Content uploaded by Jonathan W. Y. Gray
Author content
All content in this area was uploaded by Jonathan W. Y. Gray on Sep 08, 2017
Content may be subject to copyright.
Electronic copy available at: http://ssrn.com/abstract=2846398
290
Chapter eleven
Jonathan Gray, Liliana Bounegru, Stefania Milan and Paolo Ciuccarelli
Ways of seeing data: Towards a critical literacy for data visualizations as research
objects and research devices
Introduction
‘Every image embodies a way of seeing’, wrote the British art critic John Berger in his 1972
classic Ways of Seeing (Berger, 1972, p. 10). Through this book and accompanying
television series, he proposed elements of a critical literacy for making sense of the visual
landscapes that we inhabit – from reproductions of art historical masterpieces to the
advertising which adorns our cities and media environments. As well as guiding the attention
of his viewers and readers around specific images, he also sought to examine the way in
which images are reproduced and mediated (drawing on the work of the German
philosopher and critic Walter Benjamin), as well as the broader social, cultural, economic
and political contexts around them.
In the digital age, data visualizations are becoming an increasingly prominent genre for
the visual representation and mediation of collective life – from digital analytics dashboards
to interactive news graphics. Similarly, data visualizations are becoming more and more
popular in media and communications studies, and in the humanities and social sciences
more broadly. According to Stephen Few, ‘data visualization is the graphical display of
abstract information for two purposes: sense-making (also called data analysis) and
communication’ (2014, n.p.). This might include, for example, the representation of
information about numbers, words, relations, times or locations. But why have data
visualizations become so prevalent? And what does it mean to approach data visualization
as a research device? What might it offer media and communications scholars? Data
Electronic copy available at: http://ssrn.com/abstract=2846398
291
visualizations promise to assist us in making sense of complex data and complex
phenomena, allowing us to simplify and bring order to dense information, for explanatory
and communicative purposes. They may help us to analyze, filter, browse and explore
complex information. Paraphrasing Marshall McLuhan (1962), data visualizations are said to
amplify our senses and our abilities to make sense of the world around us. But data
visualizations are not a neutral tool. They come with particular ‘ways of seeing’, particular
analytical, mediation and narrative regimes regarding which we ought to be attentive as we
use them to do research and tell stories.
Taking inspiration from Berger’s agenda, in this chapter we argue that the use of data
visualizations as both instruments and objects of study requires commensurate forms of
critical literacy. We also draw on Philip Agre’s (1997) notion of ‘critical technical practice’ for
the social study of artificial intelligence and from Bernhard Rieder and Theo Röhle’s (2012)
notion of ‘methodological reflexivity’ in relation to digital methods in the humanities and
social sciences, both of which gesture towards forms of engagement with new research
methods and technologies that are accompanied by critical reflection on their uses. We seek
to extend their proposition that research practices ought to ‘oscillate […] between concrete
technical work and methodological reflexivity’ (Rieder and Röhle, 2012, p. 80) to the practice
of working with data visualizations.
We argue that data visualizations are not only bright adornments to our twenty first
century information environments. They embody and engender not only particular ways of
seeing (as Berger comments), but also ways of knowing and ways of organizing collective
life in our digital age. In other words, data visualizations reflect and articulate their own
particular modes of rationality, epistemology, politics, culture and experience. It is precisely
to emphasize these ‘world-making’ capacities, that in this chapter we prefer the term ‘device’
to the more commonly utilized term ‘tool’ to refer to data visualization. While the term ‘tool’
292
establishes the object as possessing coherence and connotes an instrumental relationship
between user and object, this chapter proposes to develop an account of data visualizations
as devices in order to draw attention to their capacities to ‘assemble and arrange the world
in specific social and material patterns’ (Law and Ruppert, 2013, p. 230). While arriving to
media and communication studies with particular inscriptions as described above, following
Law and Ruppert (2013) we argue that data visualizations as device are at the same time
indeterminate and open to multiple and diverse forms of usage. For this reason, to follow
Marres and Gerlitz (2015), deploying data visualization in media and communication
research may be thought of as a process of developing mechanisms to align its affordances
(and limitations) with the discipline-specific conventions of our fields. Given the growing role
of data visualizations in our information environments, we think it is vital to develop a critical
literacy to read, understand and work with them.
We propose a three part heuristic framework for what should be taken into account when
reading, working with and conducting research about data visualizations. In doing so this
chapter does not aim to develop a practical guide for the effective use of visualization in
media and communication research, per se. Instead we propose a framework to sensitize
researchers to forms of mediation embedded in data visualization as research devices in
order to support their critical and reflexive use.
This framework is organized around three forms of mediation that can be studied in
relation to data visualizations: (i) the mediation from world to data of the sources of
information that underpin visualizations; (ii) the mediation from data to image of the graphical
representations of this information; and (iii) the mediation from image to eye and the mind in
the socially, culturally and historically specific ‘ways of seeing’ engendered in the data
visualization. Each of these three forms of mediation can be studied with a broad range of
methods – from more familiar qualitative and quantitative approaches (such as visual and
293
textual analysis, interviews or surveys), to emerging digital and computational methods. We
describe and illustrate these different forms of mediation, and ways of studying them, with
reference to a collection of over two hundred data visualizations about public finances.
While for heuristic purposes our framework proposes the study of mediations between
‘world’, ‘data’, ‘image’ and ‘eye’, our intention is to provide a starting point to inform and
broaden inquiry rather than to propose a neat and rigid distinction between these different
elements. They are in fact mutually constitutive such that data constitute as much as they
represent the societal dynamics which they measure (Espeland and Stevens, 2008) and that
regimes of measurement and visualization are generative of specific publics, practices and
cultures (Ruppert, 2015). The forms of mediation which we propose are also not exhaustive.
