Content uploaded by Manuel Goyanes
Author content
All content in this area was uploaded by Manuel Goyanes on Jun 20, 2023
Content may be subject to copyright.
See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/371246734
Social media algorithmic versus professional journalists’ news selection:
Effects of gate keeping on traditional and social media news trust
ArticleinJournalism · June 2023
DOI: 10.1177/14648849231179804
CITATIONS
0
READS
19
3 authors:
Some of the authors of this publication are also working on these related projects:
Medios audiovisuales públicos ante el ecosistema de las plataformas: modelos de gestión y evaluación del valor público de referencia para España View project
Social Media & Personality View project
Rebecca Scheffauer
Universidad de Salamanca
5 PUBLICATIONS14 CITATIONS
SEE PROFILE
Manuel Goyanes
University Carlos III de Madrid
98 PUBLICATIONS976 CITATIONS
SEE PROFILE
Homero Gil de Zúñiga
Universidad de Salamanca
184 PUBLICATIONS11,884 CITATIONS
SEE PROFILE
All content following this page was uploaded by Rebecca Scheffauer on 14 June 2023.
The user has requested enhancement of the downloaded file.
Original Article
Journalism
2023, Vol. 0(0) 1–24
© The Author(s) 2023
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/14648849231179804
journals.sagepub.com/home/jou
Social media algorithmic versus
professional journalists’news
selection: Effects of gate
keeping on traditional and social
media news trust
Rebecca Scheffauer
Democracy Research Unit, Political Science, College of Law and Public Administration, University of Salamanca,
Salamanca, Spain
Manuel Goyanes
Democracy Research Unit, Political Science, College of Law and Public Administration, University of Salamanca,
Salamanca, Spain
Department of Communication, Carlos III University, Getafe, Spain
Homero Gil de Z ´
uñiga
Democracy Research Unit, Political Science, College of Law and Public Administration, University of Salamanca,
Salamanca, Spain
Media Effects Research Lab, Film Production & Media Studies Department, Donald P. Bellisario College of
Communications, Pennsylvania State University, State College, PA, USA
Facultad de Comunicación y Letras, Universidad Diego Portales, Santiago, Chile
Abstract
Research has shown positive attitudes toward journalists and their roles foster pro-
democratic outcomes. With the rise of social media as news sources, algorithms operate
as gatekeepers, which may alter linkages between public opinion, journalists, and media
trust. However, results from a panel-survey conducted in the U.S. underline citizens’
preference for journalist gatekeeping in fueling trust in traditional and social media news.
Conversely, preference for algorithmic news selection does not affect people’s levels of
trust. Furthermore, traditional news use moderates this relationship as those who report
higher traditional news use and a preference for professional news gatekeeping trust
Corresponding author:
Rebecca Scheffauer, Democracy Research Unit, Political Science, University of Salamanca, Casa del Bedel, c/
Benedicto XVI, 22Despacho 26, 37008 Salamanca, Spain.
Email: Rebecca.Scheffauer@gmx.at
traditional news the most. This study contributes to current discussions on the effects of
preference for journalists’or algorithmic news selection, arguing that evaluations of
journalists’editorial work remain critical to explain media trust.
Keywords
News selection, algorithms, gatekeeping, social media, news trust, journalism
Introduction
Gatekeeping theory emerged back in the 1940s and focused on how professional news
selection unfolds (Shoemaker and Riccio, 2016). Over time, it has evolved and become
more complex (Donohue et al., 1972) as it includes not only professional decisions for
news selection, but also the practices and processes associated with news production (e.g.
deciding how much space a news story is assigned or whether it returns over a period of
time) (Donohue et al., 1972). According to this theory, journalists perform a normative
role (Janowitz, 1975), take moral decisions to promote pro-democratic outcomes
(Heinderyckx and Vos, 2016), and protect their work from potentially biased influences
like advertisers or the government (Shoemaker et al., 2009). There is a voluminous
amount of research showing how journalists’news selection, the professional practices
associated, and people’s understanding of these practices play key roles in supporting a
working democracy and influencing citizens’levels of trust in media outlets (Nah and
Chung, 2012). While journalists still strive to uphold practices like transparency and
traditional gatekeeping in the editorial process, new media (e.g. social media) bring
challenges that have forced traditional media to adapt to digital transformation
(Bentivegna and Marchetti, 2018). The increasing influence of and almost instantaneous
feedback from the audience are among the major influences altering gatekeeping pro-
cesses (Malmelin and Villi, 2016).
In recent years, Artificial Intelligence (AI) has brought new challenges and oppor-
tunities for news distribution, especially on social media (Newman, 2011). Online, the
number of agents who can share and distribute information has risen significantly. These
new actors work with radically different selection standards (Meraz and Papacharissi,
2013). Algorithms distribute and filter news based on preprogrammed criteria, aimed at
selecting content that best fits readers’interests and prior behaviors (Thurman et al., 2019;
Wallace, 2018). In contrast to professional gatekeeping where editorial processes are
involved in selecting and creating content, algorithms choose from an already existing
corpus of news and distribute those to the users (Wallace, 2018). Whereas the selection
criteria applied by journalists are somewhat comprehensible for the audience (e.g. news
values), what drives algorithms if often less clear to the average user (Swart, 2021).
Algorithmic news selection has raised some concerns due to its potential effects in
nurturing filter bubbles and echo chambers (Nechushtai and Lewis, 2019;Thurman et al.,
2019). Some scholars suggest that the effects are smaller than previously assumed, while
others still advise caution in light of their findings (Flaxman et al., 2016;Sunstein, 2018).
2Journalism 0(0)
By guillotining opinion-challenging perspectives and fueling congruent news exposure,
research has suggested that algorithmic news selection may jeopardize the ethos of
journalism and its associated trust (Nechushtai and Lewis, 2019). Moreover, users of-
tentimes lack an understanding of how algorithms operate. This can be due to the opacity
of algorithms with businesses providing scarce information on how they work and what
data is used (Swart, 2021). Although algorithmic news selection might generally be more
associated with social media platforms, news organizations also make use of algorithms to
provide more tailored content to their readers. To fully personalize a reader’s experience,
news organizations use audience metrics to adapt to their audience’s interests and
preferences, thus ultimately influencing editorial decision-making (Harambam et al.,
2018).
Actual personalization of the newsfeed is, however, usually more rudimentary on the
websites of news organizations than it is for social media outlets or news aggregators.
As Harambam et al. (2018) highlight, oftentimes this is focused on a “recommended for
you”section and the ability to opt out of certain news (e.g. sports). This type of
personalization is usually referred to as explicit since the user has to consciously put
work into it (Bodó, 2019). Contrary to explicit personalization, implicit personalization
draws upon data collected from a user’s activities and infers preferences (Bodó, 2019).
This type of personalization includes metrics, such as clicks or transaction histories, and
data from third parties (Bodó, 2019). What is becoming increasingly important for news
outlets/organizations is the intersection of their content with news aggregators and
social media (Goyanes et al., 2023). A growing number of people encounter their news
on these platforms (Scheffauer et al., 2021;Thorson, 2020). News outlets and orga-
nizations have to partially rely on the algorithms provided by them to reach relevant
audiences. These platforms, however, also serve their own agendas, such as creating
advertising revenue and even restricting access to some content. Thus, while being an
important place to distribute their product, news media has to simultaneously compete
with these platforms, vying for audiences’time and attention. (Bodó, 2019).
