Misinformation

Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=witp20

Journal of Information Technology & Politics

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/witp20

Fake news self-efficacy, fake news identification, and content sharing on Facebook

Toby Hopp

To cite this article: Toby Hopp (2021): Fake news self-efficacy, fake news identification, and content sharing on Facebook, Journal of Information Technology & Politics, DOI: 10.1080/19331681.2021.1962778

To link to this article: https://doi.org/10.1080/19331681.2021.1962778

Published online: 12 Aug 2021.

Submit your article to this journal

Article views: 91

View related articles

View Crossmark data

Fake news self-efficacy, fake news identification, and content sharing on Facebook Toby Hopp

ABSTRACT This study explored the concept of self-efficacy in the context of fake news identification and sharing on Facebook. The results indicated that those scoring high on a measure of Facebook- based fake news self-efficacy (i.e., confidence in one’s ability to identify factually incorrect current events information on Facebook) performed increasingly well on a fake news identification and classification task. For its part, the ability to identify and properly classify fake news was shown to be negatively related to the self-reported likelihood of sharing of fake news on Facebook.

KEYWORDS Fake news; Facebook; social media; self-efficacy

Introduction

The presence of false, misleading, and hyper- partisan information on social media (i.e., so- called fake news) has become an increasingly salient social concern (Guess, Nagler, & Tucker, 2019). Perhaps unsurprisingly, the topic has received sub- stantial scholarly attention in recent years (e.g., Allcott & Gentzkow, 2017; Grinberg, Joseph, Friedland, Swire-Thompson, Lazer, 2019; Guess et al., 2019; Tandoc et al., 2018). Despite this increased scholarly focus, relatively little is defini- tively known about the individual-level factors that facilitate the propagation of fake news on digital platforms (Guess et al., 2019). The initial evidence that does exist suggests that fake news on social media is most likely to be consumed, believed, and, ultimately, shared by older, ideologically extreme users who often have low levels of trust in the traditional news media (e.g., Allcott & Gentzkow, 2017; Grinberg et al., 2019; Guess et al., 2019; Guess, Nyhan, & Reifler, 2018; Hopp, Ferrucci, & Vargo, 2020).

This study sought to build upon these findings by suggesting that self-efficacy beliefs (i.e., beliefs in one’s ability to affect positive task-related out- comes; Bandura, 1982) may play an important role in individual-level abilities to accurately assess fake news on Facebook. The identification of psy- chological factors underlying and/or supportive of fake news consumption and dissemination on

social media is a critical first step toward the design and implementation of intervention strategies that can effectively fight the proliferation of incorrect political information online (e.g., Pennycook & Rand, 2019). For its part, self-efficacy has extensive implications for motivation, cognition, and beha- vior, and, as such, is presumed to impact most areas of life (Carey & Forsyth, 2012). Building on prior work showing that self-efficacy beliefs are asso- ciated with information credibility motivations and evaluations in mediated environments (e.g., Hocevar, Flanagin, & Metzger, 2014; Hofstetter, Zuniga, & Dozier, 2001) and that self-efficacy beliefs may have important implications for politi- cal and politics-related beliefs and behaviors (e.g., Gil de Zúñiga & Ardévol-Abreu, 2017; Kim, Jones- Jang, & Kenski, 2020; Pingree, 2011), this study presented and employed a measure of fake news self-efficacy (FNSE). Reasoning that FNSE should have a stimulatory effect on an inter-connected body of information-relevant motivational and eva- luative processes (e.g., Bandura, 1982, 1994; Bucy & Tao, 2007), it was predicted those high in FNSE should be comparatively better suited to identify fake news on Facebook. Furthermore, this study predicted that fake news identification capabilities should, subsequently, be negatively associated with the likelihood of sharing fake news on Facebook.