For example, drawing on research on the reactivity of metrics (Espeland and Sauder, 2007;
Gerlitz and Lury, 2014), a fourth layer of mediation could be formulated, concerned with the
study of actions to intervene, respond to and modify the dynamics captured by data
visualizations. In addition to this, as the following sections will show, each of the three forms
of mediation is actually constituted by multiple sub-layers of mediation or inscription. These
will vary according to the nature of the ‘research apparatus’ in which visualizations are
embedded, in that different types of data and methods are accompanied by different types of
inscription (Ruppert, 2013). We also do not hold a strict order with regard to the proposed
layers of mediation.
The past few decades have seen the development of a body of literature dedicated to
data visualization and information graphics. In particular there is a growing body of books
and articles offering practical guidance as well as showcasing different examples,
294
techniques and approaches.1 This literature is itself interesting not only from a practical
perspective, but also as a way to understand the forms of mediation involved in the
composition of data visualizations – both those that receive attention and those that remain
neglected – as well as the aesthetics, cultures, values, ideals and practices associated with
their production. These resources can be useful as a source to disassemble and understand
the making of data visualization projects.
1 See, for example: Cairo, 2012; Card et al., 1999; Cleveland, 1993, 1994; Few, 2009, 2012, 2013; Heller and Landers, 2014;
Katz, 2012; Krum, 2013; Lima, 2011; McCandless, 2012, 2014; Meirelles, 2013; Munzner, 2014; Murray, 2013; Rendgen,
2012, 2014; Spence, 2014; Steele and Iliinsky, 2010; Tactical Technology Collective, 2014; Tufte, 1990, 1997ab, 2001;
Wong, 2013; Yau, 2011, 2013.
295
While many of these works focus on a single layer of mediation – that of data to the
image, looking at how information is translated into graphical form – we suggest that in using
and studying data visualizations it is essential to grasp not just the production of images
from data but also the datasets and data infrastructures that data visualizations draw on, as
well as the cultural practices and ideals implicated in the composition of visualizations which
invite a particular way of seeing. There have been a number of very informative works on
this latter topic (for example Drucker, 2014; Halpern, 2015), which we will explore further
below. Here our focus is less on data visualization as a field, but rather on developing the
elements of a critical reflexivity that would accompany and inform the practice of data
visualization in research and other contexts. This critical reflexivity is important not only in
the study of data visualizations as objects, but also to improve our abilities to deploy them as
research devices. Below we outline research outlooks and methods for studying all three
forms of mediation – as well as pointing to further resources that may be useful for each
one.
An Example: Visualizing Information about Public Finances
To illustrate our proposed framework we have chosen to work with a collection of data
visualization projects about public finances. These include data visualization projects from
media organizations, journalists, civil society organizations and public institutions. This
collection has been gathered in the context of research to empirically map how information
about public finances is used in the service of democratic engagement with fiscal policy
(Gray, 2015ab).
296
Figure 11.1: A selection from collection of examples of fiscal data visualizations (Gray,
2015b).
There are two main reasons why we consider this thematic focus suitable to make our case.
Firstly, there is a long tradition of work exploring public finances with information graphics.
Edward Tufte uses public finances to illustrate discussion of different techniques in his
classic The Visual Display of Quantitative Information citing a venerable tradition of
information graphics which ‘nearly always create the impression that spending and debt are
rapidly increasing’ (Tufte, 2001, p. 65). He alludes to the fiscal information graphics of the
Scottish engineer, economist and pioneer of statistical graphics William Playfair, such as this
one:
297
Figure 11.2: ‘Chart of the National Debt of England’ from William Playfair’s The Commercial
and Political Atlas (Playfair, 1801).
The second reason is that there is a heterogeneous constellation of different issues and
concerns that are associated with public finances and fiscal policy. As well as political and
economic questions from who in society pays how much tax and how public resources are
allocated – many policy areas are underpinned by discussions about public finances, from
international development to climate change. The complexity and competing narratives
298
around this topic makes it well suited to illustrate different approaches for studying data
visualizations.
Three Forms of Mediation and How to Study Them
In this section we will propose a research outlook and methods for studying each of the
three forms of mediation implicated in the creation of data visualizations that we have
outlined This may serve as a checklist of questions and a menu of different methods that
can be used.
1. From World to Data
As we shall see in the next section, many classic works on information visualization focus on
the mediation of data to image – in particular focusing on the avoidance of
misrepresentation. For example, Tufte suggests that ‘graphical excellence requires telling
the truth about data’ (Tufte, 2001, p. 51) and that data visualization designers should ‘let the
data speak for itself’ (1997a, p. 45), championing an aesthetic program of ‘graphical integrity’
that we shall challenge and explore further below. But what about the data itself? How does
the making of data shape the making of knowledge through visualizations?
Our first form of mediation to be studied is how the data used in the visualization is
generated and how this process inscribes itself in the knowledge produced through data
visualizations. This might include asking questions such as:
! What information or data is being represented in the visualization?
! What are the sources for this information? Where does the data come from?
! How is the data generated? What are the rationales, methods and standards
inscribed in the data infrastructures through which the data is generated?
! How is the data transformed or prepared?
299
! Which data sources are combined and how?
! How does the data selectively prioritize certain things over others?
The first step will often be establishing the data sources. In some cases details about the
sources of the information will be explicitly referenced or linked to. In other cases additional
work might be needed in order to identify these sources. This might be matching the precise
tables or datasets that are used from a number of possible contenders (for example, in
cases where the institution or database is given, but not the exact table or dataset); looking
in the software source code or documentation of the visualization to look for data sources; or
conducting interviews with the creators of a data visualization to establish what data was
used.
