da Silva et al. (2021) suggested that COVID-19 DVs can be used for critical
discussions of the relationships between DVs, policy and inequality.
The capability to read DVs was measured in several large-scale international surveys.
According to two of them, IALS and PIAAC, all OECD countries have a significant number of
people who perform at a problematically low level of reading documents that contain DVs,
tables, diagrams and so forth. The IALS study indicated that the scores of adults in Norway,
where the present study was conducted, were generally above the OECD average. However, an
estimated 30% can only deal with simple, clearly laid out materials and will struggle with
demands of everyday life and in work in a complex, advanced society (OECD, 2000). The more
recent PIAAC survey indicated that an estimated 40% of Norwegian adults may have struggled
with journalistic COVID-19 DVs (OECD, 2019), which is the theme of this paper.
According to Evans (2018), the advanced state of modern statistics has created a
boundary between experts and the general public. When statistics are used in public policy
debates, the gap between experts and the general public constitutes an “overt crisis of statistics”
(p. 38) because the interpretation of statistical data is shielded from criticism by its complexity,
lack of transparency, and assumed authority. On the other hand, Gal and Geiger (2022)
investigated the use of statistics and mathematics in the media coverage of COVID-19 and
found examples of news items that invite readers to critically scrutinize statistical and
mathematical issues. This was realized in various ways, for example by pointing out issues of
data quality, dissent among experts, or results of fact checking.
Jackson et al. (2018) reviewed research on numeracy practices at work and in private
life. They found that the mathematics that people use in everyday situations is (1) contextual,
that is, interrelated with personal, social, and cultural surroundings, and (2) unstable, that is,
variable over time and circumstances. Everyday numeracy practices change in response to
changes in the surroundings such as a new job, a new tool, or a significant event such as a
pandemic. Jackson et al. (2018) caution that this connectedness between numeracy practices and
their situatedness creates a challenge for survey research, such as PIAAC and IALS. To meet
high statistical standards, these studies transform literacy and numeracy practices into
competence categories, levels, and scores, and they zoom out across personal, social, and
cultural surroundings. Jackson et al. (2018) recommend large-scale studies to be complemented
by qualitative research on practices within different personal, social, and cultural contexts. With
the present study, we want to contribute to the latter.
Researchers have provided advice on guiding learning processes towards developing
effective and efficient DV reading skills (e.g., Friel et al., 2001; Shah & Hoeffner, 2002; Glazer,
2011). However, these studies focus mostly on compulsory education and do not consider adults
or lifelong learning. Studies on adults’ mathematical practices in out-of-school contexts have
found that such practices are strongly connected to the social contexts in which these occur
(e.g., Curdt et al., 2022; Jackson et al., 2018) and that school-learnt mathematics is not a
guarantee of success in out-of-school mathematics (Heyd-Metzuyanim et al., 2021).
1
The capability to use mathematics in diverse contexts has many definitions and names. In this
paper, we generally use the term mathematical literacy as an umbrella term, but we also use the
term numeracy when this is the preferred term in referenced papers.
ALM International Journal, online first
Adults Learning Mathematics – An International Journal
To study mathematical literacy practices embedded in lived experiences (Curdt et al.,
2022; Jackson et al., 2018), the media coverage of the COVID-19 pandemic offered a rich
opportunity for research. There were studies using digital surveys (Heyd-Metzuyanim et al.,
2021) and studies on how imagined readers encounter demands and opportunities when reading
COVID-19 media artefacts (Aguilar & Castaneda, 2021; Da Silva et al., 2021; Gal & Geiger,
2022; Kwon et al., 2021; Stephan et al., 2021). These studies were constrained by social
distancing measures and lockdowns. For the present study, we made use of a window between
lockdowns when it was allowed to meet people. We carried out interviews with young adults
face-to-face to study their actual engagement with COVID-19 DVs from popular news media.
So, the pandemic timing directly impacted our study. Hence, our method was different from
digital surveys and video interviews. In this way, our study validates and complements findings
from the other studies mentioned.
Theoretical framework
For our in-depth analysis, we needed a framework that accounts for the many aspects of DV
reading and the role of lived experience in sense-making processes that were highlighted in the
literature review (e.g., Jackson et al., 2018). A frequently used framework for graph reading is
from Friel et al. (2001), which distinguishes between reading the data, reading between the data,
and reading beyond the data in a DV. This framework focuses entirely on the artifacts’
characteristics and does not account for the readers and their lived experience.
The framework used for the IALS study is named Document Literacy. It describes “the
knowledge and skills needed to locate and use information contained in various formats,
including job applications, payroll forms, transportation schedules, maps, tables and charts”
(OECD, 2000, p. x). This framework highlights the cognitive underpinnings of how people
locate and use information in DVs (Kirsch & Lennon, 2017), but does not consider how people
read DVs differently according to how the information in the DVs affect their lives (Jackson et
al., 2018).
We needed a framework that considers reading and making sense of DVs as a multi-
faceted skill (Gal & Geiger, 2022; Tout et al., 2021) that includes aspects such as making sense
of mathematical models (Aguilar & Castaneda, 2021; Gal & Geiger, 2022) and critically and
reflectively use quantitative information to inform actions and judgements and reflect on their
sociopolitical implications (Da Silva et al., 2021; Gal, 2002; Gal & Geiger, 2022; Geiger et al.,
2015; Stephan et al., 2021; Tout et al., 2021; Weiland, 2017). Also, the framework should
capture interrelations between mathematical literacy practices with personal, social, and cultural
surroundings.
We found that the framework visual-numeric literacy (Tønnessen, 2020) met these
needs. Visual-numeric literacy is the capability needed for getting information from texts when
reading newspapers, websites, guidelines, and so forth, that include visual representations of
quantitative data such as graphs, diagrams, and maps. This framework is inspired by research in
language learning (Hasan, 1996) and is grounded in the theoretical perspective of social
semiotics (Van Leeuwen, 2005), which casts an eye on practices in specific social and cultural
circumstances. We regard visual-numeric literacy as a specific form of mathematical literacy. It
is similar to Document Literacy but has a special focus on the readers’ social contexts. Key to
the social context is the role of lived experience. The framework is composed of three different
aspects that together cover the nuances highlighted above. Below, we explain the three aspects,
the relevant contexts, and how we used them in this study.