Importantly, the current study focused specifi- cally on Facebook, which is the central conduit in

CONTACT Toby Hopp [email protected] College of Communication, Media and Information, Department of Advertising, Public Relations and Media Design, University of Colorado Boulder, 1211 University Ave, Boulder 80309-0401, United States

JOURNAL OF INFORMATION TECHNOLOGY & POLITICS https://doi.org/10.1080/19331681.2021.1962778

© 2021 Taylor & Francis

the dissemination of fake news and disinformation (e.g., Fourney, Racz, Ranade, Mobius, & Horvitz, 2017; Guess et al., 2019, 2018; Hopp et al., 2020). One recent study, for instance, estimated that upwards of 99% of social media-based fake news website referrals came from Facebook (Fourney et al., 2017). Facebook’s current status as a “fetid swamp of mistruths and outright lies” (Oliver, cited in Koebler, 2018) is perhaps especially concerning in light of research that has increasingly shown that Americans regularly use the platform to learn and communicate about political issues (e.g., Pew, 2016, 2019).

Literature review

Self-efficacy

Self-efficacy refers to “judgments of how well one can execute courses of action required to deal with prospective situations” (Bandura, 1982, p. 122; emphasis added). Self-efficacy beliefs have wide- reaching implications insofar as they activate and regulate motivation, affect, cognition, and behavior (Bucy & Tao, 2007). Accordingly, self-efficacy is presumed to impact many – if not most – aspects of human life (Carey & Forsyth, 2012). Self- efficacious people are able to summon and psycho- logically coordinate the internal resources neces- sary to motivate behavior, remain resilient in instances of failure, and, ultimately, achieve desired outcomes in a successful manner (Bandura, 1977, 1982, 1994, 2006). Accordingly, self-efficacy helps explain why factors such as knowledge, skill, and competency are not wholly responsible for achievement.

Self-efficacy is derived from four sources. First, self-efficacy beliefs emanate from mastery experi- ences (Bandura, 1986, 1994). Mastery experiences are prior experiences with a task or behavior that were deemed successful by the ego. The second factor that influences self-efficacy development is vicarious experience, or the observation of “others’ successful or unsuccessful performance in order to make a referential comparison and model success- ful behavior” (Hocevar et al., 2014, p. 255). A third antecedent to self-efficacy development is social persuasion (Bandura, 1994, 1986). Some authors (e.g., Hocevar et al., 2014) have conceptualized

social persuasion in terms of performance feed- back, or positive affirmation from trusted or liked others. Finally, affective, mood, and somatic factors influence self-efficacy formation and maintenance (Bandura, 1986, 1994). Negative emotional states tend to cultivate and help preserve low levels of self- efficacy while positive emotional outcomes support its growth.

As mentioned above, the effects of self-efficacy are widespread. Self-efficacy stimulates deep cogni- tive processing of information, and perhaps espe- cially so in situations marked by uncertainty, ambiguity, and challenging environmental demands (Bandura, 1994). Self-efficacy also func- tions on a motivational level such that self- efficacious individuals are likely to form and pursue specified goals and, therein, demonstrate enhanced resiliency when encountering adversity or chal- lenge (Bandura, 1982). In other words, those with domain-specific self-efficacy deficits are unlikely to be persistently motivated to behaviorally engage with an activity if they do not believe they have a high likelihood of achieving desirable outcomes (e.g., Bandura, 1986, 1994). Self-efficacy also has important implications for emotional regulation such that those high in the construct are compara- tively less likely to assign negative emotions (i.e., anxiety, fear) to target behaviors (Bandura, 1989) and, therefore, are more likely to be motivated to engage with challenging obstacles in a persistent manner (Bandura, 1994). Finally, self-efficacy informs task selection decisions. Those high in self- efficacy play an active role in the evaluation and selection of environments conducive to the obtain- ment of successful outcomes (Bandura, 1994).

Online information environments such as Facebook are complex blends of socially and i insti- tutionally produced information. The resulting pre- ponderance of cues places cognitive strain on evaluators, resulting in scenarios where users may be either unable/unwilling to engage in analytic processes (Song, Jung, & Kim, 2017) or rely heavily on surface-level social cues or cognitive heuristics (Metzger, Flanagin, & Medders, 2010). In light of such complexity, prior work on self-efficacy in online information environments has indicated that the variable may play an important role in how people select information, evaluate that infor- mation, and, ultimately, form credibility

2 T. HOPP

assessments (e.g., Khan & Idris, 2019; Hocevar et al., 2014; Yan et al., 2017).

