Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
Journal of Quantitative Description: Digital Media 3(2023), 1–34 10.51685/jqd.2023.014
Copyright © 2023 (Camila Mont’Alverne, Amy Ross Arguedas, Sumitra Badrinathan, Benjamin Toff,
Richard Fletcher, Rasmus Kleis Nielsen). Licensed under the Creative Commons Attribution-
NonCommercial-NoDerivatives 4.0 International Public License. Available at: http://journalqd.org
Domain-specific influence on Facebook: How topic matters
when assessing influential accounts in four countries
CAMILA MONT’ALVERNE
AMY ROSS ARGUEDAS
University of Oxford, United Kingdom
SUMITRA BADRINATHAN
American University, USA
BENJAMIN TOFF
RICHARD FLETCHER
RASMUS KLEIS NIELSEN
University of Oxford, United Kingdom
Against the backdrop of rising concern over misinformation and
disinformation, a growing number of studies have considered the important
role played by influential social media accounts when particular news
stories attract attention online—with special attention given to Facebook,
the most widely-used social network for news. However, little is known
about what kinds of accounts are among the most influential information
curators on Facebook, and where news organizations fit into this broader
landscape. In this study, we examine how influence on Facebook plays out
across different national contexts and different topics. We draw on a unique
dataset from CrowdTangle, sampling over a six-month period in 2021
across four countries (Brazil, India, the United Kingdom, and the United
Camila Mont’Alverne (corresponding author): camila.montalverne@politics.ox.ac.uk
Date submitted: 2023-01-04
Mont’Alverne et al. Journal of Quantitative Description: Digital Media 3(2023)
2
States). We compare what kinds of sources (e.g., news organizations,
politicians, or other kinds of influential accounts and groups) are among the
most influential accounts in each location when it comes to three specific
subjects: COVID-19, political leaders in each country, and climate
change—which we also compare to general queries that do not specify a
subject domain. Our findings show that the types of influential accounts on
Facebook vary considerably by subject domain and country. News media
accounts are among the largest share of these influential accounts in each
country, but not necessarily the types of news media organizations
presumed to be most influential offline.
Keywords: Facebook, influential accounts, social media
Introduction
Against the backdrop of rising concern over misinformation and disinformation, a
growing number of studies have considered the important role played by influential social
media accounts when particular news stories attract attention online. Studies often focus
on Facebook, the social platform used more widely than any other service as an
intermediary for news (Andı, 2021; Pew Research Center, 2021). Research has found that
Facebook diversifies people’s news exposure (Fletcher et al., 2021; Stier et al., 2022), but
also that it drives attention to low quality, emotionally charged sources that advance
extreme views on particular topics (Hagey & Horwitz, 2021; Vicario et al., 2016),
potentially contributing to eroding levels of trust in news worldwide (Kalogeropoulos et
al., 2019). In common with other social platforms, but unlike many other forms of
information access online, algorithms and networks of social ties on Facebook combine to
surface political information even when people are using it for other purposes (Boczkowski
et al., 2018; Fletcher & Nielsen, 2018).
JQD: DM 3(2023) Domain-specific influence on Facebook
3
Little is known about what kinds of accounts are among the most influential
information curators on Facebook (as in Thorson & Wells [(2016)]), promoting the stories
that go viral and driving attention to them. Prior studies examining what accounts people
interact with most on Twitter (Heiberger et al., 2021) and YouTube (Lewis, 2020) are
consistent with theories suggesting that conventional news media organizations—though
still present—may be relatively less influential in these digital spaces as agenda-setters or
gatekeepers of information than they are in other media environments. Legacy news outlets
must compete for attention with a wide range of actors on platforms including not only
“digital-born” news organizations (Nicholls et al., 2016) and partisan news organizations,
but also accounts associated with politicians and politically-aligned groups, celebrities,
sport stars, and other influential voices (Freelon et al., 2018; Park et al., 2015) who also
wield influence in these spaces. Studies have also shown that news, much less political
news or “fake news,” tends to be only a small fraction of what most people see and interact
with online (Allen et al., 2020; Beam et al., 2018; Chauchard & Garimella, 2022; Giglietto
et al., 2022; Guess, 2021; Wojcieszak et al., 2021)—or, for that matter, offline (Konitzer
et al., 2021). In fact, Facebook itself has emphasized in its public reports that the most
viewed URLs posted on the platform in the US are generally unrelated to news or politics
at all (Facebook Transparency Center, 2021).
These previous findings, however, may obscure the extent to which legacy news
organizations continue to wield influence in particular online contexts when it comes to
specific news-related subjects. After all, much of the existing research on influential
accounts on platforms does not differentiate by subject matter (e.g., Newton, 2020). Nor
are media environments outside the US typically the focus. In one study that does make
these distinctions, Majó-Vázquez and colleagues’ (2017) found that legacy news was
particularly dominant in driving interaction on Twitter during the 2017 French election
debates. More recently, when studying amplification of elite sources about COVID-19 on
Twitter in the US, Gallagher et al. (2021) also showed how different groups—including
news organizations, health professionals, and politicians—amplify messages from
different sources, most often elites that are demographically similar to them. Additionally,
Mont’Alverne et al. Journal of Quantitative Description: Digital Media 3(2023)
4
Altay and colleagues (2022) show that outlets rated as more trustworthy dominated news
use both on web and on Facebook before and during the pandemic in different countries
(though 14% of Facebook engagement was with untrustworthy outlets). Together, these
studies reinforce the notion that understanding who is influential on social media may
depend on the topic or geographic location in which influence is studied. Such differences
may be overlooked if researchers simply aggregate across entire corpuses of social media
messages or populations.