How does citizens’preference for journalists’news selection permeate public opinions
towards trust in different media outlets (traditional and social media)? Are these editorial
preferences over trust contingent upon people’s levels of traditional news consumption?
And how does the preference for algorithmic news selection relate to people’s trust in
traditional and social media news?
Drawing upon two-wave survey data from the United States (U.S.), this study ex-
amines the role of preference for journalists’and algorithmic news selection in nurturing
or hindering people’s media trust. Our findings show that preference for journalists’news
selection is positively associated to trust in traditional and social media news. Preference
for algorithmic news selection does not appear to yield a significant impact on trust in
either type of news. Finally, our findings illustrate the contingent role of traditional news
consumption on the relationship between preference for journalists’news selection and
trust in traditional news. This study contributes to ongoing discussions on the effects of
preferring journalists’or algorithmic news selection, arguing that citizens’preference for
journalists’editorial work remains a significant factor in fueling trust in different media
outlets –online and offline.
Scheffauer et al. 3
Journalists’news selection and trust in traditional news
Trust in news has been a popular research subject among scholars for almost a century.
Over time, not only has the concept evolved, but also its measurement (Fisher, 2016). This
evolution has also brought about many different research foci ranging from general media
trust to trust in a specific outlet or trust in a piece of news itself (Fisher, 2016). Some
studies point out the distinction between trust reports on news media as an institution and
a profession. Whereas trust levels for news media have been declining for a while in the
U.S. (Abdenour et al., 2020), the profession is faring better in the public eye (McKewon,
2018). Newman et al. (2021) outline that the news trust gap between countries is
widening. While over the course of the COVID-19 pandemic part of the audience seems
to have regained confidence in news, in countries with previously low scores such as the
U.S. overall trust has decreased even further.
Considering types of media outlets, audiences still place rather high levels of trust in
traditional ones –a trend occurring in a number of Western countries (Dawson, 2017;
European Broadcasting Unit, 2017). Furthermore, the trust gap between “news in
general”and “found in aggregated environments”(Newman et al., 2021: p. 10), such as
on social media platforms, has become more pronounced. Audiences reward reliable news
sources as well as established brands with more trust (Newman et al., 2021). Lower levels
of trust in traditional news are usually found in geographies with strong partisan divides
such as Italy, Hungary, or the U.S. (Newman et al., 2017). In the case of the U.S., prior
scholarship has extensively documented that trust is largely influenced by ideology. In
political news, for instance, there is extensive evidence that the Republicans are known to
be more distrustful of traditional news, whereas the Democrats appear to be more fa-
vorable (Barthel and Mitchell, 2017;Newport, 2017). Those with lower trust in news,
furthermore, more readily turn to alternative sources (Newman et al., 2021).
Public opinion and audience research have shown a positive association between news
consumption and news trust (Mourão et al., 2018). Reduced trust levels also predict a less
knowledgeable society (Kaufhold et al., 2010). Thus, decreased news use and lower levels
of news trust can in turn be harmful for democracy.
When people express favorable attitudes toward professional journalists and their
editorial work, they are more likely to report trust toward the news, resulting in higher
levels of news consumption (Gil De Z ´
uñiga and Hinsley, 2013) and learning. Traditional
media is sustained by the news published by their reporters, journalists, and editors—all
of whom are professionals. Considering the consistent empirical connection found be-
tween positive attitudes toward professional news, general professional news use, and
trust in the news (Fernandez-Planells, 2015), we expect people’s preference for jour-
nalistic news curation to be positively related to audiences’trust in news that come from
legacy media outlets curated by journalists:
H1. Preference for journalists’news selection will be positively associated to trust in
traditional news a) cross-sectionally, b) time-lagged, and c) panel autoregressive
relationships.
4Journalism 0(0)
Journalists’news selection and social media news trust
News consumption patterns have substantially changed within the last years, partially due
to the presence of various social media outlets (Ardèvol-Abreu & Gil De Z´
uñiga, 2017;
Diehl et al., 2019). However, trust in news via these platforms is mixed at best –especially
in comparison with traditional news sources (Kruikemeier and Lecheler, 2018). The
massive circulation of fake news may be a strong rationale for such an evaluation
(McKewon, 2018). Social media news are greatly affected by misinformation campaigns
and content fabrications, while traditional and mainstream news media are, overall, faring
better (Newman et al., 2017). It should be noted, however, that depending on ideology and
party affiliation, what is seen as misinformation may vary, which leads to different
evaluations of and trust levels in media outlets (Barthel and Mitchell, 2017;Newport,
2017). Considering that news online can stem from a variety of sources and that some of
those sources may be non-professional, this media ecology may leave the audiences
unsure of what to believe (Flintham et al., 2018).
However, not all news on social media come from questionable sources or alternative
media organizations. According to Emmett (2008), most traditional media outlets publish
their content on social media, allowing people to share and widely spread them to an even
bigger audience (Hermida et al., 2012). Consequently, while a significant chunk of
content users encounter on social media may derive from non-professionals or may be
directly fabricated, the vast amount of news distributed comes from legacy media
(Thurman et al., 2019).
Other important factors that can impact trust in media in general and in news, in
particular, are ideology and partisanship (Kaye and Johnson, 2016;Lee, 2010;Suiter and
Fletcher, 2020). While high levels of partisanship generally increase distrust in news
media (Suiter and Fletcher, 2020), research suggests that compared to Democrats, Re-
publicans or conservatives (Lee, 2010) exhibit lower (news) media trust. This means that
the type of news typically consumed may also vary along partisanship and ideological
lines (Hollander, 2008;Kaye and Johnson, 2016), with both sides of the spectrum fo-
cusing on news consumption in line with their political/ideological stances (Hollander,
2008;Kaye and Johnson, 2016). This divide can affect not only news consumption but
also news dissemination (Weeks and Holbert, 2013). Especially when it comes to fake
news, ideology, and partisanship play important roles in believing and sharing fabricated
information, with conservatives being more inclined to share fake news than the liberals
(Allcott and Gentzkow, 2017;Pereira et al., 2021). This cannot only drastically impact
voting behavior but also stoke prejudices and fear, ultimately leading to more polari-
zation. It is, thus, even more important to investigate potential countermeasures to combat
these negative influences on democracy. Since prior literature suggests these powerful
differences in political stances, we control for both partisanship and the strength of
partisanship in all our models.
Although it is difficult to disentangle this convoluted distribution, there are strong
reasons to expect that preferring journalists as gatekeepers may also spill over to social
media, nurturing users’trust. As Sterrett et al. (2019) point out, even on social media, the
news source is instrumental in creating trust in the information received. News on social
Scheffauer et al. 5
media are considered more credible when shared by a news organization (Tandoc Jr,
2019), which is in line with results focusing on credibility judgments based on heuristic
cues (Metzger et al., 2010;Sundar, 2008). News provided by credible, journalistic sources
should carry more weight than encountering random information on social media, es-
pecially knowing that there are many fake news circulating online.
Furthermore, people who appreciate journalistic work and professional news selection
and gatekeeping may be more likely to seek out this content online (e.g., by following
news organizations on social media). Although users may not be fully able to control what
they see (e.g. algorithms boosting certain posts), it stands to reason that those who are
inclined to consume news from traditional organizations and outlets should more fre-
quently encounter this content on social media as well, due to their personal preferences
on social media (following, liking, sharing). After all, what users see on social media is to
some extent based on their previous interactions and behaviors (Joris et al., 2021).