Wiik, A. & Vos, P. Making sense of journalistic COVID-19 data visualizations.
Adults Learning Mathematics – An International Journal
First, the recognition aspect concerns the recognition of the sign systems used for
meaning making, and for decoding the signs in order to connect them to the meaning intended
by the designer. In the case of verbal language literacy, the main sign system is the alphabet. In
the case of DVs, the sign system contains dots, lines, axes, numerals, colors and so forth.
Second, the action aspect of literacy pertains to using texts – in this case DVs – to reach
personal goals. It can entail using DVs to inform actions and decisions, such as finding a good
time for visiting elderly relatives during the pandemic. Finally, the reflection aspect pertains to
exploring and challenging the boundaries of DV practices and conventions through reflection,
critique, analysis and enquiry. Reflection literacy drives development.
Hasan (1996) distinguished between two relevant contexts when people engage with
texts, and we have applied this distinction to our analysis of interaction with DVs. First, a DV
can be understood within the context of its sign system, that is, the repertoire of available signs
and codes. It is this kind of meaning making that yields statements like “infection rates are
higher in Alta than in Kristiansand” when interpreting Figure 3 because this statement can be
inferred from the DV alone. Second, the social context is also a rich source for meaning making.
The social context entails the readers’ lived experience, history, aspirations, and environment –
family, friends, identity and so on. When the readers’ social context is used for meaning
making, statements will go beyond information inferred from the DV, for example “I need to be
extra careful around my elderly parents now that the infection rates are so high”. By combining
the three aspects with the two contexts, we created a framework consisting of six labels (Table
1) that we used for our analysis. The research question for this paper is:
What characterizes adults’ visual-numeric literacy when reading and making sense of
journalistic COVID-19-related DVs?
Table 1. Aspects and labels used for analysis of visual-numeric literacy.
Aspect
Label
Example of utterance
Recognition
literacy
Context of sign systems: Decoding without
relating it to the experienced world
Context of the readers’ social situation: Decoding
while relating it to their experienced world
When I look at the numbers, I see that the
increase is even more here.
Here in the graph, the infections are going up. I
remember that happened.
Action
literacy
Context of sign systems: Action within the DV
Context of the readers’ social situation: Action
beyond the DV
I move the mouse over the graph.
I cancelled a trip to my parents after I saw these
types of graphs.
Reflection
literacy
Context of sign systems: Reflection on the DV
Context of the readers’ social situation: Reflection
on the socio-political impact of the DV
The logarithmic scale on the vertical axis makes
the graph misleading.
DVs do not show how people experience the
pandemic.
Methods
Research context, design and participants
We held interviews in the spring of 2021, when infection rates in Norway were high but
relatively low in the municipality of the study (FHI, 2021). For the interviews, we used VG
(Verdens Gang), the most used online newspaper in Norway (Mediebedriftene, 2021). When the
first infections occurred in Norway, VG developed a web page showing health data related to
the COVID-19 pandemic (www.vg.no/special/corona). The web page contained little text and a
wide range of DVs of infections, mortality, vaccinations and so forth for Norway and the
World. The DVs were displayed in different formats, such as line graphs (see Figures 1 and 5),
and choropleth maps (see Figure 3). There were interactive features, such as mouseovers to
show exact data, like the infection rate for a specific date or a specific city (see Figures 2 and 5).
ALM International Journal, online first
Adults Learning Mathematics – An International Journal
Also, readers could change certain details in some of the DVs, for example between daily and
cumulative numbers or between linear and logarithmic scales. Below each DV, there were
clickable help-buttons that offered text boxes with legends and explanations of key concepts.
The webpage was frequently updated with new data. The webpage contained examples of what
Gal and Geiger (2022, p. 19) have called embedded criticality, that is, cues directing the
readers’ attention towards “statistical and mathematical issues that require critical scrutiny”.
Examples of this include a log of changes and errors, and mention of missing or incomplete
data. This web page attracted much traffic. Therefore, it was an important part of the media
coverage of the pandemic in Norway and an authentic arena for investigating adults’ visual-
numeric literacy.
For the data to be rich, we needed participants who were able verbalizers of their
thoughts. Also, we wanted them to be in the lower ranges of the PIAAC numeracy levels
because that would reveal more of the struggles to make sense of DVs, and it would make them
more representative of an ’average’ reader. Therefore, we approached people on the campus of a
Norwegian university asking them whether they were a student, but not in a discipline with high
mathematical entrance requirements (e.g., natural sciences, engineering, mathematics,
economics). The interviews lasted roughly one hour, and the participants received a 200 NOK
gift card. The interviews consisted of two parts: The first part was an unstructured interview
where the participants were asked to explore at their own pace and comment as they read.
Thereafter, each participant was asked open-ended questions about their reading. We chose to
start with an unstructured interview to get an impression of how they would engage with the
page with minimal interference from the researchers (Bryman, 2016).
Because the unstructured first part of the interview did not ensure comparability and
coverage of all the aspects of the framework, the second part of the interview followed a semi-
structured design (Bryman, 2016). In this second part, the interviewer asked questions that
targeted each aspect of their visual-numeric literacy, as well as relevant background questions.
The participants were asked the same main questions, but follow-up questions varied depending
on their answers. Among the questions asked in this part of the interview were questions about
the DV shown in Figure 2. Here, the interviewer asked them to elaborate on the information
they get from the DV, followed up by probing questions that targets the logarithmic scale such
as “when is the increase in the number of infected the greatest?” and “what happens if you
switch to a linear display?” Towards the end of the interview, the participants were asked
questions about their experience with school mathematics.
The two participants, hereafter anonymized as Abe and Bea, had a similar background
in that they both reported a negative experience with school mathematics. However, their
experience with COVID-19 DVs was very different: Abe reported that he avoided numerical
content in general and had never seen a DV related to COVID-19. By contrast, Bea had a strong
interest in COVID-19 DVs and had visited the web page used for the interviews many times
already.
Wiik, A. & Vos, P. Making sense of journalistic COVID-19 data visualizations.