In this study, we go beyond prior analyses of link-sharing patterns on Facebook
overall—especially those that focus only on a single country (typically the US)—and
examine how influence on the platform plays out across different national contexts. As in
Kim’s (2009) work on “issue publics,” we focus on domain-specific influence, classifying
the types of public groups and pages that receive the most interactions on the platform
when sharing top-trending links pertaining to specific topics. We draw on a unique dataset
from CrowdTangle, a social monitoring tool owned by Facebook, which tracks interactions
with all posts made by public groups and pages on the platform. By sampling over a six-
month period in 2021 across four countries (Brazil, India, the United Kingdom, and the
United States), we assess and compare what kinds of sources (e.g., news media, politicians,
or other kinds of influential accounts and groups) are among the most influential accounts
in each location when it comes to three specific subjects: COVID-19, political leaders in
each country, and climate change—which we also compare to general queries that do not
specify a subject domain.
Although we do not make causal claims about how influential these interactions
may be on audiences, nor assess relative differences in levels of influence between the
accounts we examine, we use CrowdTangle interaction data to identify and then provide
valuable descriptive analysis of the composition of influential accounts in each country.
Here we define influential accounts as those whose public posts containing links to external
content received the most engagement on the platform. The analysis that follows focuses
on assessing how prominent legacy news media may be relative to other types of accounts.
JQD: DM 3(2023) Domain-specific influence on Facebook
5
Therefore, in examining domain-specific influence on Facebook, we pose two research
questions. First, we ask: (RQ1) What proportion of influential social media accounts on
Facebook are associated with news media compared to other kinds of influential accounts
(e.g., accounts aligned with political groups or figures versus other kinds of non-political
or non-media accounts)? We ask this with respect to (RQ1a) different topics and (RQ1b)
different countries. Second, we ask: (RQ2) Among the influential social media accounts
on Facebook associated with news media, what proportion are associated with legacy
versus non-legacy news media? We also ask this with respect to (RQ2a) different topics
and (RQ2b) different countries.
Our findings show that the types of influential accounts on Facebook vary
considerably by subject domain and country. News media accounts are, in fact, among the
largest share of these influential accounts in each country, but not necessarily the types of
news media organizations presumed to be most influential offline. We also show how the
US may be highly unusual compared to other countries. In Brazil, for example, political
groups and accounts tied to Jair Bolsonaro constitute a particularly large share of these
influential accounts, more so than the other three countries. Likewise, whereas individual
(and often conservative) media personalities and digital-born outlets are especially
prominent in the US (as previously reported in the press by Roose (2021)]), legacy news
organizations (especially the BBC and the Daily Mail) tend to be particularly prominent
among influential accounts in the UK. We also find exceptionally high levels of
engagement with content shared by UNICEF in the US, a pattern that does not appear in
other countries. Some of these differences may be related to Facebook’s publicly
announced efforts to combat COVID-19 misinformation (Schechner et al., 2021),
underscoring the role played by the platform’s algorithm in shaping what domain-specific
influence looks like in practice. Others may at least in part be artifacts of how the
CrowdTangle data (and the underlying Facebook data) is structured—whereby
organizations with an international audience may list their headquarters in the US—
demonstrating how decisions made by platform companies can limit what researchers can
learn from the data they make available.
Mont’Alverne et al. Journal of Quantitative Description: Digital Media 3(2023)
6
Data and Methods
This study describes domain-specific influential accounts on Facebook in four
countries: Brazil, India, the UK, and the US. We selected these four places, which represent
a significant proportion of the world’s population, because they capture significant
variation in terms of both their political and media systems. Brazil and the US have
presidential systems—the first with a fragmented multiparty parliament and the second
characterized by a bipartisan divide—while India and the UK have parliamentary systems.
When it comes to their media systems, the UK and the US are usually described as liberal
(Hallin & Mancini, 2004) or hybrid (Humprecht et al., 2022), with medium levels of media
market inclusion, journalistic professionalism, political parallelism, and state support
(although the UK differs from the US in its strong state support for public service media).
Meanwhile, Brazil shares commonalities with the Polarized Pluralist model (Hallin &
Papathanassopoulos, 2002), characterized by high levels of political parallelism and lower
levels of market inclusion, professionalism, and state support. India’s media system has
not been classified but it includes extensive commercial offerings on television and in print
in varying languages and reporting styles. All four countries exhibit high Facebook usage
overall but with considerable variation in the proportion who use the platform to get news.
According to data from the Reuters Institute’s Trust in News Project (Mont’Alverne et al.,
2022), 52% of Brazilians, 39% of Indians, 67% of Britons, and 73% of Americans say they
used Facebook for any purpose in the previous 30 days, but as a source for news daily,
rates are higher in Brazil (34%) and India (33%) than in the UK (27%) or US (30%).
To identify influential accounts in each country, which includes both pages and
public Facebook groups, our data collection focused on six randomly selected weeks during
the first six months of 2021. For each week and for each country, we made a series of
queries using CrowdTangle. We began by identifying the top 10 trending posts each week
JQD: DM 3(2023) Domain-specific influence on Facebook
7
containing links from pages with admins based in each of the four countries.
1
We did this
for content overall (without any keyword applied) as well as for three specific sets of
keywords: (1) “covid” or “coronavirus”; (2) “Jair Bolsonaro” or “Bolsonaro” (Brazil),
“Narendra Modi” or “Modi” (India), “Boris Johnson” (UK), “Joe Biden” or “Biden” (US);
and (3) “climate change” or “global warming” (“mudança climática” or “aquecimento
global” in Portuguese). In India, we analyzed only posts in English. For each of the top-
trending stories (links) identified in this first step, we next used the CrowdTangle API to
collect lists of all accounts that posted these URLs. To limit our attention to the most
influential of these accounts, we selected no more than five accounts per link and excluded
accounts whose posts containing these links had fewer than 100 total interactions. This
process generated a list of 840 unique accounts, which posted 1,628 times and generated
75,972,324 total interactions.
2
Some of these accounts will have significant numbers of
interactions from outside the country we are looking at, and for a few sites, like UNICEF,
this number may be so large that it limits our ability to assess how important they really
are in any individual country—but the way Facebook structures the data social scientists
can access through CrowdTangle provides no way of parsing this out.