Thus, those who prefer journalists’news selection may also expect to encounter
professional news on social media and, by extension, trust social media news too:
H2. Preference for journalists’news selection will be positively associated to trust in
social media news a) cross-sectionally, b) in time-lagged relationships, and c) in panel
autoregressive relationships.
Social media algorithmic news selection and trust in traditional/
social media news
Little is known concerning audience perceptions of algorithmic news selection. Some
studies report more preference towards human recommendations (Yeomans et al., 2019),
while others suggest that readers tend to favor algorithmic curation (Thurman et al.,
2019). Complexity is added to the topic when regarding different types of algorithmic
selection, e.g. based on individuals’reading habits versus content selected based on their
friends’habits (Bachmann & Gil De Z ´
uñiga, 2013;Joris et al., 2021). There is, moreover,
a host of studies highlighting problems of users interacting with algorithms. Users exhibit
a certain form of skepticism towards (algorithmic) news selection as they do not fully
understand how the algorithms operate (Fletcher and Nielsen, 2019;Swart, 2021) and
often lack the vocabulary to articulate these matters (Swart, 2021). What is worth noting is
that several studies have demonstrated that users’acceptance of, knowledge about, and
interaction with algorithms differs substantially across the population (Gran et al., 2021;
Min, 2019). Classification attempts show that some users appear unaware or disengaged,
while others are critical, skeptical, and try to challenge the algorithms. There are,
however, also those who regard algorithms positively and deliberately engage in curation
with them (Gran et al., 2021;Min, 2019).
While empirical research on algorithmic news selection and online media is thus far
inconclusive, the potential connection with trust in traditional news is even less clear. The
MAIN Model Sundar (2008) suggests that different technological affordances can
substantially impact the credibility of digital media, going beyond mere content char-
acteristics. Thus, algorithmic news selection could have the potential to alter credibility
6Journalism 0(0)
perceptions of news. As Thurman et al. (2019) show, based on data stemming from
26 countries, users are aware of the distinction between algorithmic and editorial news
selection and value them differently.
We argue that attitudes towards algorithmic news selection might limitedly affect trust
levels in traditional media or may even fuel distrust. Algorithms apply an updated set of
news values to provide users with information, matching, among other things, likes,
dislikes, and prior engagement (DeVito, 2017). While some users might be unaware of
how algorithms operate, others engage in conscious efforts to adapt and tweak algorithmic
selection specific to their needs and wants (Gran et al., 2021;Min, 2019). Users might
prefer this type of selection and thus become wary of seeing information outside of what
they –actively or passively –trained the algorithm to show them. In a worst-case scenario,
preference for algorithmic news selection could reduce trust in traditional news by trading
factual, unbiased information from journalists for self-curated echo chambers. The in-
formation present in such echo chambers can be questionable and might even be fab-
ricated (e.g., fake news), leading to a mis- or ill-informed citizenry, which directly poses a
negative impact on democracy. However, as an important factor of algorithmic news
selection focuses on getting news online based on individuals’prior behaviors and their
friends’and contacts’behaviors, there are strong reasons to expect that the gatekeeping
function of AI is only associated to trust the digital environment, thus leaving open how
citizens’preference for algorithmic news selection affects trust in traditional news:
RQ1. How will preference for social media algorithmic news selection relate to
people’s trust in traditional news?
As for preferring algorithmic news selection and audiences’trust in social media news,
prior research provides no clear expectations. Some users may trust news on social media
because they hold favorable views toward algorithms and their curation intelligence. They
may have actively attempted to ensure the news provided by algorithmic selection reflect
their interests and trusted sources. Although users hold positive attitudes about this type of
news selection, they may also be aware that at least a chunk of what they encounter in
aggregated news environments, such as social media, can stem from non-professional
sources and low-reputation news organizations which could diminish their trust levels.
Furthermore, despite receiving news via social media, a subset of users may not be aware
of the algorithmic selection performed in the background or may not fully understand how
it works. While these users trust the news on social media, they may not associate them
with algorithmic selection. Based on these notions, we propose a measurement that
captures this construct, namely appraisals of different algorithmic selection behaviors
(based on individuals’and their friends’previous behaviors) and perception of limitations
(e.g., variety of news received). As prior findings remain largely inconsistent, we propose
the subsequent research question:
RQ2. How will preference for social media algorithmic news selection relate to
people’s trust in news encountered on social media?
Scheffauer et al. 7
Journalists’news selection, traditional news use, and trust in
traditional news
We, furthermore, seek to clarify the role traditional news use may play in explaining the
association between preference for journalists’news selection and trust in news. There are
different influential factors in fostering trust in news including political antecedents such
as ideology, partisanship, and even trust in government (Lee, 2010). However, it also
extends to news consumption (Fernandez-Planells, 2015). Several studies underline the
positive association between consuming traditional news and trust in media (Mourão
et al., 2018) and the negative relationship between mainstream news exposure and media
skepticism (Tsfati and Cappella, 2003). Several researchers highlighted the connection
between news consumption and positive evaluations of the press (Gil De Z ´
uñiga and
Hinsley, 2013;Holton et al., 2013), with increased news consumption going hand in hand
with positive attitudes towards journalism.
In line with this, it stands to reason that the effects of preference for journalists’news
selection on trust in traditional news are contingent upon citizens’levels of traditional
news consumption. Positive evaluations of journalistic work should go hand in hand with
consuming more news by journalists, influencing each other in a symbiotic
relationship. Users who favor journalists’professional work and consume more tradi-
tional news may have more positive appraisals about the news they encounter. We predict
the highest levels of trust in traditional news when both preference for journalists’news
selection and traditional news use are high:
H3. There will be a a) cross-sectional, b) time-lagged, and c) panel autoregressive,
interaction effect of traditional news use (M) on the relationship between preference for
journalists’news selection (X) and trust in traditional news (Y).
Journalists’news selection, traditional news use, and trust in
social media news
While previous research shows that reliance on online news and online interactions with
journalists positively impact audience evaluations of news (Gil De Z ´
uñiga et al., 2018),
the amalgam of news sources on social media can make it difficult for users to judge what
is credible and what not (Salwen et al., 2004). However, as a big portion of news online is
also produced by traditional news outlets (Diel, 2017;Thurman et al., 2019), the positive
connection between trust in traditional news and high levels of news consumption
(Mourão et al., 2018) may also be impactful online.
Building on the expectation that a preference for journalistic news selection positively
influences trust in news on social media (H2), we expect people, who exhibit a preference
for journalists’gatekeeping and consume more news from legacy outlets, to judge this
news to be more trustworthy even when encountered on social media. Trust in social
media news should, thus, reach the highest values when high levels of both preference for
journalists’news selection and traditional news use are reported:
8Journalism 0(0)
H4. There will be a a) cross-sectional, b) time-lagged, and c) panel autoregressive,
interaction effect of traditional news use (M) on the relationship between preference for
journalists’news selection (X) and trust in social media news (Y).
Method
Sample
The data used in this study was obtained by IPSOS and stems from a larger research project
concerned with behavioral as well as attitudinal outcomes of different types of media uses.
The Qualtrics online survey, which targeted adults in the U.S., consisted of two waves (June
and October 2019). Striving for national representativeness and thus allowing for gener-
alizable inference concerning the U.S. population, an opt-in panel curated by IPSOS
consisted of hundreds of thousands U.S. respondents. In order to match with the U.S. census
in terms of key demographic elements (e.g. gender, education, income), a subsample from
this pool, which comprised 3000 individuals, was stratified. The final numbers of the sample
were 1338 valid cases for Wave 1 and 511 cases for Wave 2. Unless otherwise stated, all
indexes were measured on Likert type scales ranging from 1 to 10.