Fake news self-efficacy

Studies on fake news identification have shown that people often have difficulty distinguishing between fake and “real” news in online contexts (e.g., Van Duyn & Collier, 2019). Research on fake news- related classification mechanisms can be linked to a larger body of work on news information cred- ibility perceptions (Tandoc et al., 2018), which has shown that people rely on factors such as institu- tional recognition, institutional trust, and author qualifications to form credibility evaluations (e.g., Hovland, Janis, & Kelley, 1953; Kiousis, 2001; Newhagen & Nass, 1989; Slater & Rouner, 1996). However, the rise of socially interactive informa- tional spaces combined with the splintering of tra- ditional news structures has helped facilitate a scenario under which individual-level traits, char- acteristics, attributes may be increasingly important factors in the processes underlying informational credibility determinations (e.g., Flanagin, Winter, & Metzger, 2018; Hocevar et al., 2014).

To that end, this work suggested that FNSE plays an important role in Facebook users’ ability to identify fake news. And, indeed, prior work has shown that self-efficacious people are perhaps bet- ter suited to make accurate news media credibility assessments (Hofstetter et al., 2001). This is because self-efficacy has been shown to promote deep cog- nitive processing and critical thinking (Bandura, 1994; Phan, 2009), factors which are supportive of the ability to assess and classify encountered infor- mation. Perhaps unsurprisingly, then, a recent study on fake news susceptibility found that the capability and motivation to engage in analytic thinking was a key factor in the ability to accurately identify and classify untrue political information (Pennycook & Rand, 2019). Moreover, self- efficacy stimulates task-specific motivation (e.g., Bandura, 1986, 1994; Hong et al., 2016; Huang, 2016). Thus, those low in FNSE are likely to avoid engaging in online credibility evaluation and clas- sification behaviors, as the activation of internal regulatory functions effectively work to shield the individual from negative affective outcomes (e.g., anxiety, frustration, disappointment, confusion,

anger; Bandura, 1994; Bandura & Locke, 2003; Schunk, 1991). For these reasons, authors like Bucy and Tao (2007) have previously described media-related forms of self-efficacy as a “motivational trait” that “activates and regulates not just behavior but also psychological responses to environmental stimuli” (p. 659).

The notion of domain specificity is an impor- tant part of self-efficacy theory (Agarwal, Sambamurthy, & Stair, 2000; Bandura, 1986, 1997, 2006; Hasan, 2006; Hopp & Gangadharbatla, 2016; Lent, Brown, & Gore, 1997; Marakas, Yi, & Johnson, 1998; Pajares, 1996). According to Bandura, self-efficacy can be understood as “a differentiated set of self-beliefs linked to distinct realms of functioning” (2006, p. 307, emphasis added). In other words, self- efficacy is a highly particularized construct (Bandura, 1986, 1997) insofar as self-efficacy assessments “should be consistent with and tai- lored to the domain of functioning and/or task under investigation” (Pajares, 1996, p. 4). The non-global nature of self-efficacy, as such, results in scenarios where one might have high levels of (for example) internal political self-efficacy (e.g., Holbert, Lambe, Dudo, & Carlton, 2007) but diminished feelings of efficacy as they pertain to effectively parsing through complex and ever- changing information environments such as Facebook (i.e., FNSE). Accordingly, the currently proposed FNSE measure is presumed to be related to but distinct from prior similar efficacy con- structs such as generalized internal political self- efficacy (e.g., Caprara, Vecchione, Capanna, & Mebane, 2009; Gil de Zúñiga et al., 2017; Holbert et al., 2007; Kim, Kim, & Lee, 2020), epistemic political efficacy (e.g., Pingree, 2011; Pingree, Brossard, & McLeod, 2014), self-efficacy beliefs related to news and information sharing online (e.g., Kim et al., 2020; Park, 2019), and forms of social media self-efficacy (e.g., Hocevar et al, 2014).1 Specifically, while general feelings of inter- nal political efficacy speak to “beliefs about one’s own competence to understand, and to participate effectively in, politics” (Niemi, Craig, & Mattei, 1991, p. 1407), FNSE refers more specifically to social media-relevant information competencies and, therein, encompasses but extends beyond purely political contexts. Likewise, while epistemic