Although we applied the same thresholds as cutoffs across each search query, there
is significant variation in the average number of total interactions these top-trending posts
received when examining domain-specific shared links (by topic and by country) versus
top-trending posts overall where no keyword was used. Top-trending posts for climate
change-related content received the lowest number of total interactions relative to the other
topics in all countries. This likely reflects how rare such posts are, especially in some
places, compared to the other topics examined.
3
1
We further restricted our lists of top-trending stories to posts that were shared in Portuguese (for
Brazil) or English (elsewhere).
2
Data was collected in September 2021. Although limiting our focus to only posts that contained
links (URLs) may have some impact on the kinds of influential accounts we identified, doing so
was necessary in order to identify all the accounts sharing the same top-trending external content
during a given week.
3
Our study is unable to differentiate between coordinated and organic activity on Facebook. In
2021, Meta published a report detailing networks that were removed from the platform due to
Mont’Alverne et al. Journal of Quantitative Description: Digital Media 3(2023)
8
Once we identified populations of domain-specific influential accounts, we next set
out to classify these accounts. Although our approach in this study draws on a
systematically collected sample of 840 accounts, the labor-intensive approach employed
for categorizing these accounts required keeping this sample at a modest scale. We rejected
alternative approaches to categorizing larger numbers of accounts given the dearth of prior
research on this subject and uncertainty about the relevant categories we might find when
examining influential accounts across the four countries. Although the sample size means
we are limited in our ability to make reliable inferences about differences between some of
the more granular subcategories, we also note that the number of accounts in many of these
categories remained roughly consistent across the six-week duration of the study. That is,
when we examine our findings at the weekly level as opposed to across the full six-week
span of our corpus, our results do not appear to be driven by any unusual activity specific
to one of the randomly selected weeks in which our CrowdTangle queries were conducted
(see the appendix for more detail). Furthermore, despite the size of the sample, the study’s
main strength comes from its comparative focus, which improves upon the limitations of
single-country studies that often raise questions about generalizability.
Two members of the research team coded accounts based on a modified version of
a categorization scheme used in Gallagher and colleagues’ (2021) study of social media
influence and COVID-19. The coders discussed the categorization using an iterative
process to reach a consensus when there was disagreement or uncertainty about how to
code an account. After a pilot stage, in which coders sought to apply the previously used
categories and revised the scheme to suit the present study, discussing discrepancies
between the coders with respect to specific accounts, the entire sample was divided
coordinated inauthentic behavior, but they do not include the countries studied here. In 2020, the
platform removed a network of social media accounts connected with Bolsonaro’s family, which
was accused of spreading misinformation and divisive political content in Brazil. See more:
https://about.fb.com/news/2021/12/metas-adversarial-threat-report/ and
https://www.reuters.com/article/ctech-us-facebook-disinformation-brazil-idCAKBN2492Y5-
OCATC.
JQD: DM 3(2023) Domain-specific influence on Facebook
9
between the two coders, who each independently coded subsets of the accounts. In
individual cases where uncertainties were raised, the coders and other members of the
research team reviewed each other’s work and came to a mutually agreed upon coding. We
opted for this approach given differences in language and domain-specific expertise.
Because many categories contained only a small number of accounts, we focus much of
our analysis on three broader categories that capture differences relevant to our research
questions: media accounts (including legacy media and non-legacy media accounts
4
),
political accounts, and other accounts. These and the additional subcategories coded are
summarized in Table 1.
5
4
Within the category of non-legacy media accounts, we classified brands as “digital news media”
even if part of larger conglomerates that owned legacy news organizations. In Brazil, for
example, web portal UOL was coded as “digital news media” even though its parent company
Grupo Folha owns both UOL and Folha de Sao Paulo, a legacy newspaper.
5
To see all accounts coded in each category, click here.
Journal of Quantitative Description: Digital Media 3(2023), 1–34 10.51685/jqd.2023.014
Copyright © 2023 (Camila Mont’Alverne, Amy Ross Arguedas, Sumitra Badrinathan, Benjamin Toff, Richard Fletcher, Rasmus Kleis
Nielsen). Licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International Public License.
Available at: http://journalqd.org
Table 1 – Categories of influential accounts.
Category
Subcategory
Description
Examples of
accounts
Media
accounts
Legacy News Media
Legacy media
News organizations with
print or television
properties
Globo, The Hindu,
BBC News, ABC
Other News Media
Media personalities
Journalists and other
media figures like
political commentators
and television hosts
José Tolentino, Sakshi
Joshi, Peter
Stefanovic, Becky
Hillier
Digital news media
Internet-based brands
UOL, Indiatimes,
LADBible, Daily
Wire
Political
accounts
Political Figures
Elected officials
Members of the
Parliament, mayors, state
governors, president,
prime minister
Marco Feliciano,
Achyuta Samanta,
Jeremy Corbyn, Ted
Cruz
Public servants
Employees of public
office, ministers,
secretaries.
Major Palumbo,
Piyush Goyal, Sid
Miller
Political candidates
Politicians who present
themselves as candidates
Lula, Dr. Shama
Mohamed, Nigel
Farage, Keith Kuder
JQD: DM 3(2023) Domain-specific influence on Facebook
11
Political figures
(other)
Political figures that do
not fit the other
categories
Ryan Fournier
Political Groups
Political
organizations and
movements
Advocacy organizations,
grassroots movements
Movimento LGBT
(LGBT Movement),
BJP West Bengal,
Leave.EU, Alexandria
Ocasio-Cortez
Progressives
Other political groups
Facebook groups whose
main theme is politics
Bolsonaro 2022,
NARENDRA MODI
ERA, Back Boris,
Trump Keep America
Great 2020
Other
accounts
Other Influential
Figures
Non-political public
figures
Athletes, entertainers,
public intellectuals,
religious leaders
Paulo Gustavo, Indian
Cricket Team,
Manchester United,
Lisa Daggs
Health experts
People that hold
positions in public health
institutions and public
health entities, medical
professionals,
epidemiologists
PAHO
NGOs
NGOSs
NGOs and charity
institutions
UNICEF,
UniteWomen.org
Other
Animals
Pages about animals
Eu amo os cachorros
(I love dogs), Justice
For Cecil The Lion,
Cat Lovers Only
Mont’Alverne et al. Journal of Quantitative Description: Digital Media 3(2023)
12
Non-political group
Facebook groups that do
not have politics as the
main theme
Reclame Aqui SJC
(Complain Here SJC),
Falkland Islands -
News & History,
Open Water
Swimming UK
Non-political shared
interests
Pages and groups that
share a specific interest
on a non-political topic
Acervo do
Conhecimento
Histórico (Historical
Knowledge Archive),
Star Wars Empire,
Catholic
Other
Accounts that do not fit
any of the other
categories
Magazine Luiza,
Restaurant Worker
News, Big Think
JQD: DM 3(2023) Domain-specific influence on Facebook
32
In the analysis that follows, we calculate the proportion of accounts in each of the
relevant categories and calculate basic inferential statistics (t-tests) to assess where
differences across subjects and countries are statistically significant.