Once datasets are established for a given visualization, there are different approaches to
studying and analyzing them. For example, the study of data sources might draw on
literature about ‘sourcing practices’ from media studies and journalism studies (Cottle, 2003;
Manning, 2001; Hall et al., 1978; Berkowitz, 2009). Research can also be undertaken on the
‘data infrastructures’ implicated in the production of the datasets that are used in the
visualizations. Drawing on previous work in this area, we take the phrase ‘data infrastructure’
to designate socio-technical systems implicated in the creation, processing and distribution
of data (see, for example, Akrich, 1992). This might include elements such as standards
bodies, software systems, administrative procedures, committees, consultancy processes
and many other things (Gray and Davies, 2015).2 Here it may be useful to draw on
approaches from scienc and technology studies (STS) to trace the politics embedded in
these systems – for example through what has been called ‘infrastructure ethnography’ (see,
for example, Star and Ruhleder, 1996; Star, 1999; Bowker and Star, 2000). The composition
2 Gray and Bounegru are currently working on another project on ‘data infrastructure literacy’, which includes further
suggestions on approaches to studying and working with data infrastructures.
300
of data infrastructures will lead to the production of different types of data – from statistical
data to transactional ‘digital traces’ extracted from digital platforms.
How might we operationalise these approaches in relation to our collection of data
visualizations about public money? Firstly, can we identify the datasets that are used in the
visualizations? The different ways of citing or linking to data sources is itself something that
can be studied – as it may reflect different kinds of ideals, norms or practices of knowledge
production. A cursory look at different data sourcing practices reveals a wide range of ways
in which data is obtained and prepared, as well as varying approaches to publishing details
about datasets, software and methodology.
Figure 11.3: Data sources for ‘The Tax Gap’ visualization from the Guardian Datablog and
Information is Beautiful.
In our collection, some of the projects simply cite the name of the department from which the
data was derived. One example from the New York Times simply says ‘Source: Office of
301
Management and Budget’, with no further details.3 Many others give much more detail –
including directly linking to datasets, or even providing documentation on methodology about
how the data was sourced and transformed, what assumptions were made, and so on. For
example a piece from the US investigative news outlet ProPublica on ‘The Millions New York
Counties Coulda Got’ is accompanied by a short methodological walkthrough titled ‘How We
Analyzed New York County Tobacco Bonds’ explaining how they obtained the data, which
the cash flow models that they used and why, and the rationales behind other assumptions
they made in the application.4
Publishing data sources has become de rigueur amongst some data journalism outlets. A
piece from the South China Morning Post explains how data was extracted from PDF files
and provides links to the ‘raw data’ used in the visualizations.5 One data visualization from
the Guardian Datablog gives an accompanying spreadsheet which lists a separate source
URL for every figure that they have used.6 If visualization projects publish the source code of
their software, this can be studied in order to see the data files which have been used. For
example, the Budget Key project by Israeli non-profit Public Knowledge Workshop link to
their source code on the social software sharing platform GitHub which makes visible the
changes and transformations that have been made on the data.7
As well as looking at the sourcing and transformation of the data, we might also look at
the selection of particular datasets and tables to highlight or narrate certain aspects of public
3 http://www.nytimes.com/interactive/2010/02/02/us/politics/20100201-budget-porcupine-graphic.html?_r=0 (accessed 10
March 2016).
4 See https://projects.propublica.org/graphics/ny-millions and https://www.propublica.org/article/how-we-analyzed-new-
york-county-tobacco-bonds (accessed 10 March 2016).
5 http://widgets.scmp.com/infographic/20140304/budget2014/data/ (accessed 10 March 2016).
6 http://www.theguardian.com/news/datablog/2010/sep/22/tax-gap-information-beautiful (accessed 10 March 2016).
7 https://github.com/OpenBudget/ (accessed 10 March 2016).
302
finances such as spending, revenue or debt. Rather than simply ‘telling the truth’ about
public finances – these selections emphasize and de-emphasize different aspects of fiscal
policy. We shall further examine the affordances of different visual forms in the next section.
Suffice to say here that the selection of different datasets and tables is an important step
that should be taken into account in the study of data visualizations – prior to their translation
into graphical form. We can compare the data mediated in the visualization to the other data
which is made available. Which indicators, subtables or items of data have been selected,
which have been left out and how does this guide our attention towards some things and not
others?
An interactive news application from The Times (UK) called ‘The Wall of Debt’ shows a
dominant wall of red bricks depicting ‘national debt’, and a comparatively small green wall
called ‘cuts’.8 The visual editorial decision to select these two items is an important one in
reading this interactive graphic, engaging with a particular political economic narrative about
tackling public debt through spending cuts. Another short video clip called ‘Debtris’ by David
McCandless presents a wide variety of other fiscal data points, ultimately highlighting how
the global cost of the credit crisis dwarfs other sums such as the Organization of the
Petroleum Exporting Countries’ climate change fund, the budget of the UN, African debt to
the West and the funds needed to ‘save the Amazon rainforest’.9 As well as looking at the
selection of data on a case by case basis for individual visualizations, comparative analysis
may be undertaken across a larger collection.
We might also study the composition of the data itself. We could study the headers,
categories and classifications within the datasets to understand which variables and
indicators are selected and prioritized. Content analysis of documentation and associated
8 http://appliedworks.co.uk/work/the-times-defining-a-new-era-of-data-journalism/ (accessed 10 March 2016).
9 https://www.youtube.com/watch?v=K7Pahd2X-eE (accessed 10 March 2016).