Adults Learning Mathematics – An International Journal
Figure 1. Miniature line graphs of infection trends per municipality,
grouped according to rising, flat or sinking trends. Downloaded from vg.no 07.04.2021.
Figure 2. Histogram of cumulative deaths in Norway with logarithmic vertical scale and
mouseover showing 251 deaths on 4 July 2020. Downloaded from vg.no 07.04.2021.
ALM International Journal, online first
Adults Learning Mathematics – An International Journal
Figure 3. Choropleth map showing infection rates in Norwegian municipalities.
Downloaded from vg.no 07.04.2021.
Wiik, A. & Vos, P. Making sense of journalistic COVID-19 data visualizations.
Adults Learning Mathematics – An International Journal
Figure 4. Barcode plot ranking municipalities by the date of the highest number of infected.
Downloaded from vg.no 07.04.2021.
ALM International Journal, online first
Adults Learning Mathematics – An International Journal
Figure 5. Combined histogram and line graph showing daily and 7-day average deaths in Norway
with mouseover showing 21 deaths for 13 January 2021. Downloaded from vg.no 07.04.2021.
Wiik, A. & Vos, P. Making sense of journalistic COVID-19 data visualizations.
Adults Learning Mathematics – An International Journal
Data and analysis
After transcribing and anonymizing the data, the analysis consisted of three rounds. First, the
transcripts were read thoroughly to get an overview. Second, each statement was labeled as
explained below. Third, for both participants, all statements were rearranged according to the
labels, which enabled us to characterize the three aspects of each participants’ visual-numeric
literacy.
The analysis was based on six labels (Table 1). For each of the three aspects of literacy,
recognition, action, and reflection, we distinguished between two contexts. When the reading
was turned towards the DVs themselves, we labeled them ‘sign system context’. When the
reading was explicitly turned towards the DVs’ role or impact on the social context, we labeled
it as ‘social context’. The label ‘recognition literacy in the sign system context’ was used when
a participant directly decoded a DV (e.g., decoding values from graphs). The label ‘recognition
literacy in the social context’ was used when decoding was supported by social context. And so
forth for action literacy and reflection literacy.
After initial analysis, both participants were sent the transcripts and a summary of the
analysis. Both agreed that they were accurately represented.
Results
In this section, we first present key findings for Abe and Bea separately. A summary of the
characteristics for each participant is provided at the end of this paragraph. This will then be
synthesized across the participants in the Discussion paragraph.
Abe
Abe was a 22-year-old male university student of Music and English. In upper secondary
school, he specialized in Music and took the lowest course in Mathematics (1P). He disliked
mathematics and statistics but considered them useful. When asked about experiences in the
COVID-19 pandemic, he highlighted feelings of loneliness, but he had a circle of close friends
to socialize with. His sources of pandemic information were popular social media, Google, and
an Instagram account from the local municipality that posted about mitigation measures. He
claimed that he never saw graphs about the pandemic before the interview.
Abe’s recognition literacy
To analyze Abe’s recognition literacy, we looked for instances where he decoded DVs, and we
were interested in which relevant contexts he used (the sign system or the social context). For
instance, when Abe looked at a DV where the Norwegian municipalities were grouped
according to their trend in infections (rising, flat, sinking) and represented by small line graphs
(Figure 1), he commented:
96 Abe: Flat trend, Trondheim has been good recently. Sinking trend, Kristiansand. … [pointing
at the peak in the curve for Kristiansand] That’s when everything was closing. Otherwise, it has
been sinking in Kristiansand. What is amusing is that I work out at the gym almost every day,
while everything is shut down. … They started shutting down everywhere except Kristiansand.
His comment on Trondheim was probably based on recent news about that
municipality. Next, he looked at the headlines, and found his university town Kristiansand in the
cluster under the headline “sinking trend”. He pointed with the cursor to a peak in infection
rates in February 2021, about six weeks before the interview. Here, he pointed at the tension
that whilst society was closing down at this point, Kristiansand remained open, which he
concretized by his habit of going to the gym every day. This instance illustrates several
ALM International Journal, online first
Adults Learning Mathematics – An International Journal
characteristics of Abe’s reading and sense-making. First, he often relied on cues from the textual
part of the DVs such as headlines, and not the numerical part, to make sense of the DVs.
Second, he used his pandemic knowledge and experiences from his social context to make sense
of the DVs and validate his interpretations.
When he discussed a DV that incorporated a histogram, showing daily COVID-19
mortality, and a line graph showing seven-day averages of the same data (Figure 5), he was
asked to explain the difference between the bars and the line. This question was aimed at
probing his recognition literacy in the context of the sign system. He said:
302 Abe: I think the dark red line is general.
303 Int: Yes, what does that mean?
304 Abe: I’m thinking … when I say general, I’m thinking of the number of infections we get in
that time, you know, rather than exact numbers we have this … you know … as long as we stay
below zero cases or percent, then infection rates go down.
Although he struggled to verbalize what he meant by ‘general’, this dialogue showed
that he understood that the red line shows something that differs from the ‘exact’ (absolute)
numbers. His statement about negative numbers points at a hurdle for Abe in expressing
decreasing infections in mathematical terms. After studying the DV further, he concluded that
the line is an average of the bars, and as he said this, he also realized that there was a legend
explaining that the curve is a seven-day average of daily deaths, and that the bars show daily
deaths. This shows that he had an intuitive and informal understanding of graphs and, through
discussion, enquiry and finding legends, was able to decode a number of signs in the DV, in
particular the difference between the bars and line graph.
When asked to explain the concept ‘per 100 000’ used in the choropleth map (Figure 3),
Abe used the terms “percentwise”, “large perspective” and “average perspective”, showing
untechnical and informal understanding of relative numbers in the DV. When discussing
further, Abe expressed that infection rates per 100 000 in municipalities with fewer than 100
000 inhabitants were “nonsensical and inaccurate”. He stated that rates per 100 000 could not be
applied for cities like Bodø with fewer than 100 000 inhabitants (lines 361, 363). This showed a
deficient capability to decode relative numbers. Nevertheless, he was able to accurately make
sense of the map by decoding the colors, where he could read darker shades of red showing
worse intensities of the pandemic. In this way, the map allowed him to avoid numbers. Thus, his
recognition literacy was functional and non-formal. With each DV he used a narrative of
experiences from his own social context to support his decoding.