Results
Examining overall differences in types of influential accounts
We begin by presenting results summarizing the proportion of influential accounts
in each major category in each country when we examine accounts identified without any
keyword applied (Figure 1). By first focusing on what percentage of influential accounts
are associated with media organizations or figures versus political organizations or figures
versus other areas when no subject domain has been applied, we establish a baseline for
comparisons with the three specific keywords that we focus on in the next part of our
findings where we more specifically take up RQ1a and RQ1b.
Figure 1 – Types of influential accounts identified with no keyword by country.
This first set of descriptive results clearly shows that accounts associated with news
media (broadly defined) account for a majority of the influential accounts identified in
Brazil and the UK and a minority (though still sizeable) in India and the US. The accounts
that appear most frequently in posts with no keyword in each country are: Caras Brasil
(Brazil), Indiatimes (India), Daily Mail (UK), and UNICEF (US). With the exception of
Mont’Alverne et al. Journal of Quantitative Description: Digital Media 3(2023)
14
UNICEF, all of these accounts belong to news organizations (and see above on the
difficulties of interpreting the numbers for UNICEF specifically). Moreover, political
accounts are consistently a relatively small share of influential accounts in all four
countries, constituting approximately 1 in 10 of these accounts. Non-political and
entertainment-related accounts (aggregated in the “other” category) also represent a
significant percentage of influential accounts when querying the CrowdTangle database
with no keywords applied. Examples of these accounts are: Eu amo os cachorros (I love
dogs), Star Wars Empire, or Cat Lovers Only.
Next we consider additional differences across subject domains (RQ1a) and
countries (RQ1b) by examining the share of domain-specific influential accounts and
compare these results to the “no keyword” condition. We plot these percentages in Figure
2.
We draw attention to two findings here. First, we note that political accounts
constitute a much larger share of influential accounts for top-trending links pertaining to
the political leaders in each country, with a statistically significant difference in the
proportion of political accounts in these posts when compared to other domains (p < 0.05)
(We summarize these t-tests in Appendix B.) In the aggregate, political accounts account
for 54% of posts about political leaders, 19% of those about COVID-19, and 23% about
climate change.
The prominence of political accounts also varies according to the country. In Brazil,
for example, two-thirds of the influential accounts identified when examining top-trending
stories that reference Bolsonaro were associated with political accounts (67%). Likewise,
in the UK, the majority of influential accounts identified for stories about Johnson were
associated with political accounts (58%), and the differences in both countries are
significant when compared to all other topics. Most of these accounts are coded under the
subcategory of political “groups” such as Fechado com Bolsonaro 2022 (Support for
Bolsonaro 2022) in Brazil (59%) and Rejoining the EU is best for Britain in the UK (56%),
JQD: DM 3(2023) Domain-specific influence on Facebook
15
rather than accounts belonging to individual political figures (8% in Brazil and 6% in the
UK). (For a breakdown of political accounts per country and topic, see appendix C.)
Second, we find that media-related accounts are a relatively larger percentage of
influential accounts for political leader stories in India (56%) and the United States (53%)
compared to the no keyword condition. In posts about political leaders, the most frequent
accounts identified in each country are Quebrando o Tabu (Breaking taboos) (Brazil),
Republic (India), Leave.EU (UK), and Ben Shapiro (US)—a mix of news organizations,
media personalities, and political groups. Media accounts are also the largest category for
top-trending stories about COVID-19 and climate change in all four countries with one
exception—COVID-19 stories in the US, where UNICEF, an NGO coded in the “other”
category, is the account that appears most frequently. Elsewhere media accounts loom
large. The accounts that appear most frequently in posts about COVID-19 in the other
countries include G1 and UOL (Brazil), The Times of India (India), and ITV News (UK).
When it comes to climate change posts, these accounts are: Economia Ecológica
(Ecological Economy) (BR), We Don’t Deserve This Planet (IN), World Economic Forum
(UK), and media personality Dan Bongino (US), demonstrating a broader mix of different
types of influential accounts.
Mont’Alverne et al. Journal of Quantitative Description: Digital Media 3(2023)
16
Figure 2 – Types of domain-specific influential accounts identified by country.
Examining differences in types of influential news media accounts
While the importance of country-level differences is apparent when it comes to
whether news media organizations are among the most influential accounts, when we drill
JQD: DM 3(2023) Domain-specific influence on Facebook
17
down and examine different types of media accounts—specifically whether or not they are
associated with legacy brands—we see further evidence of how influence often plays out
in distinct domain-specific and country-specific ways. To examine RQ2, we focus on the
set of influential accounts identified as being associated with news media and calculate the
proportion within this category that are associated with legacy news media organizations
versus other kinds of media accounts, which include both media personalities and digital-
born organizations. In Figures 3, 4, and 5 we summarize differences in the proportion of
influential accounts that are associated with legacy media in each country for each topic
area, comparing these percentages in relation to baseline results where no keyword was
applied. In Figures 6, 7, and 8 we do the same for other news media.