303
materials may provide further details about the units, forms of analysis, methodology,
software, standards and other details about how the data was generated. This might also be
supplemented by interviews with those involved in the creation of the data about their
decisions. It may be that datasets are generated from other more complex database
systems which may be of interest to study. In the case of public financial data, we might look
into politics, rationales and ways of knowing inscribed into the financial management
information systems through which datasets about public finance are generated. These may
be studied as socio-technical systems through, for example, document analysis and
interviews. Sometimes there may be documentation manuals which explain how these
systems function. In the case of public sector data infrastructures, if details are not already
published, then there may be routes for formally requesting them through access to
information laws. Researchers and civil society groups interested in studying what was the
most detailed source of information about public spending in the UK (the Combined Online
Information System, or COINS) submitted official ‘freedom of information’ requests about the
database. While their requests for its contents were initially turned down, they submitted
follow up requests asking about the database and training materials for the database, which
were successful.10
It may also be fruitful to study the categories and classification systems articulated by the
data. For example, any of the data visualizations about public money use different
categories to describe different areas of expenditure. Where do these categories come
from? How did they become the way that they are? On the one hand we can look at the
administrative contexts in which the datasets are generated – such as the organization of
public sector bodies. We might also look at which kinds of data standards shape public
sector data systems. For example, the UN COFOG (Classifications of Functions of
10 https://www.whatdotheyknow.com/request/schema_for_coins_database (accessed 10 March 2016).
304
Government) standard is widely used as a reference in order to facilitate comparability
between different national budgeting processes.11 This is particularly relevant when it comes
to looking at visualizations including multiple countries – such as international development
or climate finance. The genesis of these data standards can also be studied through
analysis of (historical) document collections, the study of information systems (for example,
using approaches from software or platform studies), interviews and ethnographic studies.
2. From Data to Image
The second form of mediation in our heuristic framework is how visualizations mediate the
data sources they draw on into graphical form. This might including addressing questions
such as:
! How is the data mediated into graphical form?
! What kinds of graphical techniques, methods and technologies have been used?
! What are their affordances? How do they guide our attention towards different
aspects of the data?
! What design decisions have been taken? What are their consequences?
These concerns are prominent in both research and practical literature around data
visualizations. Edward Tufte talks of ‘graphical methods that organize and order the flow of
graphical information presented to the eye’ (Tufte, 2001, p. 154). He places a significant
premium on ‘graphical integrity’ of information visualizations such that they ‘defeat graphical
distortion and ambiguity’ (Tufte, 2001, p. 77). While this is one important way of looking at
how data is mediated into graphical form, there are many other important aspects to study.
Rather than just looking at accuracy, fidelity, and how graphics may be truthful or untruthful
about data – we can also step back and look at their affordances – such as how they
11 http://unstats.un.org/unsd/cr/registry/regcst.asp?Cl=4 (accessed 10 March 2016).
305
articulate structures and relationships or how they organize space, time, quantity and
categories in relation to the data.
In order to study these things we may learn a great deal from contemporary literature
about the creation of data visualizations, as well as classic literature on information graphics.
These can be used in order to understand the composition of data visualizations, or to
‘reverse engineer’ them, which helps us to develop a critical and reflexive approach to the
use of data visualization as a research device. Jacques Bertin’s 1967 Semiology of Graphics
proposes a series of ‘retinal variables’ (1983, p. 9) including size, value, texture, color,
orientation, shape as a starting point for analyzing and working with different types of
information graphics. This work provides an extensive and richly illustrated overview of how
different ‘components’ of information can be mapped onto different visual forms. In a similar
vein, Tufte’s The Visual Display of Quantitative Information contends that ‘data graphics
visually display measured quantities by means of the combined use of points, lines, a
coordinate system, numbers, symbols, words, shading and color’ (2001, p. 9).
While Bertin’s work is explicitly limited to print graphics which fit on a ‘sheet of white
paper’ (1983, p. 42) and many of Tufte’s seminal works also focus on print graphics, many of
the elements they discuss have been adopted and developed in relation to digital and
interactive data visualizations. Thus we may read these influential works alongside other
more recent works that cover digital tools and methods in order to understand how different
visual forms are implicated in the visualizations under study – including table graphics,
maps, timelines, sparklines, networks, graphs, flow diagrams, small multiples, motion charts,
bubble charts and treemaps. These different forms can also be combined to highlight
different aspects of the data. We may also look at the affordances of the software or the
platforms that were used to create data visualizations (see, for example, Wright, 2008;
Manovich, 2002, 2011, 2014). As well as desktop software applications, there is a growing
306
number of visualization tools which enable visualizations to be generated and embedded.
There is also a growing number of software libraries and components which are widely used
to translate data into different graphical forms.
How might we use these approaches to study our collection of data visualizations about
public money? We could look at the different graphical forms used in data visualizations to
organize attention around different aspects of public finances. For our collection, this could
include examining how different graphical elements are used in order to:
! Show a ‘bigger picture’ of breakdowns of totals into different categories;
! Put different figures into context;
! Show the geographical distribution of funds;
! Show trends or developments over time;
! Show breakdown of funds by sector or recipient;
! Show networks of financial flows;
! Compare different parts of the budgeting cycle (for example, commitments and
actual expenditure);
! Compare revenues, expenditures and debts;
! Show allocations per capita.
The formal characteristics of different visualizations in a collection can be studied with
reference to either a pre-defined vocabulary of elements or through an ‘open coding’ or
‘emergent coding’ process (see Strauss & Corbin, 1998). As similar visual elements can play
different roles in different contexts, it is crucial to note not only their presence or absence,
but their relations with other elements in the visualization as a whole. For example, compare
the use of bubbles in the following visualizations:
307
Figure 11.4: Total UK public spending (The Guardian);
http://www.theguardian.com/news/datablog/2011/oct/26/government-spending-department-
2010-11 (accessed 10 March 2016).
Figure 11.5: Geographical distribution of international development spending (UNDP);
http://open.undp.org/#2014 (accessed 10 March 2016).