Independent and dependent variables
Variables of interest
Preference for Algorithmic News Selection. This variable taps into respondents’attitudes
towards algorithmic news selection (Gil De Z ´
uñiga et al., 2022): “Having stories au-
tomatically selected for me on the basis of what I have consumed in the past is a good way
to get the news,”“News based on an algorithm does not limit my exposure to important
news,”“Relying on a news diet based on algorithms ensures that I receive news that is
most relevant to me,”“Having stories automatically selected for me on the basis of what
my friends have consumed is a good way to get news,”“News based on an algorithm of
what my friends consume does not limit my exposure to important news”(Cronbach’s
α= 0.89; M = 4.28; SD = 2.29).
We conducted two confirmatory factor analyses (CFAs) to compare the model pro-
posed for this study (consisting of the dimensions preference for algorithmic and
journalists’news selection) with the alternative model (one latent variable for news
selection preference). The analyses reveal that the two-factors-model provides indeed a
better fit for the data (see Appendix Tables A2 and A3,χ
2
= 197.8, df = 19, p< 0.001,
CFI = 0.953, TLI = 0.931, RMSEA = 0.084, SRMR = 0.031, BIC = 44,015.14.
AIC = 43,885.19).
Preference for Journalists’News Selection. Asking respondents for the level of (dis-)
agreement for the following three items, an index for journalists’news selection was
created: “Having stories selected for me by editors and journalists is a good way to get
news,”“Editors and journalists know best what is most relevant to me,”“News selection
Scheffauer et al. 9
by editors and journalists ensure that I get exposed to important news”(Cronbach’s
α= 0.91; M = 4.12; SD = 2.51).
Social Media News Trust. This construct was measured using three items inquiring how
much respondents trust news on social media, such as Facebook or Twitter (W
1
:
Cronbach’sα= 0.86; M = 4.02; SD = 2.37; W
2
: Cronbach’sα= 0.90; M = 3.77;
SD = 2.47).
Traditional News Trust. To capture trust in traditional news, respondents were asked how
much they generally trust news “that comes from mainstream news media (e.g.,
newspapers, TV newscasts, online news sites),”“that is fact-checked?”(W
1
: Spearman-
Brown coefficient = 0.68; M = 6.24; SD = 2.25; W
2
: Spearman-Brown coefficient = 0.80;
M = 6.36; SD = 2.41).
Control variables
Demographics. To control for any possible intervening factors we include a host of
demographic variables such as age (Median: 3 [36-55]), gender (46.7% males), education
(Median: 3 [Some college], range: 1 = “less than high school”to 8 = “doctoral degree”),
family income (Median: 4 [$ 50,000 to $ 99,999], range: 1 = “0 to $ 14,999”to 7 = “$
200,000 or more”), and race (75.2% white).
Social Media Use. We also control each respondent’s overall social media use which was
measured as single item. (M = 2.21, SD = 0.02, range: one to 3)
Traditional Media Use. In order to create an index of respondents’traditional media use, we
tapped their frequency of getting news from the following sources: “network TV (e.g.
ABC, CBS, NBC),”“local television news (cf. local affiliate stations),”“MSNBC cable
news, CNN cable news,”“FOX cable news,”“television,”“national newspapers (e.
g. The New York Times,The Washington Post,USA Today),”“local newspapers (e. g. The
Oregonian,Houston Chronicle,The Miami Herald),”“printed,”“online news sites (e.
g. Politico, VOX, BuzzFeed),”“local news online sites (online sites related to news in
your local community),”“radio news (e. g. NPR, talk shows),”“radio?”(Cronbach’s
α= 0.88; M = 4.46; SD = 1.90).
Social Media News Use. This construct was measured with 14 items to assess how often
respondents got different kinds of news from social media sources, such as Facebook,
Twitter, or WhatsApp (Cronbach’sα= 0.91; M = 3.33; SD = 1.99).
Political Interest. For this construct, two items were computed: “How interested are you in
information about what’s going on in politics and public affairs,”“How closely do you
pay attention to information about what’s going on in politics and public affairs?”
(Spearman-Brown coefficient = 0.94; M = 6.13; SD = 2.72).
10 Journalism 0(0)
Political Discussion. Creating an index of online and offline discussion, we assessed re-
spondents’frequency of talking about politics or public affairs with the following:
“spouse/partner, family, relatives,”“friends,”“neighbors, co-workers you know well,”
“acquaintances,”“strangers,”“neighbors, co-workers you don’t know well?”(Cron-
bach’sα= 0.93; M = 3.38; SD = 2.03.)
Political Knowledge. In order to determine respondents’levels of political knowledge,
respondents were asked eight questions concerning the political system and important
political actors. This measurement is drawn from prior work (e. g., Delli Carpini and
Keeter, 1993) and correct answers were coded as 1, incorrect ones as 0 (M = 2.77,
SD = 2.02, Guttman λ= 0.71).
Strength of Partisanship. Drawing from previous research (Lee et al., 2013), we asked
respondents about their level of party identification on a scale of 0 (= Strong Democrat) to
10 (=Strong Republican), with five representing Independent. This variable was,
moreover, recoded into a 6-point scale to capture weak to strong partisanship (M = 2.21;
SD = 1.44).
Partisanship. We measure this concept by asking respondents about their party identi-
fication (1 = Democrat, 2 = Independent, 3 = Republican, M = 2.02, SD = 0.02).
Network Size (log). To measure respondents’network size, they were asked two questions
about their behaviors within the last month of our survey: “About how many total people
have you talked to about politics or public affairs: (a) face-to-face or over the phone, (b)
via the Internet, including email, chat rooms, social networks?”
(M = 0.42; SD = 0.42; Spearman-Brown coefficient: 0.48).
Results
While the constructs have theoretical face validity, we opted for a confirmatory factor
analysis to provide additional empirical validation. The measures of fit showed robust
factors (Appendix Table A2), indicating reliable measures for preference of journalists’
and algorithmic news selection (χ
2
= 197.8; df = 19; p= 0.001; RMSEA = 0.08; CFI =
0.953; TLI = 0.931; SRMR = 0.03). We addtionally provide zero-order-correlations for all
key variables included in our analyses (Appendix Table A1).
We first investigated the positive connection between preference for journalists’news
selection and trust in traditional news (H1). The regression analysis shown in Table 1
demonstrated significant positive results cross-sectionally (β= 0.170, p< 0.001,
ΔR2 = 3.0%). This finding is consistent in the lagged (β= 0.221, p< 0.01, ΔR2 = 2.4%)
and autoregressive (β= 0.117, p< 0.05, ΔR2 = 0.5%) models. Our results suggest that
preference for journalists’news selection is an important factor in fostering trust in
traditional news even after controlling for demographics, media antecedents, political
orientations, and prior levels of trust (W
1
, for the autoregressive model).
Scheffauer et al. 11
For our second hypothesis (H2), we proposed a positive connection between pref-
erence for journalists’news selection and trust in social media news. As shown in Table 2,
our predictions were confirmed as it demonstrated a positive association between
preference for journalists’news selection and trust in social media news cross-sectionally
Table 1. Cross-sectional, lagged, and autoregressive regression models testing preferences for
social media algorithmic and journalists news selection, and traditional news trust.