JOURNAL OF INFORMATION TECHNOLOGY & POLITICS 3

political efficacy refers broadly to “confidence in one’s own ability to determine the truth about factual aspects of politics” (Pingree, 2011, p. 23), FNSE focuses much more narrowly on the ability to make credibility ascertainments within affor- dance-bounded social media environments. And, again, while fake news frequently involves political topics, it also commonly speaks to other topics of social concern. Self-efficacy beliefs regarding infor- mation sharing behaviors (e.g., Kim et al., 2020; Yilmaz, 2016) characteristically describe psycholo- gical mechanisms pertaining to external knowl- edge transfer activities rather than internal processing and classification functions. Finally, social media self-efficacy beliefs such as those described by Hocevar et al. (2014) speak to con- fidence in one’s ability to complete generic tasks on social (e.g., creating a post, searching for infor- mation) and do not specifically references beha- viors related to credibility assessment.

In the context of fake news, therefore, the fore- going theorization suggests that those with robust FNSE resources should be comparatively more motivated to systematically analyze the various informational cues accompanying a bit of news/ news-like information. From a mechanistic stand- point, such motivated processing is presumed include an enhanced desire to closely evaluate a news producing source (including relevant brand details) and of the language-related attributes of the message. Notably, fake news stories tend to emerge from little-known sources (Allcott & Gentzkow, 2017) and regularly contain highly emo- tional language and grammatical and syntactical elements not present in objective news content (Aldwairi & Alwahedi, 2018; Mourão & Robertson, 2019; Ylä-Anttila, Bauvois, & Pyrhönen, 2019). A Facebook user that possesses high levels of FNSE is – theoretically speaking – better equipped to make note of these informa- tional attributes. Relatedly, it is logical to presume that the mastery experiences supporting high levels of identification FNSE are derived from frequent contact with the mainstream news. Evaluation of and learning from the news equips people with the ability to contrast “real” and fake news on the basis of not only informational cues but also in terms of current events and political knowledge. For its part, political knowledge has previously been associated

with the ability to properly identify and categorize fake news, misinformation, and disinformation (Van Duyn & Collier, 2019).

Hypothesis 1: FNSE will be positively associated with the ability to accurately identify Facebook- based fake news.

While this study’s primary interest relates to the relationship between FNSE and fake news identifi- cation capabilities, it is also of critical importance to assess the implications that stem from an indivi- dual’s in/ability to identify false, misleading, and grossly biased news (i.e., answering the so-called “so what?” question). Fake news is a viral phenom- enon insofar as its deleterious social and demo- cratic potentials rest jointly upon consumption and dissemination behaviors. Stated differently, the foregoing theorization regarding the presumed association between FNSE and identification out- comes is of bounded consequence if assessment outcomes are not, themselves, linked to a decrease in fake news sharing behaviors. Because perceived veracity (however determined) is central to the concept of informational credibility and trust (e.g., Hovland et al., 1953), it stands to reason that infor- mation known or suspected to be false or otherwise untrustworthy will be of diminished social and informational value. Prior work on fake news has shown that social media users recognize that shar- ing low-quality information such as fake news is risk-laden act insofar as the act can “prompt social repercussions in the form of negative feedback from recipients” (Duffy, Tandoc, & Ling, 2020, p. 1974). Indeed, according to Lazer et al. (2017), “there is a real threat of embarrassment for sharing news that one’s peers perceive as fake” (p. 6). Because people are motivated to present a positive articulation of the self on social media platforms such as Facebook (e.g., Ellison, Heino, & Gibbs, 2006; Gil-Or, Levi-Bel, & Turel, 2015), research indicates that most social media users tend to avoid spreading information they know or suspect is incorrect (Barthel, Mitchell, & Holcomb, 2016; Chadwick, Vaccari, & O’Loughlin, 2018; Duffy et al., 2020). Translated into the context of the current study, it is therefore predicted those who are able identify fake news on Facebook will be comparatively disinclined to share such content

4 T. HOPP

with their network. In this sense, FNSE’s primary contribution to the fake news diffusion outcomes is presumed to relate to its fortification of identifica- tion capabilities.