308
Figure 11.6: Health spending per person compared with income per person over time
(Gapminder World); http://www.gapminder.org/world/ (accessed 10 March 2016).
309
Figure 11.7: Spending by different government departments over time (Government of
Paraguay); https://www.contrataciones.gov.py/datos/visualizaciones/contratos (accessed 10
March 2016).
Figure 11.8: Network of Dutch public spending on IT services and software (Adriana
Homolova); http://www.homolova.sk/dh/it/# (accessed 10 March 2016).
310
Figure 11.9: Comparing commitments and spending in Israeli budget (Budget Key);
http://www.obudget.org (accessed 10 March 2016).
With these examples we can see how the bubbles are arranged, scaled and colored in order
to draw out different characteristics of public finances – from representing hierarchical
relationships in national budgets, to the locations of development funds, to health spending
per capita over time, to spending per department over time, to relationships between
recipients of public contracts, to the differences between commitments and spending. In
each case, the bubbles are used in different ways to represent different things – with the
position, size and color having varying significance for each graphic. These different ways of
organizing visual phenomena thus articulate and emphasize different ways of organizing
knowledge about public finances.
As well as studying the characteristics of visualizations, their creators could be
interviewed regarding their techniques, methods, software tools and design choices. Where
software repositories are available these can also be analyzed in order to understand how
311
data sources are being mediated into graphics – whether for individual repositories or for
collections. We might, for example, study repositories associated with fiscal data
visualization projects on GitHub.12
3. From Image to Eye
The third and final form of mediation in our heuristic framework is how the image appears to
the eye and to the mind. As we saw in the last section, visualizations emphasize different
aspects of data through design choices, techniques and software which organize data into
graphical form. But these graphical forms also engender and depend on different socially,
culturally and historically contingent ways of seeing data. Hence we might ask:
! What kinds of visual cultures and practices are implicated or reflected in the data
visualization? Where do these come from?
! What forms of usage are inscribed in the visualization? Who are the publics of the
data visualization? How is it circulated, cited and shared?
Rather than seeing the visual forms as neutral instruments for making evidence visible –
or as Tufte puts it, ‘instruments for reasoning about quantitative information’ (Tufte, 2001, p.
9) – we can study their genealogies, aesthetics and epistemological affordances to situate
them in relation to other ideals, values and practices. For example, visualizations can be
read against the background of histories of science, technology and modernity which explore
the relationship between vision, knowledge and image making practices – from accounts of
occidental ‘ocularcentricism’ (see Jay, 1988) to the role of visual cultures in the development
of conceptions of objectivity (see Daston and Galison, 2010).
12 Bounegru and Gray are currently involved in developing tools and methods for working with metadata from GitHub as
part of other research projects.
312
There have been several recent works which make the case for drawing on
hermeneutical approaches from the humanities to enhance the study of data visualizations.
For example, in her book Graphesis Johanna Drucker advocates ‘critical study of visuality
from a humanistic perspective’ in order to ‘de-naturalize the increasingly familiar interface
that has become so habitual in daily use’ (Drucker, 2014, p. 9–10). She explores the
emergence of contemporary ‘visual epistemology’ with reference to a broad range of
developments in art, architecture, design, industry, philosophy and computer science – from
the emergence of graphical design to interdisciplinary deliberations about visual abstraction
between artists and designers associated with the Bauhaus school in Weimar Germany. In
her Beautiful Data, Orit Halpern also similarly aims to ‘denaturalize and historically situate’
ideals and practices of data visualization (Halpern, 2015). She traces the genesis of what
she calls ‘communicative objectivity’ that has come to be associated with data visualizations
drawing on a different disciplinary constellation focusing on cybernetics, communication
science, behavioral science, engineering, management studies, urban planning and military
research. There is also a growing body of literature focusing on the aesthetics of data
visualization (see Manovich, 2002; Jevbratt, 2004; Lau and Vande Moere, 2007; Whitelaw,
2008; Sack, 2011; Cubitt, 2015) as well as on the development of different visual forms such
as the timeline (see Rosenberg and Grafton, 2013) or those associated with network
analysis (see Freeman, 2004). The narrative dimension of data visualizations can also be
studied (see Segel and Heer, 2010; Venturini et al., 2016).
How might these kinds of approaches be adopted to study specific data visualizations –
such as our collection about public finances? We could study the aesthetics of these
visualizations – for example, the clean, minimalistic style adorned with primary color palettes
and icons that has become so widely adopted in many contemporary information graphics.
313
Figure 11.10: Government spending over £25,000 (The Guardian);
http://www.theguardian.com/news/datablog/2010/nov/19/government-spending-information-
beautiful (accessed 10 March 2016).
Figure 11.11: Openaid.se (Swedish International Development Cooperation Agency);
http://www.openaid.se/aid/2014/ (accessed 10 March 2016).
This is similar to the aesthetic that is advocated by Tufte: championing efficiency and
parsimony, maximizing the ‘data-ink’ ratio and eliminating ‘chartjunk’. He proposes that ‘the
design of statistical graphics is a universal matter […] like mathematics’ and that insight into
314
the design of visualizations may be obtained through the study of ‘excellence in art,
architecture and prose’ (Tufte, 2001, p. 10). Peter Galison (1990) has previously studied the
links between architectural modernism and the universal aspirations of logic, mathematics
and philosophy in the first few decades of the twentieth century. Many of the fiscal data
visualizations in our collection look to share this aesthetic. Several of them incorporate icons
or pictograms reminiscent of those associated with the Isotype Institute of Marie and Otto
Neurath, which has exercised an important influence on the contemporary aesthetics of data
visualization (Zambrano and Engelhardt, 2008; Mayr and Schreder, 2014; Headrick, 2000;
Rayward, 2008).