Figure 3 – Proportion of influential media accounts associated with legacy news
organizations by country (for all posts versus posts that reference political leaders).
Note: Percentages shown are as share of media-related influential accounts identified.
Figure 4 – Proportion of influential media accounts associated with legacy news
organizations by country (for all posts versus posts that reference COVID-19).
Note: Percentages shown are as share of media-related influential accounts identified.
Mont’Alverne et al. Journal of Quantitative Description: Digital Media 3(2023)
18
Figure 5 – Proportion of influential media accounts associated with legacy news
organizations by country (for all posts versus posts that reference climate change).
Note: Percentages shown are as share of media-related influential accounts identified.
Several trends are apparent here. First, relative to searches without any keyword
and for all three topics, legacy media are a significantly larger proportion of domain-
specific influential accounts in India and the UK, constituting a majority of media-related
influential accounts for two of the three topics in India and roughly three-quarters of all
media-related influential accounts in the UK. In part this captures the importance of the
BBC in the UK’s media landscape; however, other legacy media organizations such as the
Daily Mail also appear frequently across these topics. In Brazil, 6-in-10 of the influential
media accounts identified were associated with legacy news organizations in the no
keyword condition; however, this share was significantly smaller for stories referencing
Bolsonaro (22%) or COVID-19 (29%), as other news media accounts, including those
associated with media personalities and digital born organizations such as UOL,
constituted a larger share of influential accounts. In the US, on the other hand, legacy news
media were a smaller share of the influential media accounts identified, and this is true for
both the baseline condition and the topic-specific queries – the difference is non-significant
only for posts about COVID-19. This reflects the prominence of individual news media
personalities such as Dan Bongino and digital-born organizations, who frequently top the
lists of influential accounts across these topic domains.
Figures 6, 7, and 8 summarize differences in the proportion of influential accounts
that are associated with other news media in each country for each topic area, comparing
JQD: DM 3(2023) Domain-specific influence on Facebook
19
these percentages in relation to baseline results where no keyword was applied. In Brazil
and the US, posts from other news media represent the majority of media accounts in posts
about the presidents of these countries (78% and 94%, respectively) and those about
COVID-19 (71% and 78%). This contrasts with India and especially the UK, where other
news media account for a smaller share of posts about these topics (24% of posts from
media accounts about Johnson, 43% about Modi, 19% about COVID-19 in the UK and
48% about COVID-19 in India) but are responsible for a significant percentage of posts
from media accounts about climate change (60% in the UK and 76% in India).
Figure 6 – Proportion of influential media accounts associated with other news
organizations by country (for all posts versus in posts that reference political
leaders).
Note: Percentages shown are as share of media-related influential accounts identified.
Figure 7 – Proportion of influential media accounts associated with other news
organizations by country (for all posts versus posts that reference COVID-19).
Note: Percentages shown are as share of media-related influential accounts identified.
Mont’Alverne et al. Journal of Quantitative Description: Digital Media 3(2023)
20
Figure 8 – Proportion of influential media accounts associated with other news
organizations by country (for all posts versus in posts that reference climate
change).
Note: Percentages shown are as share of media-related influential accounts identified.
Discussion
In response to our research questions, our analysis of influential accounts on
Facebook in Brazil, India, the UK, and the US demonstrates the continued centrality of
accounts belonging to news media organizations across all countries, but in some places
more so than others and generally for specific topics rather than for top-trending links
overall. In other words, even if most people consume little news in general online (Guess,
2021; Wojcieszak et al., 2021), news media can still be central to understanding
information flows on platforms. Although we did not formally code accounts for ideology,
our results suggest that the prominence of highly partisan brands and accounts on
Facebook, as found by Hiaeshutter-Rice and Weeks (2021) in the US, does not necessarily
generalize to other countries, at least not those we examined, as established, non-partisan
brands such as the BBC in the UK, UOL in Brazil, or The Times of India in India, were
found to be particularly prominent in these markets alongside entertainment-related and
tabloid websites.
These results build on and help contextualize previous findings that have also
shown the continued importance of legacy news when it comes to specific topics on Twitter
(e.g., Gallagher et al., 2021; Majó-Vázquez et al., 2017), demonstrating similar dynamics
JQD: DM 3(2023) Domain-specific influence on Facebook
21
extend to Facebook—a platform that is much more widely used, even as much Facebook
news use is incidental. At the same time, the considerable variation we found by country
and by topic underscores the importance of domain specificity when thinking about how
to define the nature of social media influence and the role of news organizations as a force
on platforms for driving the public agenda (Heiberger et al., 2021). Furthermore, when we
differentiate legacy media from other kinds of media pages and accounts, the former
constitute a majority of influential media accounts primarily in the UK and India—at least
for news-related subjects—while the latter are more prominent in the US for all subjects.
These results corroborate the centrality of legacy news organizations—especially the
BBC—for how people in the UK consume news online found by tracking studies (Fletcher
et al., 2021) while broadening our understanding of how these dynamics may differ in other
country contexts.
Although much of our analysis focuses on media-related influential accounts, it is
worth noting the importance of other kinds of influential accounts identified as part of this
study, including political groups and pages who constitute a particularly large share of the
most influential accounts in all four countries for top-trending links referencing political
leaders. In Brazil and the UK in particular, political groups are the most prominent type of
account for top-trending posts about Bolsonaro in Brazil and Johnson in the UK. These
results may well reflect the fact that many of the reasons people say they use social media
have little to do with news (Mont’Alverne et al., 2022), and when it comes to political
content on platforms, many accounts that operate on them and may be influential on them
have objectives in mind such as political mobilization rather than journalism. While these
other uses may nonetheless be intertwined with exposure to information (Lewis, 2020), our
findings raise questions about the consequences of Meta’s decision to downgrade news in
users’ Facebook feeds.
6
Such actions could in effect give a relative boost to these other
political groups and pages.