Hypothesis 2: There will be a negative relationship between the ability to accurately classify fake news on Facebook and Facebook-based fake news sharing intentions.

Notably, research on external context-specific interventions such as platform-generated fake news disclaimers and fact-checks has shown such techniques have limited effectiveness when it comes to either correcting false beliefs or prevent- ing the spread of fake news online (e.g., Clayton et al., 2019; Flynn, Nyhan, & Reifler, 2017). Pennycook, Cannon, and Rand (2018), for exam- ple, found evidence that belief in false political information stems partly from the illusory truth effect, or the idea that prior exposure to a statement of purported fact increases the prob- ability that the statement will be judged as a factually accurate in the future. Across a series of studies, the authors found that “warning indi- viduals that the fake news headlines had been disputed by third-party fact-checkers did not abol- ish or even significantly diminish this effect” (p. 1875), suggesting that attempts to label fake news articles as false may not necessarily be an effective means of correcting untrue political beliefs (and, ultimately, inhibiting sharing beha- viors). Thorson’s (2016) concept of belief echoes speaks to a similar set of conclusions, namely that even thoroughly discredited information con- tinues to exert a shaping effect over political atti- tudes. Importantly, FNSE, as a resource drawn from the psychological interior, presents a potentially fruitful mechanism for ameliorating the spread of fake news. If shown to be a mechanistic contributor to the misidentification spread of fake news on social media, social media platforms, news organizations, and other inter- ested parties can focus on the stimulation of fake news-related efficacy beliefs rather than challenges to factual content which, in addition to having generally limited and short-term ameliorative atti- tudinal effects, are frequently subject to

politicization (e.g., Brummette, DiStasio, Vafeiadis, & Messner, 2018).

Method

The survey document was hosted on the research- er’s institutional Qualtrics server. Participant recruitment was managed by Dynata. A quota sam- ple was constructed using the demographic esti- mates of the US-based Facebook-using population. Weighted data taken from Pew Research’s (2018) Core Trends Survey was used to determine quotas. . Quotas were employed for age, education, income, biological sex, and racial/ethnic identification. All participants were required to be current Facebook users. Project approval was obtained from the researcher’s institutional review board.

After meeting the inclusion criteria, respondents provided basic information on political interest/ identity and trust in the media. Qualifying respon- dents were then tasked with reviewing 12 Facebook-based stories on the COVID-19 pan- demic, which was ongoing at the time of data col- lection (see Appendix A). Half of these stories were excerpts of published news articles taken from six large news organizations (USA Today, The Wall Street Journal, The New York Times, The Los Angeles Times, and the StarTribune) while the other six stories were fake news stories marked as false by major fact-checking organizations such as Snopes, FactCheck.org, PolitiFact, and Poynter. Article presentation order was randomized. Because prior work suggests that the use of the phrase “fake news” might prime evaluators (Van Duyn & Collier, 2019), a random half of the sample answered the FNSE questions before viewing and classifying the articles and a random half of the sample answered the FNSE questions after evaluat- ing the articles.

Each article included a descriptive headline, a short paragraph that articulated the article’s core claim, and a relevant image. Consistent with prior work on fake news credibility assessments (e.g., Fazio, 2020; Guess et al., 2020; Van Duyn & Collier, 2019), source information (i.e., originating brand identity) was provided for realism purposes . Text and imagery were taken from “authentic” fake

JOURNAL OF INFORMATION TECHNOLOGY & POLITICS 5

news websites linked to or described in the fact- checking materials. This information tended to take on a form that was hyperbolic, descriptively shal- low, and oriented toward the stimulation of nega- tive affect. Notably, fake news, as an epistemological phenomenon, has inherent meme- like qualities insofar as it tends to be affectively activating information that frequently (if not char- acteristically) is built around facially specious claims. As such, the fake news stimuli used in this study were selected – in large part – on the basis of their perceived ability to represent typical instances of COVID-19 fake news. Source text was not mate- rially altered, though in some cases it was slightly modified for succinctness. Text and imagery were embedded in mock Facebook post. All posts were associated with similar numbers of (fictional) reac- tions, comments, and shares. Articles were shown in random order to avoid ordering effects. After assessing the articles, respondents provided data on socio-demographic factors. At the conclusion of the survey, respondents were provided with a disclosure statement that identified the stories that were incorrect and provided corrective infor- mation. This statement included links to relevant fact-check articles for all fake news stories. A total of 1,016 complete responses were obtained.