Figure 11.12: Science spending in the UK (Scienceogram);
http://scienceogram.org/summary/ (accessed 10 March 2016).
315
Figure 11.13: ‘Home and Factory Weaving in England’ (Neurath, 1939).
Genealogical study may help to enrich research about what has given rise to the contingent
forms that data visualizations may take – identifying a range of different influences and
origins of different visual forms. Starting points may be provided through interviews with
designers as well as content analysis of relevant design materials – which may be
complemented with historical texts and archival research.
We might also study the specific visual forms which are transposed to mediate public
money. Many visualizations are described and/or organized as dashboards – giving users
an overview of multiple key indicators and trends over time in a single viewing pane, inviting
narratives of oversight, optimization, balance and control (see Tkacz, 2015).
316
Figure 11.14: City of Boston’s Open Budget Application;
http://budget.data.cityofboston.gov/#/ (accessed 10 March 2016).
As well as studying data visualizations as cultural forms from a humanistic perspective, we
can also trace their circulation, reception and how they are used and viewed by different
publics (see Kennedy et al., 2016). In addition to interviews and workshops, digital methods
could be utilized to extract and analyze traces from digital platforms and online spaces in
order to review the contexts in which visualizations are being used and shared (Rogers,
2013). For example, we could query for names or URLs of projects on social media, search
engine results, news media, or collections of documents. This may also help to chart the
particular publics and context of usage of the visualizations – which can assist with their
study as social and cultural forms.
Conclusion
In this chapter we have proposed a heuristic framework that may be used to develop critical
reflexivity around the use of data visualization as an object of research or a communicative
317
or analytical device in research. We have illustrated this framework with reference to a
collection of data visualizations about public money, and suggested some methods and
approaches that could be used to study them.
Firstly we looked at the study of data sources, focusing on research approaches to
examine how they selectively articulate and mediate different aspects of the world –
including how to identify sources, tracing how they have been transformed and studying data
infrastructures. Secondly we looked at ways to study how the data sources are mediated
into graphical form – including through the analysis of their visual properties, software and
design choices. Thirdly and finally, we looked at ways of studying graphical forms as
socially, culturally and historically contingent forms engendering different ways of seeing –
including by tracing their diverse influences, and by analyzing their circulation and contexts
of usage.
We hope that the study of these three forms of mediation through some of the
approaches that we have discussed in this chapter may provide a useful starting point for
researchers who wish to use data visualization as a research object or device. As data
visualizations become more and more central, prominent and familiar as ways of knowing
and organizing phenomena, we think it becomes imperative to develop a richer
understanding of the ways of seeing and ways of knowing that they engender. Just as
Berger’s Ways of Seeing helped to advance broader awareness of the critical study of
images and visual culture, so we hope that further research in this area will advance literacy
around ways of seeing data and ways of seeing with and through data visualizations. As
visualization tools and practices become more and more ubiquitous, this might include not
only the development of a critical hermeneutics, but also new kinds of self-reflexive praxis
for the creation and reconfiguration of visualizations which are attentive to the forms of
mediation that we have outlined. Experimentation in this direction might be informed by calls
318
for non-reductive visualization (Manovich, 2002), humanistic interfaces (Drucker, 2014),
feminist data visualization (D’Ignazio, 2015), inventive methods (Lury and Wakeford, 2012),
rethinking dashboards (Tkacz, 2015), and critical analytics (Rogers, 2015).
References
Agre, P. E. (1997) Toward a Critical Technical Practice: Lessons Learned in Trying to
Reform AI. In G. C. Bowker, S. Star, L. Gasser et al. (Eds.), Social Science,
Technical Systems, and Cooperative Work: Beyond the Great Divide (Mahwah:
Lawrence Erlbaum), pp. 131–58.
Akrich, M. (1992) The De-Scription of Technical Objects. In W. Bijker and J. Law (Eds.),
Shaping Technology/Building Society: Studies in Sociotechnical Change
(Cambridge, MA: MIT Press), pp. 205–24.
Berger, J. (1972) Ways of Seeing (London: Penguin).
Berkowitz, D. (2009) Reporters and Their Sources. In K. Wahl-Jorgensen and T. Hanitzsch
(Eds.), The Handbook of Journalism Studies (New York: Routledge), pp. 102–15.
Bertin, J. (1983) Semiology of Graphics: Diagrams, Networks, Maps. (Madison: University of
Wisconsin Press).
Bowker, G., and Star, S. (1999) Sorting Things Out: Classification and its Consequences
(Cambridge, MA: MIT Press).
Cairo, A. (2012) The Functional Art: An Introduction to Information Graphics and
Visualization (Berkeley: New Riders).
Card, S. K., Mackinlay, J., and Schneiderman, B. (Eds.). (1999) Readings in Information
Visualization: Using Vision to Think (San Francisco: Morgan Kaufmann).
Cleveland, W. S. (1993) Visualizing Data (Murray Hill: Hobart Press).
Cleveland, W. S. (1994) The Elements of Graphing Data (Murray Hill: Hobart Press).
319
Cottle, S. (2003). News, Public Relations and Power: Mapping the Field. In S. Cottle (ed.),
News, Public relations and Power (London: Sage), pp. 3–24.
Cubitt, S. (2015) Data Visualization and the Subject of Political Aesthetics. In D. Berry and
M. Dieter (Eds.), Postdigital Aesthetics: Art, Computation And Design (London:
Palgrave Macmillan), pp. 179–90.
D’Ignazio, C. (2015) What Would Feminist Data Visualization Look Like? MIT Center for
Civic Media 20 December: https://civic.mit.edu/feminist-data-visualization (accessed
10 March 2016).
Daston, L. and Galison, P. (2010) Objectivity (Cambridge, MA: MIT Press).