When Abe was challenged to make sense of Figure 2, which shows deaths in a diagram
with a logarithmic vertical scale, his initial impression was that “a lot of people have died” (line
200). However, after studying the concrete numbers from the mouseovers, he realized that the
graph showed cumulative numbers. Initially, he claimed that the greatest increase was when the
curve was steepest, even after studying the concrete numbers through mouseovers. It was only
when he and the interviewer toggled between the logarithmic and linear scale that he realized
that there was an even larger increase in the winter 2020/21. This shows that he had difficulties
with the formal numerical codes.
Abe claimed several times that the graphs repeated themselves. This was elaborated:
290 Abe: It [Figure 2 and 5] is even more repetition because it shows that it was like I thought
that there was a lot of infections in the [Christmas] holidays.
Hence, Abe interpreted the DVs holistically as telling a story of how the pandemic
developed in waves and disregarded the concrete numeric information underlying each DV.
Wiik, A. & Vos, P. Making sense of journalistic COVID-19 data visualizations.
Adults Learning Mathematics – An International Journal
Abe’s action literacy
First, we focus on Abe’s action literacy in the context of the sign system, that is, his actions with
the DVs only. While exploring the page, he scrolled up and down a lot and would often
continue talking about a DV while scrolling further and making mouseovers. Unless prompted,
he never opened expandable information boxes or switched toggles. His scrolling and use of
mouseovers supported him in finding relevant information, although he did not utilize the full
potential in information boxes and toggles.
When looking at Abe’s action literacy in the social context, we analyze how he used
DVs to inform actions and decisions, and how he saw the DVs as reflecting events happening
around him. Talking about the page in general, he said:
176 Abe: It [the content of the page] shows … where it is worst … at the same time it shows what
we have to do, we must be more careful when we are in large cities. At the same time, the cities
must think clearly and not lose the grip only because the virus is spreading in one place.
Thus, to Abe the contents of the page called for certain forms of action by authorities.
He also said that:
462 Abe: [I must be careful] on behalf of myself, others, and so forth. And if I go and visit my
family, then I must take them into account.
Thus, by finding relevant information, Abe felt ready to act and be extra careful about
visiting areas with high infection rates on behalf of himself and others.
When asked why a peak in a graph for infections occurred roughly two weeks before a
peak in a mortality graph, he reasoned:
385 Abe: They may have been fighting for their lives and died after some time.
So, when prompted, he was able to reason causally across the DVs, and for this he used
his pandemic knowledge. This reasoning supported by the social context was a central feature in
both his recognition and action literacy.
Abe’s reflection literacy
First, we present findings about how Abe reflected on the DVs themselves, that is, reflection
literacy in the context of the sign system. While discussing a DV showing infection peaks in
Norwegian municipalities (Figure 4), Abe initiated a discussion on the limitations of using
relative numbers. In this DV, the municipalities were ranked by the most recent date of the
highest infection rate. Thus, small and large municipalities were mixed. The DV used a color
code where the hue varied according to the infection rate, and the total of infected was given in
numerals. Kvam, a rural municipality, was ranked second because they just experienced a large
outbreak. Oslo, the capital and largest city in Norway, was ranked much lower despite having a
higher absolute number of infected.
102 Abe: These are very small numbers when you look closer.
103 Int: Are Oslo’s numbers also small? You just found Oslo.
104 Abe: They [Oslo] have 29 000 [infected]. A lot more than the 95 [infected] that are in Kvam.
That’s a little unclear. But still …
105 Int: Yes. What do you mean by unclear?
106 Abe: I think that you can perfectly well look in terms of percentages that, all right, now there
are 95 percent infected, oh my! But then you see that only two out of three are infected. That’s a
little unclear in my opinion.
In this instance, Abe pointed at the complexity of him reading this DV. He indicated
that the ordering contradicted the ordering he knew. Then he pointed out that it was the use of
relative numbers that created this tension. The change of 95 infections to 95 percent and his
ALM International Journal, online first
Adults Learning Mathematics – An International Journal
ratio “two out of three” in line 106 were constructed for his argument. His perception of
unclarity seemed to stem from his mixing of relative and absolute numbers.
He also reflected on the opportunities offered by DVs:
258: Abe: Just now, it was new to me that it [the latest wave of infections] started in November
and not in February.
He considered the DVs as a helpful and trustworthy addition to other means of reporting
about the pandemic because they illuminate patterns.
Now we turn to findings about Abe’s reflection literacy in the social context. Abe gave
his opinion on news media:
268 Abe: [the degree to which the media exaggerate] varies of course, but that’s why I like to
look at multiple perspectives.
269 Int: And that means through social media?
270 Abe: Yes.
Although he thought that news media tend to exaggerate and used social media as his
main source of information instead, he still considered the page to be trustworthy because of the
richness of perspectives it provided and because he considered the Norwegian Institute of Public
Health (FHI, the data source) reliable. He commented that the page offered him new
information. Despite spending a considerable amount of time exploring the page, he did not
attend to the instances of embedded criticality.
Summary of Abe’s visual-numeric literacy
Abe relied to a large extent on his knowledge and experience of the pandemic to make sense of
the DVs.
• Recognition literacy: He decoded DVs functionally, non-formally and in a way that was
dependent on concrete interpretations of the social context of the data. He had a limited
understanding of mathematical concepts such as relative numbers. Unless prompted, he did
not fine-read DVs and often interpreted them based on general impressions and headers. He
decoded the particular meanings only when there were absolute numbers given, which he
used to create narratives (e.g., telling personal memories from his gym, or visiting his
family). While avoiding quantities, he interpreted the DVs qualitatively, which led him to
interpret the DVs as telling the same story even though they show different data sets.
• Action literacy: He avoided DVs in his everyday media habits and preferred non-numeric
content. When looking at DVs in the interview, he actively scrolled and looked for
mouseovers, which helped him find relevant information. When he encountered
contradictory interpretations of the DVs, he did not look for resolutions. When prompted, he
could see causal relations in the data. He saw how actions and events such as Christmas
holidays left an imprint in the data, and how data can inform action.