Traditional news
trust (W
1
)
Traditional new
trust (W
2
lagged)
Traditional new
trust (W
2
autoregressive)
Block 1: Autoregressive Term
Trust traditional
news
W1
—— 0.524***
ΔR
2
—— 43.6%
Block 1: Demographics
Age 0.043 0.028 0.022
Gender (female) 0.102*** 0.095* 0.034
Education 0.021 0.012 0.001
Income 0.035 0.035 0.032
Race (white) 0.021 0.014 0.021
ΔR
2
2.2% 3.3% 0.7%
Block 2: Media Antecedents
Social media use freq. 0.055* 0.084 0.073
Traditional news use 0.357*** 0.281*** 0.081
Social media news use 0.007 0.026 0.013
ΔR
2
(%) 19.3% 13.4% 0.8%
Block 3: Political Orientations
Political interest 0.179*** 0.129* 0.058
Network size (log) 0.041 0.033 0.011
Political discussion 0.149*** 0.128* 0.058
Political knowledge 0.020 0.100* 0.075
Strength of partisanship 0.012 0.014 0.021
Partisanship 0.225*** 0.299*** 0.150***
ΔR
2
8.8% 12.0% 2.7%
Block 4: Variables of Interest
Preference for
algorithmic news
selection
0.057 0.045 0.049
Preference for
journalists’news
selection
0.170*** 0.221** 0.117*
ΔR
2
3.0% 2.4% 0.5%
Total R
2
33.4% 31.1% 48.4%
Note: Sample size = 1338 (Wave 1); 511 (Wave 2). Cell entries are final-entry OLS standardized Beta (β)
coefficients. *p< 0.05; **p< 0.01; ***p<0 .001.
12 Journalism 0(0)
(β= 0.095, p< 0.01, ΔR2 = 3.2%), and lagged (β= 0.172, p< 0.01, ΔR2 = 3.0%). The
autoregressive model, however, is slightly above the significance threshold (β= 0.100,
p= 0.053, ΔR2 = 0 .7%).
In order to investigate the potential influence of preference for algorithmic news
selection on trust in news, we posed two research questions. The results for the association
with traditional news showed no significant results in either cross-sectional, lagged, or
autoregressive models (RQ1). As for the influence of favoring algorithmic news selection
on social media news, the results were positive but only significant in the cross-sectional
model (β= 0.144; p< 0.001, ΔR2 = 3.2%). Hence, we could not confirm any robust
connection between preference for algorithmic news selection and trust in social media
news (RQ2).
Our third hypothesis (H3) aimed at investigating the interaction effect of traditional
news use on the relationship between preference for journalists’news selection and trust
in traditional news. We expected users, who prefer journalists’news selection and are avid
consumers of traditional news, to also exhibit the highest levels of trust in traditional
news. Table 3 illustrates that the interaction is not significant in the cross-sectional or the
lagged model. However, in the autoregressive model (β=0.060, p< 0.05, ΔR2 = 1.0%)
our moderating variable exhibited a statically significant interaction effect. We plotted the
interaction term in Figure 1 to draw more precise conclusions. Respondents, who were
high in traditional news consumption and exhibited higher levels of preference for
journalists’news selection, were indeed individuals who trusted traditional news more
than those scoring low.
Finally, for the fourth hypothesis (H4), the same prediction was made by testing the
interaction influence on trust in social media news. Users, who held journalists’news
selection in high regard and consumed a lot of traditional news, were expected to exhibit
the highest trust levels in social media news. Here, we obtained significant results only in
the cross-sectional model (β= 0.028, p< 0.05, ΔR2 = 1.2%), but not for the lagged or
autoregressive model.
Discussion
As trust in news is significantly declining in the U.S., the ideal of an informed citizenry
may be under further threat. On this account, it is important to understand the dynamics
that may reignite audiences’faith in the news, either when individuals are actively
consuming or inadvertently exposed. This paper sought to better uncover what sparks
levels of trust in news, relying on a two-wave U.S. panel survey. Our findings flesh out the
idea that, in a growing media ecology shaped by algorithmic news selection, journalists’
editorial work remains a determinant factor in explaining people’s trust in news (both
online and offline). Considering the low news trust ratings public opinion polls report for a
great many countries, focusing on the journalistic craft could, thus, be a way to combat
unfavorable news trust assessments.
When it comes to examining the impact of people’s preference for algorithmic news
selection, the results lend support to a different interpretation. Either audiences cogni-
tively discern between the role of algorithmic and journalists’gatekeeping or simply
Scheffauer et al. 13
ignore the effects and functions of AI in news curation processes. Algorithms are usually
designed to unobtrusively operate in the background, with users not fully understanding
or being aware of the processes (Fletcher and Nielsen, 2019;Swart, 2021). Thus, users
appear to be largely indifferent and perhaps unconcerned about the potential effects of
Table 2. Cross-sectional, lagged, and autoregressive regression models testing social media
algorithmic and journalists news selection, and traditional news trust.
Social Media news
trust (W
1
)
Social Media news
trust (W
2
lagged)
Social Media news
trust (W
2
autoregressive)
Block 1: Autoregressive Term
Social media traditional
news
W1
0.494***
ΔR
2
53.7%
Block 1: Demographics
Age 0.024 0.081* 0.063
Gender (female) 0.029 0.071* 0.049
Education 0.017 0.040 0.035
Income 0.008 0.058 0.035
Race (white) 0.030 0.054 0.049
ΔR
2
12.9% 19.1% 2.1%
Block 2: Media Antecedents
Social media use freq. 0.142*** 0.197*** 0.115**
Traditional news use 0.101*** 0.015 0.019
Social media news use 0.427*** 0.361*** 0.118*
ΔR
2
(%) 40.0% 27.1% 1.7%
Block 3: Political Orientations
Political interest 0.001 0.027 0.024
Network size (log) 0.034 0.001 0.019
Political discussion 0.013 0.014 0.021
Political knowledge 0.057* 0.008 0.032
Strength of partisanship 0.042* 0.061 0.032
Partisanship 0.001 0.004 0.014
ΔR
2
0.8% 0.5% 0.3%
Block 4: Variables of Interest
Preference for
algorithmic news
selection
0.144*** 0.059 0.008
Preference for
journalists’news
selection
0.095** 0.172** 0.100
¤
ΔR
2
3.2% 3.0% 0.7%
Total R
2
56.8% 49.7% 58.4%
Note: Sample size = 1338 (Wave 1); 511 (Wave 2). Cell entries are final-entry OLS standardized Beta (β) coefficients.
*p < 0.05; **p < 0.01; ***p < 0.001;
¤
p = 0.053.
14 Journalism 0(0)
Table 3. Cross-sectional, lagged, and autoregressive interaction effects test.
Traditional
media news
trust (cross.)
Traditional media
news trust
(lagged)
Traditional media
news trust
(autoregr.)
Social media
news trust
(cross.)
Social media
news trust
(lagged)
Social media
news
trust (autoregr.)
Block 1: All Prior Blocks Table 1
ΔR
2
33.4% 31.1% 48.8% 56.8% 49.7% 58.4%
Block 2: Interaction
Pref. For journalists’Selec.
*Trad. News
0.009 (0.014) 0.047 (0.027) 0.060 (0.024)* 0.028 (0.012)* 0.014 (0.023) 0.018 (0.020)
ΔR
2
2% 1.7% 1% 1.2% 7.7% 10.2%
Total R
2
35.4% 32.8% 47.8% 55.6% 57.4% 68.6%
Note: Estimates are unstandardized Beta coefficients. Standardized errors between brackets. Interaction effects based on bootstrapping to 5000 samples with biased
corrected confidence intervals. The effects account for the same demographic, political antecedents and media orientations control variables as found in Tables 1 and 2
Sample-W
1
= 1338; Sample-W
2
= 511.