Drucker, J. (2014) Graphesis: Visual Forms of Knowledge Production (Cambridge, MA:
Harvard University Press).
Espeland, W. N. and Stevens, M. L. (2008) A Sociology of Quantification. European Journal
of Sociology 49(3): 401–36.
Espeland, W. N. and Sauder, M. L. (2007) Rankings And Reactivity: How Public Measures
Recreate Social Worlds. American Journal of Sociology 113(1): 1–40.
Few, S. (2009) Now You See It: Simple Visualization Techniques for Quantitative Analysis
(Oakland: Analytics Press).
Few, S. (2012) Show Me the Numbers: Designing Tables and Graphs to Enlighten. 2nd ed
(Burlingame: Analytics Press).
Few, S. (2013) Information Dashboard Design: Displaying Data for At-a-Glance Monitoring
(Burlingame: Analytics Press).
Few, S. (2014) Data Visualization for Human Perception. In M. Soegaard and R. Dam
(Eds.), The Encyclopedia of Human-Computer Interaction. 2nd Ed. (Aarhus: The
Interaction Design Foundation), available at: https://www.interaction-
320
design.org/literature/book/the-encyclopedia-of-human-computer-interaction-2nd-
ed/data-visualization-for-human-perception (accessed 20 March 2016).
Freeman, L. C. (2004) The Development of Social Network Analysis (Vancouver:
Booksurge).
Galison, P. (1990) Aufbau/Bauhaus: Logical Positivism and Architectural Modernism. Critical
Inquiry 16(4): 709–52.
Gerlitz, C and Lury, C. (2014) Social Media and Self-Evaluating Assemblages: On Numbers,
Orderings and Values. Distinktion 15(2): 174–88.
Gray, J. (2015a) Open Budget Data: Mapping the Landscape. Available at:
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2654878 (accessed 15 March
2016).
Gray, J. (2015b) Examples of Fiscal Data Visualisations. figshare. Available at:
https://figshare.com/articles/Examples_of_Fiscal_Data_Visualisations/1548331
(accessed 15 March 2016).
Gray, J. and Davies, T. (2015) Fighting Phantom Firms in the UK: From Opening Up
Datasets to Reshaping Data Infrastructures? Working paper presented at the Open
Data Research Symposium at the 3rd International Open Government Data
Conference (Ottawa). Available at:
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2610937 (accessed 10 March
2016).
Hall, S., Critcher, C., Jefferson, T. et al. (1978) Policing the Crisis: Mugging, the State, and
Law and Order (London: Macmillan).
Halpern, O. (2015) Beautiful Data: A History of Vision and Reason Since 1945 (Durham:
Duke University Press).
321
Headrick, D. R. (2000) When Information Came of Age: Technologies of Knowledge in the
Age of Reason and Revolution, 1700-1850 (Oxford: Oxford University Press).
Heller, S. and Landers, R. (2014) Raw Data: Infographic Designers’ Sketchbooks (London:
Thames and Hudson).
Jay, M. (1988). The Rise of Hermeneutics and the Crisis of Ocularcentrism. Poetics Today
9(2): 307–26.
Jevbratt, L. (2004) The Prospect of the Sublime in Data Visualizations. `
Katz, J. (2012) Designing Information: Human Factors and Common Sense in Information
Design (Hoboken: Wiley).
Kennedy, H., Hill, L., Aiello, G. et al. (2016) The Work that Visualisation Conventions Do.
Information, Communication & Society 19(6): 715–35.
Krum, R. (2013) Cool Infographics: Effective Communication with Data Visualization and
Design (Indianapolis: Wiley).
Lau, A. and Vande Moere, A. (2007) Towards a Model of Information Aesthetics in
Information Visualization. In Proceedings of the 11th International Conference
Information Visualization (Washington, DC: IEEE Computer Society), pp. 87–92.
Law, J. and Ruppert, E. (2013) The Social Life of Methods: Devices. Journal of Cultural
Economy 6(3): 229–40.
Lima, M. (2011) Visual Complexity: Mapping Patterns of Information (New York: Princeton
Architectural Press).
Lury, C. and Wakeford, N. (Eds.) (2012) Inventive Methods: The Happening of the Social
(London: Routledge).
Manning, P. (2001) News and News Sources: A Critical Introduction (London: Sage).
322
Manovich, L. (2002). Data Visualisation as New Abstraction and Anti-sublime. In B. Hawk, D.
Rieder and O. Oviedo (Eds.), Small Tech: The Culture of Digital Tools, Electronic
Mediations (Minneapolis: University of Minnesota Press), pp. 3–9.
Manovich, L. (2011) What is Visualisation? Visual Studies 26(1): 36–49.
Manovich, L. (2014) Visualization Methods for Media Studies. In C. Vernallis, A. Herzog and
J. Richardson (Eds.), Oxford Handbook of Sound and Image in Digital Media
(Oxford: Oxford University Press), pp. 253–78.
Marres, N. and Gerlitz, C. (2015) Interface Methods: Renegotiating Relations Between
Digital Social Research, STS and Sociology. The Sociological Review 64(1): 21–46.
Mayr, E. and Schreder, G. (2014) Isotype Visualizations. A Chance for Participation & Civic
Education. JeDEM – eJournal of eDemocracy and Open Government 6(2), 136–50.
McCandless, D. (2012) Information is Beautiful (New York: Collins).
McCandless, D. (2014) Knowledge is Beautiful (New York: Collins).
McCosker, A. and Wilken, R. (2014) Rethinking ‘Big Data’ as Cisual Knowledge: The
Sublime and the Diagrammatic in Data Visualisation. Visual Studies 29(2): 155–64
McLuhan, M. (1962) The Gutenberg Galaxy: The Making of Typographic Man (Toronto:
University of Toronto Press).