• Reflection literacy: He was critical of some design choices when the DVs did not reflect his
impression of reality. Nevertheless, he considered the page as a whole as trustworthy, and it
could offer him new insights into the pandemic.
• Cross-aspect: We observed that in the context of the sign system, Abe’s action literacy
(mouseovers, toggles) supported his recognition literacy. For each of the aspects of literacy
(recognition, action and reflection), his lived experience of the pandemic helped him give
meaning to the sign system.
Bea
Bea was a 23-year-old female university student of Gender Studies, with a bachelors’ degree in
Global Development. In upper secondary school, she chose the theoretical stream, specialized in
Wiik, A. & Vos, P. Making sense of journalistic COVID-19 data visualizations.
Adults Learning Mathematics – An International Journal
Social Sciences and took the lowest course in Mathematics (1P). She had negative experiences
with school mathematics, she said she knew the basic concepts such as plus or minus, and
beyond these she said to have low confidence. When asked about her pandemic experiences, she
answered that it had caused some strain on her studies, work and social life, but she considered
herself relatively unaffected by the pandemic. She regularly consulted a wide array of sources
about the pandemic, and she was a regular reader of VGs COVID-19 webpage. In the beginning
of the pandemic, she visited the page several times a day. At the time of the interview, she
visited it 2-3 times a week. She regularly encountered statistics and DVs in her studies and in
social media.
Bea’s recognition literacy
As with Abe, we looked for instances where Bea decoded DVs, and were interested in which
relevant contexts she used (the sign system or the social context). When she first explored the
page, she did so very quickly because she was already familiar with the content and did not need
much time to get an overview. This shows that her familiarity with the page enabled her to
decode many of the DVs quickly and effortlessly. When she was challenged to make sense of
the DV in Figure 2 which shows cumulative numbers of deaths in a logarithmic vertical scale,
she started by looking up the headline. Her first impression was that “it looks like a lot”, adding
that the red area under the graph increased the overwhelmingness. However, she proceeded to
decode data values through mouseovers showing the concrete numbers, which changed her
impression:
58 Bea: … it is not quite so overwhelming after all when you look at the numbers, I think.
Here, she used her action literacy in the sign system to support her recognition literacy
in the sign system by looking up numbers in mouseovers. To further probe her decoding of the
logarithmic scale, she was asked to identify when the number of deaths grew fastest. First, she
answered March and April 2020 when the curve was steepest. When she then was asked to
compare the increase in this period with the increase in the most recent period, she concluded:
68 Bea: … when I look at the numbers, I see that it [the number of deaths] is even more here [in
winter/spring of 2021] even though the curve is flatter.
This contradiction between the visual shape and numerical information prompted a
search for a resolution. After considering time intervals, she looked at the vertical axis:
74 Bea: It goes straight from 10 to 100 here and from 100 to 1000. And 1 to 10. So, it makes
sense that it is steep here [in March and April 2020]. But, if you don’t look at the numbers, only
the shape, you may not get that.
By pointing at the characteristic of the logarithmic scale, she inferred that this design
feature had misled her initial decoding of the shape of the curve. Hence, aided by a prompt from
the researcher and quality control of her own decoding, she developed an understanding of
diagrams with logarithmic scales. This shows that she was capable of learning to decode new
signs related to DVs through reflection and enquiry into her own decoding. Hence, she used her
reflection literacy to expand her recognition literacy. Her decoding of Figure 2 relied
consistently on signs in or surrounding the DV, and there were no explicit references to the
social context during this sense-making process.
Later, Bea discussed the concepts ‘per cent’ and ‘per 100 000’, used in some DVs.
Elaborating on the meaning of ‘per 100 000’ in a world map showing the countries, which were
colored according to infection rates per 100 000 inhabitants, she explained:
116 Bea: I understand it as something used to make it easier to understand percentwise how many
infected there are. So, it’s not right to say that if we are 700 infected then … it can be that they
have equally many infected in a country percentwise, but twice the number of inhabitants.
ALM International Journal, online first
Adults Learning Mathematics – An International Journal
Her mention of “percentwise” suggests that her understanding of ‘per 100 000’ was
modeled on an understanding of percent and useful for comparing across, for example, countries
of different size.
117 Int: Is it possible to talk about number of infected per 100 000 inhabitants in a country with
less than 100 000 inhabitants?
118 Bea: Yes, they just need to calculate it.
119 Int: Yes. Can you elaborate on that? Take for example a fictitious country with 10 000
inhabitants.
120 Bea: Yeah, then they need to calculate from 10 000. They have to, kind of, multiply their
number of infected by 10.
And somewhat later, she discussed differences between percent and ‘per 100 000’:
146 Bea: The difference is that 10 percent is a lot more [than 10 per 100 000].
With this, she showed to have understood the concepts of ‘per 100 000’ and percent,
their mutual relation, and their usefulness for comparisons. She uses an example from the social
context, Norway versus the US, to elaborate her understanding of ‘per 100 000’, but there was
no indication that she depends on the social context to decode these numbers. Thus, while she
was able to connect the signs to the social context, she did not depend on the social context to
decode the DVs.
Bea’s action literacy
First, we focus on Bea’s action literacy in the context of the sign system. While exploring the
DVs, she often used mouseovers, which she described like this:
98 Bea: [To find the number of infected] I go in here and it says registered infected. Then I take
the cursor over and see how many there are.
102 Bea: … for positive tests and such, which I don’t really look at, it’s the same thing. You just
take the cursor over and kind of look at the date and number of infected, and the same for the other
categories.
Using mouseovers to extract numbers was a crucial component of her reading habits.
The quote from line 102 also suggests that she habitually checked infection rates and skipped
content that she deemed less relevant. This was elaborated later:
106 Bea: The number of people on respirator isn’t crucial to me. It is of course sad that it
happens, but it is not something that I … yeah.
236 Bea: [When I don’t immediately understand something] I move on rather than spend time
trying to understand it.