Scheffauer et al. 15
algorithms in shaping their news diets online and offline. This once more underscores the
crucial gatekeeping role of professional (i.e., human) journalists.
Our findings additionally suggest that the moderating effects of traditional news consumption
are perhaps more intricate than expected. Future research should investigate the influence of
online or social media news consumption that might prove to be more important for predicting
trust in online news than traditional news consumption. Users who trust traditional news may not
be the same people who trust online news. Furthermore, additional investigation is needed to
determine how citizen journalism factors into these relationships. As partisanship emerged as a
significant predictor of trust in traditional news, future research should pay attention to dis-
tinctions among those lines. As prior literature has evidenced, there are differences in media
consumption as well as media trust based on a person’s ideology and partisanship. Thus, a closer
investigation may uncover nuances that are beyond the scope of this paper.
Overall, our results showed the strong impact preference for journalists’news selection
has on trust in online and offline news. Hence, it is vital that news outlets focus on
professionals to maintain and raise trust in their published content, reinforcing the notion
that professional journalists are paramount in fostering an informed citizenry. This condition
holds true in the case of socialmedia, an environment where a plethora of agents can publish
information. News passed along by professionals appear to be the most trusted.
Apart from the impacts of our research, this study has some limitations. All our
measurements are self-reported and based on surveys. Therefore, future research may
implement experimental studies to validate with greater certainty the causal associations.
However, in order to alleviate this issue, this study relied on a two-wave panel survey data,
which allowed for more stringent causal order relationships with autoregressive models
(See for a thorough explanation, Regis et al., 2022). Additionally, our questionnaire did
Figure 1. Autoregressive Panel Interaction Effect of Traditional News Use (M) on the Relationship
Between Preference for Journalists’News Selection (X) and Traditional News Trust (Y). The
theoretical model controls for the same demographic, media antecedents and political orientations
control variables as found in Table 3, as well as the autoregressive term for the dependent variable.
Group is the mean plus/minus SD.
16 Journalism 0(0)
not include any inquiries about respondents’understanding of how algorithms work and
may impact society. A lack of correct knowledge on the subject might have influenced
their attitudes and perceptions. While we tried to provide enough information to our
respondents to ensure they grasped the concept that we intended to measure (see Newman
et al., 2016;Thurman et al., 2019), the participants’unawareness of algorithms and how
they work in daily life cannot be entirely ruled out. Thus, people might not consciously
deliberate on who selects their news and whether to trust those sources to the degree we
assume. Similarly, the term news may mean a variety a of things and thus be interpreted as
“news story,”“news organization,”or “news outlet.”As is the case with algorithms, we
did not include any questions to inquire about the respondents’understanding of these
terms. Nevertheless, we believe our measurements have merit as we tried to use ex-
planations instead of labels that might be confusing to individuals.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/
or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or
publication of this article: This work has benefited from the support of the Spanish National
Research Agency’s Program for the Generation of Knowledge and the Scientific and Technological
Strengthening Research + Development Grant PID2020-115562GB-I00. The last author is funded
by the ‘Beatriz Galindo Program’from the Spanish Ministry of Science, Innovation & Universities,
and the Junta de Castilla y León. Responsibility for the information and views set out in this study
lies entirely with the authors.
ORCID iDs
Rebecca Scheffauer https://orcid.org/0000-0003-4545-2062
Manuel Goyanes https://orcid.org/0000-0001-6537-9777
Homero Gil de Zuniga https://orcid.org/0000-0002-4187-3604
References
Abdenour J, McIntyre K and Dahmen NS (2020) Seeing eye to eye: a comparison of audiences’and
journalists’perceptions of professional roles and how they relate to trust. Journalism Practice
15: 329–347. DOI: 10.1080/17512786.2020.1716828
Allcott H and Gentzkow M (2017) Social media and fake news in the 2016 election. Journal of
Economic Perspectives 31(2): 211–236. Article 2. DOI: 10.1257/jep.31.2.211
Ardèvol-Abreu A and Gil de Z´
uñiga H (2017) Effects of editorial media bias perception and media
trust on the use of traditional, citizen, and social media news. Journalism and Mass Com-
munication Quarterly 94(3): 703–724. Article 3. DOI: 10.1177/1077699016654684
Scheffauer et al. 17
Bachmann I and de Z´
uñiga HG (2013) News platform preference as a predictor of political and civic
participation. Convergence: The International Journal of Research into New Media Tech-
nologies 19(4): 496–512.
Barthel M and Mitchell A (2017, May 10) Americans’attitudes about the news media deeply
divided along partisan lines. In: Pew Research Center’s Journalism Project.https://www.
journalism.org/2017/05/10/americans-attitudes-about-the-news-media-deeply-divided-along-
partisan-lines/
Bentivegna S and Marchetti R (2018) Journalists at a crossroads: are traditional norms and practices
challenged by Twitter? Journalism 19(2): 270–290. DOI: 10.1177/1464884917716594
Bodó B (2019) Selling news to audiences –a qualitative inquiry into the emerging logics of al-
gorithmic news personalization in European quality news media. Digital Journalism 7(8):
1054–1075. DOI: 10.1080/21670811.2019.1624185
Dawson A (2017 March 16) Public trust in Australian media falls to all-time low. Mumbrella https://
mumbrella.com.au/trust-in-daily-and-local-newspapers-on-a-steady-decline-according-to-
essential-research-432670
Carpini MXD and Keeter S (1993) Measuring political knowledge: putting first things first.
American Journal of Political Science 37(4): 1179–1206. DOI: 10.2307/2111549
DeVito MA (2017) From editors to algorithms. Digital Journalism 5(6): 753–773. Article 6. DOI:
10.1080/21670811.2016.1178592
Diehl T, Barnidge M and Gil de Zuniga H (2019) Multi-platform news use and political participation
across age groups: toward a valid metric of platform diversity and its effects. Journalism and
Mass Communication Quarterly 96(2): 428–451.
Diel S (2017) New media, legacy media and misperceptions regarding sourcing. Kome 5(1):
104–120.
Donohue GA, Tichenor PJ and Olien CN (1972) Gatekeeping: mass media systems and information
control. Current Perspectives in Mass Communication Research 1: 41–70.
Emmett A (2008) Networking news: traditional news outlets turn to social networking Web sites in
an effort to build their online audiences. American Journalism Review 30(6): 40–44.
European Broadcasting Unit (2017) Trust in Media 2017: EBU Report. EBU Media Intelligence
Service. https://www.epra.org/news_items/trust-in-media-2017-ebu-report
Fernandez-Planells A (2015) Factors Influencing Trust in Media: Exploring the Association be-
tween Media Consumption and News about the 15M Movement. Hipertext net Revista
Acad´
emica Sobre Documentación Digital y Comunicación Interactiva Vol 13: 1–14.