Measures

Fake news self-efficacy FNSE was measured using three items that were designed to tap self-appraisals relating to identifi- cation of fake news on Facebook. Questionnaire items were developed using Bandura’s (2006) guide for instrument construction. The length and semantic qualities of the inventory items were similar to other self-efficacy measures per- taining to informational assessment and political behavior (e.g., Hoffman, Lutz, & Meckel, 2015; Kushin & Yamamoto, 2010; Newhagen, 1994; Park, 2019; Pingree, 2011). These items asked respondents to indicate how confident they were in several critical aspects of fake news evaluation, including the ability to identify news-like informa- tion that may be intentionally misleading or other- wise incorrect (i.e., disinformation; Wardle, 2017; “I am confident in my ability to spot fake news on Facebook”), the ability to distinguish between fake

news and intuitionally-produced (and normatively objective) news content (“I am confident in my ability to distinguish between legitimate and fake news on Facebook”), and the ability to identify news that may be unintentionally incorrect (i.e., misinformation; Wardle, 2017; “I am confident in my ability to identify inaccurate news content on Facebook”). Each item was on a seven-point scale were 1 = strongly disagree and 7 = strongly agree. The items were collapsed into a single composite measure (M = 4.98, SD = 1.23, skewness = −0.47, α = .93). 2, 3

Article classification For each of presented articles, respondents were asked if they thought the article was “real news” or “fake news.” A “don’t know” option was also provided. This approach was similar to the approach recently used by Van Duyn and Collier (2019). The correct classification percentage for the fake news articles was between 33.7% and 59.9% (M = 2.62, SD = 1.86, skewness = 0.20; KR-20 = .71) and between 33.8% and 50.6% for the news articles (M = 2.62, SD = 1.92, skewness = 0.14, KR-20 = .73). The total number of fake news and news articles correctly classified was summed to create a single additive measure (M = 5.24, SD = 3.00, skewness = 0.02, KR-20 = .74).

Article sharing likelihood Respondents were asked how likely (1 = not at all likely, 7 = very likely) they would be to share each of the evaluated articles with their Facebook followers. Individual composite measures of sharing likeli- hood were created for the six fake news (M = 2.45, SD = 1.65, skewness = 0.98, α = .91) and six news (M = 2.88, SD = 1.77, skewness = 0.54, α = .92) articles.

Covariates In addition to the variables of primary interest, the survey instrument also measured socio- demographic, political, and media use factors. Specifically, in terms of socio-demographic vari- ables, information was collected on age (M = 42.91, SD =14.70, skewness = 0.22), biological sex (53.6% female), race (75.8% white), years of formal education (1 = 1–5 years, 6 = 26 or more years; M = 3.24, SD = 1.01, skewness = −0.24, median = 11–

6 T. HOPP

15 years), and estimated annual income (1 = 0.00 USD – 25,000 USD, 7 = Greater than 200,000 USD; M = 3.01, SD = 1.89, skewness = 1.26, median = 50,001 USD – 75,000 USD).

In terms of political factors, the survey assessed political interest (“I’m interested in politics,” 1 = strongly disagree, 7 = strongly agree; M = 4.67, SD = 1.97, skewness = −0.59), political party iden- tification (38.1% democrat/lean democrat, 32.3% republican/lean republican, 29.6% independent/ other), ideological self-placement (i.e., conserva- tism; 1 = very liberal, 11 = very conservative; M = 6.25, SD = 2.90, skewness = −0.08), and ideological extremity (created by recoding the ideological self- placement scale such that extreme scores on either end of the political spectrum were assigned higher scores; possible value range = 1– 6; M = 3.18, SD = 1.92, skewness = 0.21). Political knowledge was measured using four items asking various questions about current political affairs. These questions asked respondents who Sonia Sotomoyer is, what party currently controls the U.S. House of Representatives, who the current Senate majority leader is, and who the current U.S. Secretary of State is. All questions were multiple choice format and included a “don’t know” option. Respondents had 20 seconds to answer each question before the survey instrument was automatically advanced. Correct answers were marked as 1 while incorrect, while “don’t know” and blank responses were assigned a zero. An aggregate measure was created by summing the total number of correct responses (M = 1.91, SD = 1.45, skewness = 0.11, KR- 20 = .71).