Thus, from her extensive experience with the page, she selected information based on
pertinence and how readily she understood the content. When asked to explain the difference
between the pale pink bars and the red line in Figure 5, she demonstrated how she found
explanations:
284 Bea: The thick red line is, it’s written here and when I hold the cursor over, it explains what
it is. It is the average over seven days. So, it’s the average. The pale pink [bars] are the number of
new cases per day. So even though some of them [bars] are very high, they don’t reflect the
averages.
Thus, she had developed effective reading habits that involved interactive features and
content filtering that enabled her to efficiently find relevant content and make sense of the DVs.
In other words, her action literacy supported her recognition literacy.
Now we turn to Bea’s action literacy in the social context. Related to this, she said:
18 Bea: … It has been a lot about [COVID-19] in everyday life, casual conversations and when
I’m at work and with friends. It has been a lot of talk about it. Now, the numbers are falling.
Wiik, A. & Vos, P. Making sense of journalistic COVID-19 data visualizations.
Adults Learning Mathematics – An International Journal
In this quote, Bea connected her knowledge of statistics on infection rates to her personal life
and mitigation policies. She indicated that the falling numbers could lead to less invasive
mitigation policies in the near future. This shows that she was capable of connecting the
information from the DVs to her social context.
Bea’s reflection literacy
First, we present findings about Bea’s reflection literacy in the context of the sign system. After
the exploration, Bea shared her thoughts on the limitations of DVs and statistics:
40 Bea: It shows a lot of statistics. But it is … a lot of numbers, and numbers don’t always
represent reality the way we experience it.
While looking at Figure 2, she reflected on how DVs can be misleading:
58 Bea: It looks like a lot. First, it’s entirely red and you don’t usually associate red with
something good. […] The first impression is like ‘wow it looks like a lot’. But then you look
closer and it’s not so overwhelming when you look at the numbers. I think.
74 Bea: […] If you look at the shape and not the numbers, you may not understand it.
Here, she reflected on impressions based on salient visual features such as the color red
and the steepness of graphs. She described the logarithmic scale as misleading and not as
wrong. This suggests that she understood that the logarithmic scale did not change the data
themselves, only their visual representation. Hence, she reflected on how challenging design
choices such as logarithmic scales can mislead readers.
Next, we present findings relating to her reflection literacy in the social context.
Discussing a bar chart of registered infected in Norwegian municipalities, she said:
100 Bea: For today it [the number of registered infected] says zero but it’s certainly not updated
because for yesterday it says 421.
Here, she concluded that an anomalous value was due to lags in the updating of data.
Thus, she showed how she could critically analyze a DV and use her understanding of the
processes behind the data collection to understand anomalies.
She generally trusted the page to be accurate but would often compare it with other
sources. She did not attend to the instances of embedded criticality on the page, but her use of
multiple sources to validate the data suggests that she is aware of the possibility of missing,
inaccurate or incomplete data. She suggested that the use of DVs can make webpages appear
more scientific, but that DVs can be confusing for people who do not understand them. Further,
42 Bea: Thinking about the next few days and the coming week, when we just had a large
breakout here earlier this week and they are opening up this weekend. I don’t think it adds up
when the numbers are high but we are opening up instead of staying locked down.
This quote shows how she sees connections between statistics and events and how she can
critically reflect on the relationship between policies and statistics. Thus, her judgment of
appropriate actions, which pertains to her action literacy in the social context, triggered her
reflection literacy.
Summary of Bea’s visual-numeric literacy
Bea was able to decode DVs without explicit reference to her knowledge and experience of the
pandemic.
• Recognition literacy: She decoded DVs functionally, effectively and formally. For relative
numbers, she had an accurate understanding that involved definitions, calculation
procedures as well as meta-knowledge of their usefulness. She showed a flexible and open
attitude towards logarithmic scales, which she learned to decode during the interview. She
ALM International Journal, online first
Adults Learning Mathematics – An International Journal
was capable of decoding DVs within the context of the sign system with minimal reference
to the social context. She was able to use the social context to explain the usefulness of
relative numbers.
• Action literacy: Before the interview, she already had developed an extensive familiarity
with the page, and reading habits that helped her select and locate desired content such as
legends and explanations. This helped her to make sense of DVs quickly and effortlessly.
She used mouseovers, scrolling and visual searches for legends and explanatory text to
explore DVs analytically and systematically. She could see how actions and events such as
vaccinations and holidays could leave an imprint in the data, and how data can be used to
inform actions and decisions.
• Reflection literacy: She was critical of some design choices. Her criticism was based on the
impressions that the DVs gave her, potentials for misunderstanding, and the validity and
accuracy of the data. She was aware of limitations of DVs. Through critical analysis, she
could perform quality control of her own reading and when encountering contradictions, she
was able to resolve these.
• Cross-aspect: We observed that her reflection literacy supported her recognition and action
literacy; that her action literacy triggered her reflection literacy; and that she could fluently
translate between the sign system context and her social context literacy.
Discussion and conclusion
The research question for this paper was “what characterizes adults’ visual-numeric literacy
when reading and making sense of COVID-19 data visualizations?”. Visual-numeric literacy is
the form of mathematical literacy that is specific to getting information from/making sense of
DVs, and our answer is framed by the framework of Hasan (1996) and Tønnessen (2020) that
distinguishes between recognition, action, and reflection literacy, and between the contexts of
the sign system and the readers’ social context (Table 1).
A first characteristic of visual-numeric literacy was that the three aspects (recognition,
action and reflection literacy) and the two contexts (the context of the sign system and the social
context) all interacted and were mutually reinforcing. Hasan (1996) and Tønnessen (2020)
suggested that action literacy builds on recognition literacy and that reflection literacy builds on
action literacy, and not the other way around. Our study contradicts this hierarchy. For example,
action literacy (e.g., using interactive features to look up legends) supported recognition
literacy, and knowledge of the depicted social context through lived experience supported
recognition of signs. These interactions between different aspects of literacy are not captured by
frameworks that are useful for large-scale surveys such as IALS and PIAAC (OECD, 2000,
2019). There, standardized tasks are administered to participants across countries and cultures
and are therefore unlikely to capture themes that are deeply connected with the participants’
social context. For example, IALS applied a task about firework related injuries (Kirsch, 2001),
which could be comprehensible to many participants. However, it is unlikely that such a social
context is really significantly impacting the participants’ social lives, in the same way as the
COVID-19 pandemic impacted the participants in our study. Therefore, our qualitative study
could capture the role of the social context in making sense of DVs. This demonstrates how
qualitative research can complement quantitative research such as IALS and PIAAC (OECD,
2000, 2019).