6
https://www.wsj.com/articles/facebook-shifts-resources-from-news-tab-and-bulletin-to-focus-
on-creator-economy-11658250433
Mont’Alverne et al. Journal of Quantitative Description: Digital Media 3(2023)
22
Our findings with respect to the composition of influential accounts in Brazil is
particularly noteworthy, where we see a large percentage of legacy news media accounts
when examining top-trending links overall, but relatively smaller percentages when it
comes to stories about political leaders, where accounts associated with political groups
are far more common. These results are in line with those documented by Santos Júnior
(2023; 2019), suggesting that traditional brands may be becoming less dominant on digital
platforms over time when compared to alternative sources and politicians, at least unless
they in their rankings prioritize one or more signals that clearly benefits long-standing
brands (as many newer voices feel that Google Search does through PageRank, for
example). They also are consistent with results looking specifically at information about
Bolsonaro in Brazil found on other platforms. In WhatsApp groups that support Bolsonaro,
partisan personalities (or “political influencers”) were previously identified as a
particularly important source for shared content (Santos et al., 2022). Likewise, on Twitter,
the former president’s supporters were shown to be less likely to refer to legacy media
organizations as well (Santos, 2021). While the individual influential political groups we
identified sharing links referencing Bolsonaro on Facebook included both supporters and
opponents, our findings point to the need for more research that seeks to better understand
why such groups and figures are as prominent as they appear to be relative to news media—
whether due to deliberate political strategies, reasons involving the political information
environment, factors involving the platforms themselves, or a combination of these forces.
When it comes to differences across topics, the findings showing the importance of
media-related accounts (and political accounts to some extent) on posts about COVID-19
aligns with Gallagher and colleagues’ (2021) previous findings on the amplification of elite
posting about the pandemic in the US. However, our results for the US also differ
considerably from the other countries, since we see other kinds of accounts, namely
UNICEF, constituting a particularly large share of the most influential accounts (though a
significant share of interactions with this page are likely to be international, complicating
interpretations of the result). These cross-country differences underscore the problem with
relying on (and extrapolating from) US data only when studying social media influence.
JQD: DM 3(2023) Domain-specific influence on Facebook
23
While CrowdTangle is one of the few options available to study what widely
circulates on Facebook, several caveats are important to acknowledge in relation to the data
accessed through this tool. First, CrowdTangle only provides interaction metrics for public
pages and groups, not which stories are most widely seen in users’ feeds, which inevitably
shapes the populations of influential accounts on which we have based our inferences. We
also do not know to what extent is a proxy for reach or influence. Second, the reasons why
CrowdTangle interaction metrics vary the way they do are often opaque—that is, we cannot
tell what may be due to differences in the content posted, algorithmic ranking or “shadow
banning,” numbers of followers each page or group has previously amassed and how, or
other factors involving metrics not provided by CrowdTangle (such as click-through rates
or time spent dwelling on posts versus scrolling). Third, and related, our data are
constrained by what Meta chooses to make available, which cannot be independently
audited, to whom they choose to grant access, as well as how they choose to structure their
data, as illustrated by the inability to differentiate from which countries interactions with
global pages such as UNICEF may be coming from. Ultimately, relying on tools owned by
platforms means that studies are subject to such decisions—a particularly salient point
when it comes to CrowdTangle, which Meta has reportedly considered shutting down
entirely.
7
All of these considerations underscore the real vulnerabilities of doing research
in the current platform environment, and there are few reasons to believe this will improve
as people increasingly use more closed networks like TikTok.
We have also gathered only modestly-sized samples for just three topics over a six-
month period rather than examining a wider array of subjects or time periods. Specifically
in the case of India, our sample is also limited to English-speaking accounts, which only
covers a fraction of what is being posted in the country. Considering the degree to which
influence appears to be so dependent on context, it is plausible that different choices around
dates, topics or countries might produce different results, making it difficult to generalize
7
https://www.theverge.com/2022/6/23/23180357/meta-crowdtangle-shut-down-facebook-
misinformation-viral-news-tracker
Mont’Alverne et al. Journal of Quantitative Description: Digital Media 3(2023)
24
from these specific findings—although, as we noted, there is some stability in the patterns
we found when disaggregating our data week by week. Our focus on political leaders may
also limit how well the accounts in the sample represent the broader political discussion in
these countries—although it is worth pointing out that all of them recently had (or still
have) populists in power, and these figures tend to be at the center of the political debates
even for topics that would not typically be divided along partisan lines, which we believe
provides important insight into how public political discussion on Facebook works in these
countries. Our coding scheme is also limited to specific kinds of influencers and could be
further refined or adapted for future studies. We also note that although our analysis points
to differences in the kinds of accounts that received the most interactions when posting
top-trending links on Facebook, we are not able to assess how influential these accounts
may be or why some accounts received more attention on the platform than other kinds of
accounts. Higher rates of interactions may be due to how Facebook’s algorithm surfaces
stories or privileges some sources over others, or they might reflect different levels of
interest from those encountering the news/information on platforms. Differences may also
be due to strategic factors made by organizations themselves in how they post on the
platform, sometimes using coordinated activity to boost their own engagement numbers
(Giglietto et al., 2020). We are unable to evaluate the reasons behind the patterns we
identify in the types of influential accounts we found.
Despite these limitations, this empirical study provides a much-needed systematic
comparative assessment of the types of accounts that receive the most interactions with
public content on Facebook, clarifying not only the continued importance of news media
in these spaces but also of non-legacy organizations and media personalities in some media
systems. As Facebook remains one of the most prominent platforms for information
discovery in many countries, understanding what actors play influential roles in curating
information there is increasingly important when considering how platforms affect
information environments more broadly. Considering how opaque distribution of content
on social media is, we believe this study contributes to our understanding of the information
environments in these countries, adding nuance to what we know about influential accounts
JQD: DM 3(2023) Domain-specific influence on Facebook
25
on Facebook and also underscoring the importance of examining that influence in a
domain-specific manner and across countries.
Funding
This research was funded by the Meta Journalism Project (CTR00730).