Finally, a number of media and platform usage variables were assessed. Political news surveillance was measured using three items that asked respon- dents to indicate how often they read the news- paper (either online or in hardcopy), watch cable news, and watch broadcast news (1 = very infre- quently, 7 = very frequently; M = 4.06, SD = 1.74, skewness = −0.12, α = .72). Media trust was mea- sured using four-items taken from Kohring and Matthes (2007; all measures were on seven-point scales where higher scores were indicative of higher levels of trust in the media). Collectively, these measures assessed the degree that participants felt the mainstream news media could be trusted to report on important issues, to provide important

facts, to accurately depict reality, and offer appro- priate evaluations of news events (M = 4.18, SD = 1.49, skewness = −0.31, α = .90). General Facebook usage intensity was measured by asking respon- dents (1) how often they log into Facebook and (2) how often they post content on Facebook (1 = very infrequently, 7 = very frequently; M = 4.84, SD = 1.65, skewness = −0.54, r = .55) while general Facebook self-efficacy was evaluated using three items tapping confidence in one’s ability to success- fully execute common Facebook tasks (all on 7-point scales where higher values were indicative of higher self-efficacy beliefs; M = 5.94, SD = 1.24, skewness = −1.36, α = .80). Finally, political Facebook usage intensity was assessed by asking respondents to indicate how often they use Facebook to learn about political issues and how often they use Facebook to post about their political views (1 = very infrequently, 7 = very frequently; M = 3.02, SD = 1.92, skewness = 0.57, r = .71).

Results

As seen in Figure 1, a fairly large percentage of respondents – around 9% – did not correctly clas- sify any of the evaluated articles. This was largely due to respondents repeatedly selecting the “don’t know” option on the article classification questions. Such inability to form an evaluative decision is a potentially important manifestation of low levels of self-efficacy (i.e., amotivation; Bandura, 1986, 1994). From a distributional standpoint, the large number of zero values suggested that the data was zero-inflated (i.e., excess zero counts are present; Zorn, 1998), indicating that the current data was ill- suited for traditional approaches to modeling count data such as Poisson or negative binomial regres- sion. As such, a Poisson Logit Hurdle (PLH) model was used. PLH models are comprised of two parts. First, a logit regression model is employed to dis- tinguish between those who did not classify any articles correctly (0) and those who classified one or more article correctly (1). Next, using the sub- sample of those who classified at least one article correctly, a zero-truncated Poisson model (with the log link function) is employed to predict the posi- tive counts. In this study, both the logit and trun- cated Poisson models controlled for the full complement of socio-demographic, political, and

JOURNAL OF INFORMATION TECHNOLOGY & POLITICS 7

media use factors. Additionally, both models con- trolled for FNSE presentation order (0 = after arti- cle evaluation, 1 = after article evaluation).

Table 1 shows the exponentiated coefficients for each portion of the PLH model. For the logit model component, the provided coefficients are odds ratios (ORs), which represent the odds of classify- ing at least one article correctly. For the zero- truncated Poisson model, the furnished coefficients are incidence rate ratios (IRRs). In both cases, para- meter estimates smaller than 1 represent a negative relationship between the independent and response variables while coefficients that are greater than 1 represent a positive relationship between the variables.

As shown, the FNSE identification factor was positively related to the ability to properly classify at least 1 of the presented articles, OR = 1.35, p = .002 (χ2[18] = 99.84, p <.001; McFadden R2 = .16). Likewise, in the zero-truncated Poisson model, the identification factor was positively associated to the number of articles correctly classified, IRR = 1.04, p = .002 (χ2[18] = 291.11, p <.001; McFadden R2 = .07). A full reporting of these models is provided in Table 1. These findings supported Hypothesis 1.