What also characterized visual-numeric literacy was that it interacts with other forms of
literacy. For example, when the participants scrolled and used toggles, they used their digital
media literacy. The interaction between visual-numeric literacy and digital literacy was also
Wiik, A. & Vos, P. Making sense of journalistic COVID-19 data visualizations.
Adults Learning Mathematics – An International Journal
observed by Tønnessen (2020). We also observed that visual-numeric literacy interacted with a
more general mathematical literacy when making sense of relative numbers and logarithmic
scales. A limited understanding of these concepts was a hurdle for the sensemaking process.
However, with some guidance, participants were able to develop this understanding. Visual-
numeric literacy also interacted with media literacy, when participants showed skills in filtering
and organizing the web page. Neither of our participants explicitly attended to the instances of
embedded criticality (Gal & Geiger, 2022), but one participant showed a critical awareness of
the data quality. The fact that they did not critique or enquire into the underlying data
infrastructures supports Evans’ (2018) concerns about the non-transparent boundary between
lay people and people with statistical expertise. The interaction between multiple forms of
literacy agrees well with Gal and Geiger’s (2022) call for blended knowledge where
mathematical and statistical knowledge are integrated with other forms of knowledge such as
language, contextual knowledge and critical demands.
A further characteristic of visual-numeric literacy relates to the diversity and design of
DVs. We observed a diverse range of DV formats including a line graph with missing vertical
axis (Figure 1), bar charts, choropleth maps and a barcode plot (Figure 4). These illustrate how
media (1) continuously introduce new formats and (2) break conventions taught in schools. This
diversity and unconventionality mean that readers need to continually and flexibly adapt and
develop their visual-numeric literacy to account for new formats and new contexts. At the same
time, we also observed that the DVs were designed with journalistic insight that supported
readers. For example, the use of colours to express danger (red) or calm (white), the inclusion of
legends and explanatory text to guide readers, and the use of headings to create structure and
coherence, were helpful cues and lowered demands to visual-numeric literacy.
In addition, we found two characteristics of how readers can approach their visual-
numeric literacy. The first can be characterized as avoiding DVs and consequently not using
opportunities to develop visual-numeric literacy. The second approach can be characterized as
being interested in social phenomena depicted by DVs, whereby a need for information leads to
frequently visiting DV-filled media pages that are regularly updated and relevant to a reader’s
life. One case in our study shows that this second approach can lead to the development of a
sophisticated visual-numeric literacy.
Of course, our study has limitations through the selection of DVs and the low number
of participants. However, the low number of participants enabled an in-depth analysis. Further,
the research setting, where a participant reads DVs under the watching eyes of a researcher, is
artificial and does not match the circumstances under which readers typically interact with
digital DVs. A further limitation in this study is the theoretical framework; it is well suited for
describing the participants’ capacity to read and make sense of the DVs as well as describing the
sense-making processes, but it does not reveal how visual-numeric literacy can be developed.
Our study sheds light on the need to think of graph reading as a practice that is deeply
connected to the readers’ social context. Therefore, we recommend mathematics educators to go
beyond teaching graph reading as a mechanical decoding of signs, and to connect signs to
events that are authentic and relevant to the students’ lives. The DVs that people encounter in
everyday life are constantly changing. Therefore, readers must constantly adapt their visual-
numeric literacy to new demands. Lifelong learning of visual-numeric literacy can start in
schools but must be reinforced by regular participation in activities that involve DVs, such as
reading the news. Also, DV designers should be aware of and attend to the challenges that
readers face when reading DVs. These challenges depend on the readership and the social
context visualized (e.g., COVID-19, climate issues). We recommend that DV designers
consider further developing appropriate reader guidance by offering explanations of key
ALM International Journal, online first
Adults Learning Mathematics – An International Journal
concepts, clear structure and guidelines for how to read DVs, as well as meta-information about
the data (sources, methods for handling, etc.).
In terms of recommendations for theory and research, this study sheds light on the need
for researchers to use theoretical frameworks with a broad scope that can capture the many
facets of reading and making sense of DVs. In particular, the role of lived experience was found
to be significant and is easily overlooked in frameworks that focus on cognitive skills. We
encourage future research to focus on why some people productively engage with everyday
mathematics while others avoid it. We also encourage future research on finding ways of
teaching mathematics in mandatory, vocational, and adult education that can promote
participation and lifelong learning in everyday mathematics activities. In this regard, we expect
that teaching approaches that involve practical and relevant activities can be beneficial, and that
emotions play a vital role.
DVs are ubiquitous in contemporary society, and are used to offer information of
democratic, social and economic relevance (Engebretsen & Kennedy, 2020). People who are
unable to make sense of DVs can therefore have problems participating in important discourses.
The high number of people who perform poorly on surveys of relevant skills are worrying
(OECD, 2000, 2019), and the present study adds to the concern by showing how one
participant, who habitually avoided DVs and other numerical content in everyday life, struggled
to make sense of many DVs. As with other mathematical practices in everyday life, DV reading
is context dependent and in constant development (Jackson et al., 2018). Therefore, we
encourage efforts to make school mathematics more socially relevant with the aim of preparing
students for lifelong participation and lifelong learning in adult life.
References
Aguilar, M. S., & Castaneda, A. (2021). What mathematical competencies does a citizen need to interpret
Mexico’s official information about the COVID-19 pandemic? Educational Studies in Mathematics,
108(1–2), 227–248. https://doi.org/10.1007/s10649-021-10082-9
Bryman, A. (2016). Social Research Methods (5th ed.). Oxford University Press.
Curdt, W., Schreiber-Barsch, S., & Angermeier, K. (2022). Numeracy practices and vulnerability under
conditions of limited financial means: “Without money, you can't survive or do anything or develop
yourself.” Adults Learning Mathematics: An International Journal, 17(1), 27–40.