References
Allen, J., Howland, B., Mobius, M., Rothschild, D., & Watts, D. J. (2020). Evaluating the
fake news problem at the scale of the information ecosystem. Science Advances,
6(14), 1–7.
Altay, S., Nielsen, R. K., & Fletcher, R. (2022). Quantifying the “infodemic”: People
turned to trustworthy news outlets during the 2020 coronavirus pandemic. Journal
of Quantitative Description: Digital Media, 2.
Andı, S. (2021). How and Why do Consumers Access News on Social Media? In N.
Newman, R. Fletcher, A. Schulz, S. Andı, C. T. Robertson, & R. K. Nielsen (Eds.),
Digital News Report (pp. 52–57). Reuters Institute for the Study of Journalism.
Beam, M. A., Hutchens, M. J., & Hmielowski, J. D. (2018). Facebook news and
(de)polarization: reinforcing spirals in the 2016 US election. Information
Communication and Society, 21(7), 940–958.
Boczkowski, P. J., Mitchelstein, E., & Matassi, M. (2018). “News comes across when
I’m in a moment of leisure”: Understanding the practices of incidental news
consumption on social media. New Media and Society, 20(10), 3523–3539.
Chauchard, S., & Garimella, K. (2022). What Circulates on Partisan WhatsApp in India?
Insights from an Unusual Dataset. Journal of Quantitative Description: Digital
Media, 2, 1–42.
Mont’Alverne et al. Journal of Quantitative Description: Digital Media 3(2023)
26
Facebook Transparency Center. (2021). Widely Viewed Content Report: What People See
on Facebook. https://transparency.fb.com/en-gb/data/widely-viewed-content-report/
Fletcher, R., Kalogeropoulos, A., & Nielsen, R. K. (2021). More diverse, more politically
varied: How social media, search engines and aggregators shape news repertoires in
the United Kingdom. New Media and Society.
Fletcher, R., & Nielsen, R. K. (2018). Are people incidentally exposed to news on social
media? A comparative analysis. New Media and Society, 20(7), 2450–2468.
Freelon, D., Lopez, L., Clark, M. D., & Jackson, S. J. (2018). How Black Twitter and
Other Social Media Communities Interact With Mainstream News.
https://repository.upenn.edu/asc_papers
Gallagher, R. J., Doroshenko, L., Shugars, S., Lazer, D., & Welles, B. F. (2021).
Sustained Online Amplification of COVID-19 Elites in the United States. Social
Media + Society, 1–16.
Giglietto, F., Farci, M., Marino, G., Mottola, S., Radicioni, T., & Terenzi, M. (2022).
Mapping nefarious social media actors to speed-up Covid-19 fact-checking.
Guess, A. M. (2021). (Almost) Everything in Moderation: New Evidence on Americans’
Online Media Diets. American Journal of Political Science, 65(4), 1007–1022.
Hagey, J., & Horwitz, J. (2021, September 15). Facebook Tried to Make Its Platform a
Healthier Place. It Got Angrier Instead. The Wall Street Journal.
Hallin, D. C., & Mancini, P. (2004). Comparing Media Systems: Three Models of Media
and Politics. Cambridge University Press.
Hallin, D. C., & Papathanassopoulos, S. (2002). Political clientelism and the media:
southern Europe and Latin America in comparative perspective. Media, Culture &
Society, 24(2), 175–195.
JQD: DM 3(2023) Domain-specific influence on Facebook
27
Heiberger, R., Majó-Vázquez, S., Castro, L., Nielsen, R. K., & Esser, F. (2021). Don’t
blame the media! The role of politicians and parties in fragmenting online political
debate. International Journal of Press/Politics, 32.
Hiaeshutter-Rice, D., & Weeks, B. (2021). Understanding Audience Engagement with
Mainstream and Alternative News Posts on Facebook. Digital Journalism, 9(5),
519–548.
Humprecht, E., Castro Herrero, L., Blassnig, S., Brüggemann, M., & Engesser, S. (2022).
Media Systems in the Digital Age: An Empirical Comparison of 30 Countries.
Journal of Communication, 00, 1–20.
Kalogeropoulos, A., Suiter, J., Udris, L., & Eisenegger, M. (2019). News Media Trust
and News Consumption: Factors Related to Trust in News in 35 Countries.
International Journal of Communication, 13, 3672–3693.
Kim, Y. M. (2009). Issue publics in the new information environment: Selectivity,
domain specificity, and extremity. Communication Research, 36(2), 254–284.
Konitzer, T., Allen, J., Eckman, S., Howland, B., Mobius, M., Rothschild, D., & Watts,
D. J. (2021). Comparing Estimates of News Consumption from Survey and
Passively Collected Behavioral Data. Public Opinion Quarterly, 85(1), 347–370.
Lewis, R. (2020). “This Is What the News Won’t Show You”: YouTube Creators and the
Reactionary Politics of Micro-celebrity. Television and New Media, 21(2), 201–217.
Majó-Vázquez, S., Zhao, J., & Nielsen, R. K. (2017). The Digital-Born and Legacy News
Media on Twitter during the French Presidential Elections (Issue June).
http://reutersinstitute.politics.ox.ac.uk/sites/default/files/2017-11/Digital-Born and
Legacy News Media UK Factsheet.pdf
Mont’Alverne et al. Journal of Quantitative Description: Digital Media 3(2023)
28
Mont’Alverne, C., Badrinathan, S., Arguedas, A. R., Toff, B., Fletcher, R., & Nielsen, R.
K. (2022). The Trust Gap: How and Why News on Digital Platforms Is Viewed
More Sceptically Versus News in General.
Newton, C. (2020, July 22). Why no one knows which stories are the most popular on
Facebook. The Verge.
https://www.theverge.com/interface/2020/7/22/21332774/facebook-crowdtangle-
kevin-roose-nyt-tweets-interactions-reach-engagement
Nicholls, T., Shabbir, N., & Nielsen, R. K. (2016). Digital-born news media in Europe.