Fisher C (2016) The trouble with ‘trust’in news media. Communication Research and Practice
2(4): 451–465. DOI: 10.1080/22041451.2016.1261251
Flaxman S, Goel S and Rao JM (2016) Filter bubbles, echo chambers, and online news con-
sumption. Public Opinion Quarterly 80(S1): 298–320. DOI: 10.1093/poq/nfw006
Fletcher R and Nielsen RK (2019) Generalised scepticism: how people navigate news on social
media. Information, Communication and Society 22(12): 1751–1769. DOI: 10.1080/
1369118X.2018.1450887
Flintham M, Karner C, Bachour K, et al. (2018) Falling for fake news: investigating the con-
sumption of news via social media. In: Proceedings of the 2018 chi conference on human
factors in computing systems, Montreal QC Canada, April 21-26, 2018, 1–10. DOI: 10.1145/
3173574.3173950
18 Journalism 0(0)
Gil de Z´
uñiga H, Cheng Z and Gonz´
alez-Gonz´
alez P (2022) Effects of the news finds me perception
on algorithmic news attitudes and social media political homophily. Journal of Communication
72(5): 578–591.
Gil de Z´
uñiga H, Diehl T and Ardèvol-Abreu A (2018) When citizens and journalists interact on
Twitter: expectations of journalists’performance on social media and perceptions of media
bias. Journalism Studies 19(2): 227–246.
Gil de Z´
uñiga H and Hinsley A (2013) The press versus the public. Journalism Studies 14(6):
926–942. DOI: 10.1080/1461670X.2012.744551
Goyanes Manuel, Scheffauer Rebecca and Gil de Z ´
uñiga Homero (2023) News Distribution and
Sustainable Journalism: Effects of Social Media News Use and Media Skepticism on Citizens’
Paying Behavior. Mass Communication and Society. doi: 10.1080/15205436.2023.2169164
Gran A-B, Booth P and Bucher T (2021) To be or not to be algorithm aware: a question of a new
digital divide? Information, Communication and Society 24(12): 1779–1796. DOI: 10.1080/
1369118X.2020.1736124
Harambam J, Helberger N and van Hoboken J (2018) Democratizing algorithmic news recom-
menders: how to materialize voice in a technologically saturated media ecosystem. Philo-
sophical transactions. Series A, Mathematical, physical, and engineering sciences 376(2133):
20180088. Article 2133. DOI: 10.1098/rsta.2018.0088
Heinderyckx F and Vos T (2016) Reformed gatekeeping. CM: Communication and Media 11(36):
29–46.
Hermida A, Fletcher F, Korell D, et al. (2012) Share, Like, Recommend: decoding the social media
news consumer. Journalism Studies 13(5–6): 815–824. DOI: 10.1080/1461670X.2012.
664430
Hollander BA (2008) Tuning out or tuning elsewhere? Partisanship, polarization, and media
migration from 1998 to 2006. Journalism and Mass Communication Quarterly 85(1): 23–40.
DOI: 10.1177/107769900808500103
Holton AE, Coddington M and Gil de Z´
uñiga H (2013) Whose news? Whose values? Citizen
journalism and journalistic values through the lens of content creators and consumers.
Journalism Practice 7(6): 720–737.
Janowitz M (1975) Professional models in journalism: the gatekeeper and the advocate. Journalism
Quarterly 52(4): 618–626. DOI: 10.1177/107769907505200402
Joris G, Grove FD, Van Damme K, et al. (2021) Appreciating news algorithms: examining au-
diences’perceptions to different news selection mechanisms. Digital Journalism 9(5):
589–618. DOI: 10.1080/21670811.2021.1912626
Kaufhold K, Valenzuela S and de Z´uñiga HG (2010) Citizen journalism and democracy: how user-
generated news use relates to political knowledge and participation. Journalism and Mass
Communication Quarterly 87(3–4): 515–529. DOI: 10.1177/107769901008700305
Kaye BK and Johnson TJ (2016) Across the great divide: how partisanship and perceptions of media
bias influence changes in time spent with media. Journal of Broadcasting and Electronic
Media 60(4): 604–623.
Kruikemeier S and Lecheler S (2018) News consumer perceptions of new journalistic sourcing
techniques. Journalism Studies 19(5): 632–649.
Scheffauer et al. 19
Lee N-J, Shah DV and McLeod JM (2013) Processes of political socialization: a communication
mediation approach to youth civic engagement. Communication Research 40(5): 669–697.
Article 5. DOI: 10.1177/0093650212436712
Lee T-T (2010) Why they don’t trust the media: an examination of factors predicting trust. American
Behavioral Scientist 54(1): 8–21. DOI: 10.1177/0002764210376308
Malmelin N and Villi M (2016) Audience community as a strategic resource in media work.
Journalism Practice 10(5): 589–607. DOI: 10.1080/17512786.2015.1036903
McKewon E (2018) Public Trust in Journalism: An Annotated Bibliography. Au: Centre for Media
Transition, University of Technology Sydney. Available at: Cmt. Uts. Edu.
Meraz S and Papacharissi Z (2013) Networked gatekeeping and networked framing on #Egypt. The
International Journal of Press/Politics 18(2): 138–166. DOI: 10.1177/1940161212474472
Metzger MJ, Flanagin AJ and Medders RB (2010) Social and heuristic approaches to credibility
evaluation online. Journal of Communication 60(3): 413–439. DOI: 10.1111/j.1460-2466.
2010.01488.x
Min SJ (2019) From algorithmic disengagement to algorithmic activism: charting social media
users’responses to news filtering algorithms. Telematics and Informatics 43: 101251. DOI: 10.
1016/j.tele.2019.101251
Mourão RR, Thorson E, Chen W, et al. (2018) Media repertoires and news trust during the early
trump administration. Journalism Studies 19(13): 1945–1956. DOI: 10.1080/1461670X.2018.
1500492
Nah S and Chung DS (2012) When citizens meet both professional and citizen journalists: social
trust, media credibility, and perceived journalistic roles among online community news
readers. Journalism 13(6): 714–730. DOI: 10.1177/1464884911431381
Nechushtai E and Lewis SC (2019) What kind of news gatekeepers do we want machines to be?
Filter bubbles, fragmentation, and the normative dimensions of algorithmic recommendations.
Computers in Human Behavior 90: 298–307. DOI: 10.1016/j.chb.2018.07.043
Newman N (2011) Mainstream Media and the Distribution of News in the Age of Social Media.
https://ora.ox.ac.uk/objects/uuid:94164da6-9150-4938-8996-badfdef6b507
Newman N, Fletcher R, Kalogeropoulos A, et al. (2017) Reuters Institute Digital News Report.
Reuters Institute.
Newman N, Fletcher R, Levy DAL, et al. (2016) Reuters Institute Digital News Report.https://
reutersinstitute.politics.ox.ac.uk/our-research/digital-news-report-2016.
Newman N, Fletcher R, Schulz A, et al. (2021) Reuters Institute Digital News Report 2021. Reuters
Institute for the Study of Journalism.
Newport F (2017 June 26) Americans’Confidence in Institutions Edges up. Gallup.Com. https://
news.gallup.com/poll/212840/americans-confidence-institutions-edges.aspx
Pereira A, Harris E and Van Bavel JJ (2021) Identity Concerns Drive Belief: The Impact of Partisan
Identity on the Belief and Dissemination of True and False News. 1st ed. Group Processes and
Intergroup Relations, 26, 24–47. Available at: https://doi.org/10.1177/13684302211030004
Regis M, Serra P and van den Heuvel ER (2022) Random autoregressive models: a structured
overview. Econometric Reviews 41(2): 207–230. DOI: 10.1080/07474938.2021.1899504
Salwen MB, Garrison B and Driscoll PD (2004) Online News and the Public. Mahwah, New Jersey:
Lawrence Erlbaum Associates, Inc.