Meirelles, I. (2013) Design for Information: An Introduction to the Histories, Theories, and
Best Practices Behind Effective Information Visualizations. Beverly, MA: Rockport
Publishers.
Munzner, T. (2014) Visualization Analysis and Design (Boca Raton: CRC Press).
Murray, S. (2013) Interactive Data Visualization for the Web (Sebastopol: O’Reilly).
Neurath, O. (1939) Modern Man in the Making (New York: Knopf).
323
Playfair, W. (1801) The Commercial and Political Atlas, Representing, by Means of Stained
Copper-Plates, the Progress of the Commerce, Revenues, Expenditure, and Debt of
England during the Whole of the Eighteenth Century. 3rd ed (London: Wallis).
Rayward, W. B. (Ed.). (2008) European Modernism and the Information Society: Informing
the Present, Understanding the Past (Aldershot: Ashgate).
Rendgen, S. (2012) Information Graphics (Köln: Taschen).
Rendgen, S. (2014) Understanding the World: The Atlas of Infographics (Köln: Taschen).
Rieder, B. and Röhle, T. (2012) Digital Methods: Five Challenges. In D. M. Berry (Ed.),
Understanding Digital Humanities (London: Palgrave Macmillan), pp. 67–84.
Rogers, R. (2013) Digital Methods (Cambridge, MA: MIT Press).
Rogers, R. (2015) From Dashboards to Critical Analytics. Paper presented at Data Power
Conference (University of Sheffield).
Rosenberg, D. and Grafton, A. (2013) Cartographies of Time: A History of the Timeline (New
York: Princeton Architectural Press).
Ruppert, E. (2015) Doing the Transparent State: Open Government Data as Performance
Indicators. In: R. Rottenburg, S. E. Merry, S.-J. Parkand and J. Mugler (Eds.), A
World of Indicators: The Making of Governmental Knowledge through Quantification.
(Cambridge: Cambridge University Press), pp. 1–18.
Ruppert, E., Law, J. and Savage, M. (2013) Reassembling Social Science Methods: The
Challenge of Digital Devices. Theory, Culture & Society 30(4): 22–46.
Sack, W. (2011) Aesthetics of Information Visualization. In M. Lovejoy, C. Paul and V.
Vesna (Eds.), Context Providers: Conditions of Meaning in Media Arts (Bristol:
Intellect), pp. 123–50)..
Schoffelen, J., Claes, S., Huybrechts, L. et al. (2015) Visualising Things. Perspectives on
How to Make Things Public Through Cisualisation. CoDesign 11(3–4): 179–92.
324
Segel, E. and Heer, J. (2010) Narrative Visualization: Telling Stories with Data. IEEE
Transactions on Visualization and Computer Graphics 16(6): 1139–48.
Spence, R. (2014) Information Visualization: An Introduction (New York: Springer).
Star, S. (1999) The Ethnography of Infrastructure. American Behavioral Scientist 43(3): 377–
91.
Star, S. and Ruhleder, K. (1996) Steps Toward an Ecology of Infrastructure: Design and
Access for Large Information Spaces. Information Systems Research 7(1): 111–34.
Steele, J. and Iliinsky, N. (Eds.) (2010) Beautiful Visualization: Looking at Data through the
Eyes of Experts (Sebastopol: O’Reilly).
Strauss, A. and Corbin, J. (1998) Basics of Qualitative Research. Techniques and
Procedures for Developing Grounded Theory. 2nd ed (London: Sage).
Tactical Technology Collective (2014) Visualising Information for Advocacy.
Tkacz, N. (2015) Connection Perfected: What the Dashboard Reveals. Keynote Address at
Digital Methods Winter School 2015 (University of Amsterdam).
Tufte, E. R. (1990) Envisioning Information (Cheshire: Graphics Press).
Tufte, E. R. (1997a) Visual Explanations: Images and Quantities, Evidence and Narrative
(Cheshire: Graphics Press).
Tufte, E. R. (1997b) Visual and Statistical Thinking: Displays of Evidence for Making
Decisions (Cheshire: Graphics Press).
Tufte, E. R. (2001) The Visual Display of Quantitative Information. 2nd ed (Cheshire:
Graphics Press).
Tukey, J. (1977) Exploratory Data Analysis (Reading: Addison-Wesley).
Venturini, T., Bounegru, L., Jacomy, M. and Gray, J. (forthcoming) How to Tell Stories with
Networks: Exploring the Narrative Affordances of Network Graphs with the Iliad. In
325
M. T. Schaefer and K. van Es (Eds.), The Datafied Society: Social Research in the
Age of Big Data (Amsterdam: University of Amsterdam Press).
Whitelaw, M. (2008) Art Against Information: Case Studies in Data Practice. Fibreculture 11:
http://eleven.fibreculturejournal.org/fcj-067-art-against-information-case-studies-in-
data-practice/ (accessed 10 March 2016).
Wong, D. M. (2013) The Wall Street Journal Guide to Information Graphics: The Dos and
Don’ts of Presenting Data, Facts, and Figures (New York: Norton).
Wright, R. (2008) Data Visualization. In M. Fuller (ed.), Software Studies: A Lexicon.
Cambridge, MA: MIT Press), pp. 78–87.
Yau, N. (2011) Visualize This: The FlowingData Guide to Design, Visualization, and
Statistics (Indianapolis: Wiley).
Yau, N. (2013) Data Points: Visualization That Means Something (Indianapolis: Wiley).
Zambrano, R. N. and Engelhardt, Y. (2008) Diagrams for the Masses: Raising Public
Awareness – From Neurath to Gapminder and Google Earth. In G. Stapleton, J.
Howse and J. Lee (Eds.), Diagrammatic Representation and Inference (Berlin:
Springer), pp. 282–92.