Da Silva, A. S., Barbosa, M. T. S., De Souza Velasque, L., Da Silveira Barroso Alves, D., & Magalhães,
M. N. (2021). The COVID-19 epidemic in Brazil: How statistics education may contribute to unravel
the reality behind the charts. Educational Studies in Mathematics, 108(1–2), 269–289.
https://doi.org/10.1007/s10649-021-10112-6
Engebretsen, M. & Kennedy, H. (Eds.). (2020). Data Visualization in Society. Amsterdam University
Press. https://doi.org/10.2307/j.ctvzgb8c7
Evans, J. (2018). Statistics in public policy debates: Present crises and adult mathematics education.
Adults Learning Mathematics: An International Journal, 13(1), 38–45.
FHI (2021, October 08). Daily report and statistics about coronavirus and COVID-19. Accessed 22
October 2021 at https://www.fhi.no/en/id/infectious-diseases/coronavirus/daily-reports/daily-reports-
COVID19/
Friel, S. N., Curcio, F. R., & Bright, G. W. (2001). Making sense of graphs: Critical factors influencing
comprehension and instructional implications. Journal for Research in Mathematics Education,
32(2), 124–158. https://jstor.org/stable/749671
Gal, I. (2002). Adults’ statistical literacy: Meanings, components, responsibilities. International
Statistical Review, 70(1), 1–25. https://doi.org/10.1111/j.1751-5823.2002.tb00336.x
Gal, I., & Geiger, V. (2022). Welcome to the era of vague news: A study of the demands of statistical and
mathematical products in the COVID-19 pandemic media. Educational Studies in Mathematics,
111(1), 5–28. https://doi.org/10.1007/s10649-022-10151-7
Wiik, A. & Vos, P. Making sense of journalistic COVID-19 data visualizations.
Adults Learning Mathematics – An International Journal
Geiger, V., Goos, M., & Forgasz, H. (2015). A rich interpretation of numeracy for the 21st century: A
survey of the state of the field. ZDM Mathematics Education, 47(4), 531–548.
https://doi.org/10.1007/s11858-015-0708-1
Glazer, S. (2011). Challenges with graph interpretation: A review of the literature. Studies in Science
Education, 47(2), 183–210. https://doi.org/10.1080/03057267.2011.605307
Hasan, R. (1996). Literacy, everyday talk and society. In R. Hasan & G. Williams (Eds.), Literacy in
society (pp. 377–424). Longman.
Heyd-Metzuyanim, E., Sharon, A. J., & Baram-Tsabari, A. (2021). Mathematical literacy in the COVID-
19 pandemic and its relation to school mathematics education. Educational Studies in Mathematics,
108(1–2), 201–225. https://doi.org/10.1007/s10649-021-10075-8
Jackson, K., Rogers, A., & Yasukawa, K. (2018). Expanding and deepening the terrain: Numeracy as
social practice. In K. Yasukawa, A. Rogers, K. Jackson & B. V. Street (Eds.), Numeracy as social
practice: Global and local perspectives (pp. 243–354). Routledge.
Kirk, A. (2019). Data visualization: A handbook for visual design. (2nd ed.). Sage.
Kirsch, I. (2001). The international adult literacy survey (IALS): Understanding what was measured.
Educational testing service.
Kirsch, I. & Lennon, M. L. (2017). PIAAC: A new design for a new era. Large Scale Assessment in
Education, 5, article 11. https://doi.org/10.1186/s40536-017-0046-6
Kwon, O. N., Han, C., Lee, K., Kim, K., Jo, G., & Yoon, G. (2021). Graphs in the COVID-19 news: A
mathematics audit of newspapers in Korea. Educational Studies in Mathematics, 108(1–2), 183–200.
https://doi.org/10.1007/s10649-021-10029-0
Li, N. & Molder, A. L. (2021). Can scientists use simple infographics to convince? Effects of the “flatten
the curve” charts on perceptions of and behavioral intentions toward social distancing measures
during the COVID-19 pandemic. Public Understanding of Science, 30(7), 898–912.
https://doi.org/10.1177/09636625211038719
Mediebedriftene (2021). Lesertall. Accessed 22 October 2021 at https://www.mediebedriftene.no/tall-og-
fakta/lesertall/
OECD. (2000). Literacy in the information age: Final report on the international adult literacy survey.
Statistics Canada. https://www.oecd.org/education/skills-beyond-school/41529765.pdf
OECD. (2019). Skills matter: Additional results from the survey of adult skills. OECD Publishing.
https://doi.org/10.1787/1f029d8f-en
Romano, A., Sotis, C., Dominioni, G., & Guidi, S. (2020). The scale of COVID-19 graphs affects
understanding attitudes, and policy preferences. Health Economics, 29(11), 1482–1494.
https://doi.org/10.1002/hec.4143
Shah, P., & Hoeffner, J. (2002). Review of graph comprehension research: Implications for instruction.
Educational Psychology Review, 14(1), 47–69. https://doi.org/10.1023/A:1013180410169
Stephan, M., Register, J., Reinke, L., Robinson, C., Pugalenthi, P., & Pugalee, D. (2021). People use math
as a weapon: Critical mathematics consciousness in the time of COVID-19. Educational Studies in
Mathematics, 108(3), 513–532. https://doi.org/10.1007/s10649-021-10062-z
Tout, D., Demonty, I., Díez-Palomar, J., Geiger, V., Hoogland, K., & Maguire, T. (2021). PIAAC Cycle 2
assessment framework: Numeracy. In OECD (Ed.), The Assessment Frameworks for Cycle 2 of the
Programme for the International Assessment of Adult Competencies (pp. 65–154). OECD Publishing.
https://doi.org/10.1787/c4221062-en
Tønnessen, E. S. (2020). What is visual-numeric literacy, and how does it work? In M. Engebretsen & H.
Kennedy (Eds.), Data Visualization in Society (pp. 189–205). Amsterdam University Press.
https://doi.org/10.2307/j.ctvzgb8c7.18
Van Leeuwen, T. (2005). Introducing Social Semiotics. Routledge.
Weiland, T. (2017). Problematizing statistical literacy: An intersection of critical and statistical literacies.
Educational Studies in Mathematics, 96(1), 33–47. https://doi.org/10.1007/s10649-017-9764-5