20 Journalism 0(0)
Scheffauer Rebecca, Goyanes Manuel and Gil de Z´
uñiga Homero (2021) Beyond social media news
use algorithms: how political discussion and network heterogeneity clarify incidental news
exposure. Online Information Review 45(3): 633–650. https://doi.org/10.1108/OIR-04-2020-
0133
Shoemaker PJ and Riccio JR (2016) Gatekeeping. The International Encyclopedia of Political
Communication. Wiley-Blackwell, 1–5. Available at: https://doi.org/10.1002/9781118541555.
wbiepc202
Shoemaker PJ, Vos TP and Reese SD (2009) Journalists as gatekeepers. The Handbook of
Journalism Studies. New York: Routledge, 93–107.
Sterrett D, Malato D, Benz J, et al. (2019) Who shared it?: deciding what news to trust on social
media. Digital Journalism 7(6): 783–801. DOI: 10.1080/21670811.2019.1623702
Suiter J and Fletcher R (2020) Polarization and partisanship: key drivers of distrust in media old and
new? European Journal of Communication 35(5): 484–501. DOI: 10.1177/
0267323120903685
Sundar SS (2008) The MAIN Model: A Heuristic Approach to Understanding Technology Effects
on Credibility. Digital Media, Youth, and Credibility. Cambridge, MA: The MIT Press.
Sunstein CR (2018) #Republic: Divided Democracy in the Age of Social Media. Princeton, NJ:
Princeton University Press.
Swart J (2021) Experiencing algorithms: how young people understand, feel about, and engage with
algorithmic news selection on social media. Social Media + Society 7(2): 205630512110088.
DOI: 10.1177/20563051211008828
Tandoc EC Jr. (2019) Tell me who your sources are: perceptions of news credibility on social media.
Journalism Practice 13(2): 178–190.
Thorson K (2020) Attracting the news: algorithms, platforms, and reframing incidental exposure.
Journalism 21(8): 1067–1082. DOI: 10.1177/1464884920915352
Thurman N, Moeller J, Helberger N, et al. (2019) My friends, editors, algorithms, and I. Digital
Journalism 7(4): 447–469. DOI: 10.1080/21670811.2018.1493936
Tsfati Yand Cappella JN (2003) Do people watch what they do not trust?: exploring the association
between news media skepticism and exposure. Communication Research 30(5): 504–529.
DOI: 10.1177/0093650203253371
Wallace J (2018) Modelling contemporary gatekeeping. Digital Journalism 6(3): 274–293. DOI:
10.1080/21670811.2017.1343648
Weeks BE and Holbert RL (2013) Predicting dissemination of news content in social media: a focus
on reception, friending, and partisanship. Journalism and Mass Communication Quarterly
90(2): 212–232. DOI: 10.1177/1077699013482906
Yeomans M, Shah A, Mullainathan S, et al. (2019) Making sense of recommendations. Journal of
Behavioral Decision Making 32(4): 403–414. DOI: 10.1002/bdm.2118
Scheffauer et al. 21
Author biographies
Rebecca Scheffauer is a PhD student at the University of Salamanca. Before coming to
Spain to join the DRU, she studied at the University of Vienna where she got a Magister’s
degree in Communication Science and a Master’s degree in Contemporary History and
Media. Her research so far has mostly focused on quantitative methods and her fields of
interest include new media such as social networking sites, how those relate to political
participation, incidental news exposure, and the influence of algorithms.
Manuel Goyanes (PhD) serves as an assistant professor at Carlos III University in Madrid
and is a former visiting fellow at both the London School of Economics (LSE) and the
University of Vienna. His research addresses the influence of journalism and new
technologies over citizens’daily lives, as well as the effects of news consumption on
citizens’political knowledge and participation. He is also interested in global inequalities
in academic participation, the systematic biases towards global South scholars, and
publication trends in Communication. His works appeared in top-tier journals such as
News Media and Society, Information, Communication and Society, Scientometrics, etc.
Homero Gil de Z´
uñiga is a Distinguished Research Professor at the University of Sal-
amanca and serves as director of the Democracy Research Unit (DRU). He is also a Media
Effects Professor at Pennsylvania State University, and Senior Research Fellow at
Universidad Diego Portales, Chile. In general, his work draws from theoretically driven
research, aiming to shed an empirical social scientific light over how social media, al-
gorithms, AI, and other technologies affect society and democracy.
22 Journalism 0(0)
Appendix
Table A1. Wave one zero-order correlations among demographic and key variables in the study.
Variables 1 2 3 4 5 6 7 8 9 10
1. Age —
2. Gender (female) 0.14*** —
3. Education 0.20*** 0.01 —
4. Income 0.16*** 0.08** 0.48*** —
5. Race (white) 0.33*** 0.12*** 0.09** 0.16*** —
6. Traditional news use 0.03 0.13*** 0.10*** 0.12*** 0.13*** —
7. SM news use 0.43*** 0.12*** 0.06 0.07* 0.23*** 0.57*** —
8. Pref. For journalists’news sel. 0.34*** 0.13*** 0.10*** 0.10*** 0.24*** 0.38*** 0.51*** —
9. Pref. For SM algorithmic news sel. 0.32*** 0.09** 0.13*** 0.12*** 0.23*** 0.41*** 0.56*** 0.82*** —
10. Trust traditional news 0.04 0.05 0.11*** 0.10*** 0.05 0.46*** 0.22*** 0.29*** 0.26*** —
11. Trust SM news 0.34*** 0.05 0.10** 0.09** 0.22*** 0.46*** 0.73*** 0.52*** 0.57*** 0.39***
Note: Sample size = 1338. Cell entries are two-tailed zero order correlation coefficients.
*p< 0.05; **p< 0.01; ***p<0 .001. Pearson coefficients based on bootstrapping to 5000 samples with confidence intervals set at 95%.
Scheffauer et al. 23
Table A2. Structural equation modeling confirmatory factor analysis of attitudes toward news
selection.
Factor/Measure
2 factor
Loading
Preference for journalists’news selection
Having stories selected for me by editors and journalists is a good way to get news 0.887
Editors and journalists know best what is most relevant to me 0.879
News selection by editors and journalists ensure that I get exposed to important news 0.870
Preference for social media algorithmic news selection
Having stories automatically selected for me on the basis of what I have consumed in
the past is a good way to get the news
0.792
News based on an algorithm does not limit my exposure to important news 0.696
Relying on a news diet based on algorithms ensures that I receive news that is most
relevant to me
0.796
Having stories automatically selected for me on the basis of what my friends have
consumed is a good way to get the news
0.866
News based on an algorithm of what my friends consume does not limit my exposure
to important news
0.870
Measures of fit
χ
2
197.8
df 19
p-value 0.000
CFI 0.953
TLI 0.931
RMSEA 0.084
SRMR 0.031
Note: Sample size = 1338. Cell entries are standardized Structural Equation Modeling coefficients for Con-
firmatory Factor Analysis at p< 0.001.
Table A3. Confirmatory factor analysis of news selection (preference for journalists vs
algorithmic news selection).
Model χ
2
(df) CFI TLI RMSEA SRMR BIC AIC
One Factor 274.9 (20)*** 0.933 0.907 0.098 0.039 44,170.32 44,045.56
Two-Factors 197.8 (19)*** 0.953 0.931 0.084 0.031 44,015.14 43,885.19
Note: CFI = comparative fit index; TLI = Tucker–Lewis index; RMSEA = root mean square error of ap-
proximation; SRMR = standardized root-mean-square residual.
24 Journalism 0(0)
View publication stats