Park, S., Lee, J., Ryu, S., & Hahn, K. S. (2015). The Network of Celebrity Politics:
Political Implications of Celebrity Following on Twitter. Annals of the American
Academy of Political and Social Science, 659(1), 246–258.
Pew Research Center, P. (2021). News Use Across Social Media Platforms in 2020 (Vol.
12).
Roose, K. (2021, July 14). Inside Facebook’s Data Wars. The New York Times.
Santos Junior, M. A. dos. (2023). As Flutuações de Longo Prazo da Polarização no Brasil
– Análise do Compartilhamento de Informações Políticas Entre 2011 e 2019. Dados,
66(2), 1–45.
Santos Júnior, M. A. (2019). Desarranjo da visibilidade, desordem informacional e
polarização na política brasileira entre 2013 e 2018. Universidade Federal
Fluminense.
Santos, N. (2021). Fontes de informação nas redes pró e contra o discurso de Bolsonaro
sobre o Coronavírus. E-Compós, 24.
Santos, N., Chagas, V., & Marinho, J. (2022). De onde vem a informação que circula em
grupos bolsonaristas no WhatsApp. Intexto, 123603.
Schechner, S., Horwitz, J., & Glazer, E. (2021, September 17). How Facebook Hobbled
Mark Zuckerberg’s Bid to Get America Vaccinated. The Wall Street Journal.
JQD: DM 3(2023) Domain-specific influence on Facebook
29
Stier, S., Mangold, F., Scharkow, M., & Breuer, J. (2022). Post Post-Broadcast
Democracy? News Exposure in the Age of Online Intermediaries. American
Political Science Review, 116(2), 768–774.
Thorson, K., & Wells, C. (2016). Curated Flows: A Framework for Mapping Media
Exposure in the Digital Age. Communication Theory, 26(3), 309–328.
Vicario, M. Del, Bessi, A., Zollo, F., Petroni, F., Scala, A., Caldarelli, G., Stanley, H. E.,
& Quattrociocchi, W. (2016). The spreading of misinformation online. Proceedings
of the National Academy of Sciences of the United States of America, 113(3), 554–
559.
Wojcieszak, M., de Leeuw, S., Menchen-Trevino, E., Lee, S., Huang-Isherwood, K. M.,
& Weeks, B. (2021). No Polarization From Partisan News: Over-Time Evidence
From Trace Data. The International Journal of Press/Politics.
Appendix
APPENDIX A. Variation of types of accounts per week
As we can see in Figure 1, the frequency of accounts (in absolute numbers) sharing
top-trending links does not seem to vary drastically depending on the week when they were
posted, indicating some stability in the results. However, there is no way of demonstrating
how representative they are in comparison to the full universe of Facebook accounts over
time.
Mont’Alverne et al. Journal of Quantitative Description: Digital Media 3(2023)
30
Figure A1 – Variations of types of accounts per week.
APPENDIX B. T-tests results
The results below show which significant differences exist between countries and
keywords. Frequencies of media accounts for most keywords across countries are
significantly different from the no keyword condition, while there is also significant
variation comparing keywords against each other. There are also significant country
differences when it comes to the frequency of media accounts per keyword, except in posts
about climate change.
JQD: DM 3(2023) Domain-specific influence on Facebook
32
Table B1. T-test results.
i.
Brazil
ii.
India
iii.
United Kingdom
iv.
United States
a. No keyword % Media
% Politics
% Other
51.3
8.9
39.9
b, c
b, c
c
ii, iv
–
ii, iv
29.0
9.3
61.7
b, c, d
c, d
b, c, d
i, iii
–
i, iii
61.2
4.1
33.7
c
b, c
c
ii, iv
iv
ii, iv
30.3
10.6
59.1
c, d
c, d
c, d
i, iii
i
i, iii
b. COVID % Media
% Politics
% Other
40.5
25.6
33.3
a, c
a, c, d
c
ii
iv
iv
59.5
14.3
26.2
a
c, d
a, c
i, iv
–
iv
51.2
17.1
31.7
c
a, c
c
iv
–
iv
30.5
11.0
58.5
c, d
c, d
c, d
ii, iii
i
i, ii, iii
c. Political leaders
% Media
% Politics
% Other
24.6
67.3
8.0
a, b, d
a, b, d
a, b, d
ii, iv
ii, iv
–
56.5
38.7
4.8
a
a, b
a, b
i, iii
i, iii
iv
29.6
57.8
12.7
a, b, d
a, b, d
a, b, d
ii, iv
ii, iv
–
52.8
31.5
15.7
a, b
a, b
a, b
ii, iii
i, iii
ii
d. Climate change
% Media
% Politics
% Other
46.3
12.2
41.5
c
b, c
c
–
ii, iv
ii, iv
48.3
36.6
15.0
a
a, c
a
–
i, iii
i, iv
59.7
9.7
30.6
c
c
c
–
ii, iv
–
47.3
30.4
22.3
a, b
a, b
a, b
–
i, iii
i, ii
Note. Statistically significant differences in percentages (p < 0.05) are denoted using a, b, c, d for comparisons
between keyword queries and using i, ii, iii, iv for comparisons between countries.
JQD: DM 3(2023) Domain-specific influence on Facebook
32
APPENDIX C. Proportion of influential political accounts associated with political
groups or political figures
When detailing the categories aggregated as political accounts, it becomes clear the
relevance of political groups in posts about political leaders and covid, especially in Brazil,
the UK, and the US. In India, political groups represent a majority of political accounts
posting about climate change only. In addition, political figures represent a significant
share of accounts posting about covid and political leaders in the country. In the US,
political figures account for the majority of political accounts posting about climate change
only.
JQD: DM 3(2023) Domain-specific influence on Facebook
33
Figure C1 – Proportion of influential political accounts associated with political
groups by topic area and country.
Note: Percentages shown are as share of political-related influential accounts identified.
Mont’Alverne et al. Journal of Quantitative Description: Digital Media 3(2023)
34
Figure C2 – Proportion of influential political accounts associated with political
figures by topic area and country.
Note: Percentages shown are as share of political-related influential accounts identified.