Next, the article sharing likelihood measures were assessed. The fake news and news article

sharing intentions were addressed in separate mod- els. Both models included all of the covariates used in the PLH model and the FNSE variable. While the fake news model was of theoretical primary inter- est, examination of the news sharing model allowed for exploration into the degree that fake news- related self-efficacy beliefs and identification abil- ities contribute to the sharing of institutionally produced news content. As described above, the sharing measures were substantially positively skewed. Using the process described by Ng and Cribbie (2016), a variety of plausible generalized linear model configurations were estimated and comparatively assessed using deviance residuals and information criterion indicators (see the sup- plemental materials for more information). In both cases, these analyses suggested that a Gamma regression model with the log link function per- formed best.

Table 2 shows the exponentiated coefficients produced by the Gamma regression models. These coefficients can be interpreted as multiplicative terms describing the amount of change in the cri- terion variable given a 1-unit increase on the regressor. As such, parameter estimates that are less than 1 represent a negative relationship, while values greater than 1 are indicative of a positive

Figure 1. Histogram showing number of articles correctly classified.

8 T. HOPP

relationship between the variables under considera- tion. In support of Hypothesis 2, the variable describing the number of correctly classified arti- cles was negatively associated with the likelihood of sharing the fake news articles, exp(b) = 0.95, p <.001 (χ2[19] = 206.71, p <.001; McFadden R2 = .20). Alternatively, in the news sharing model, the num- ber of correctly classified articles was positively associated with sharing intentions, exp(b) = 1.02, p =.004 (χ2[19] = 168.82, p <.001; McFadden R2 = .14). FNSE was not significantly associated with sharing likelihood in either model (fake news model: exp[b] = 1.00, p = .932; news model: exp [b] = 1.00, p = .848). 3,4,5,6,7,8

Results summary

The results provided support for the Hypotheses 1 and 2. The data indicated the presence of a robust and positive relationship between FNSE and the ability to accurately classify fake news on Facebook. Identification capabilities were shown to be negatively related to the likelihood of sharing fake news and positively related to the likelihood of sharing news. As a whole, the data provided initial and tentative indication that FNSE impacts news sharing outcomes primarily via its ability to support information identification/classification processes (in other words, a direct relationship between FNSE and sharing likelihood was not observed in the current data).

Discussion

The results of this work suggest that fake news- related self-efficacy beliefs might serve as an

Table 1. Poisson logit hurdle (PLH) model predicting article classification capabilities.

Logit Component Poisson Component OR IRR

FNSE Measure Presentation Order (1 = Before)

0.90 1.11***

Age 1.00 1.00 Sex (1 = Female) 0.82 0.99 White (1 = Yes) 0.85 1.11** Formal

Education (Years)

1.21 1.00

Income 1.07 1.02** Political Interest 1.00 1.01 Democrat-

Republican Contrast

1.19 0.89**

Democrat – Independent/ Other Contrast

0.70 0.92*

Ideological Self- Placement (Conservatism High)

0.98 0.98***

Ideological Extremity

1.07 0.99

Political Knowledge

1.60*** 1.09***

Political News Surveillance

1.22* 0.99

News Trust 1.08 1.07*** Facebook Usage

Intensity 0.94 0.99

Facebook Self- Efficacy

1.23* 1.03*

Political Facebook Usage Intensity

0.97 0.99

FNSE 1.35** 1.04** Model Likelihood

Ratio χ2(18) = 99.84*** χ2(18) = 291.11***

McFadden R2 .16 .07

Notes: * = p <.05; ** = p < .01; *** p < .001; OR = odds ratio; IRR = incidence rate ratio (exponentiated coefficients reported); FNSE = fake news self- efficacy

Table 2. Gamma regression models predicting fake news and news sharing likelihood.

Fake News Model News Model exp(b) exp(b)

FNSE Measure Presentation Order (1 = Before)

0.95 1.01

Age 0.99*** 0.99*** Sex (1 = Female) 0.89*** 0.90** White (1 = Yes) 0.90** 0.88** Formal Education (Years) 0.96* 0.98 Income 0.98 0.97** Political Interest 0.99 0.99 Democrat-Republican

Contrast 1.11* 1.02

Democrat – Independent/Other Contrast

0.98 0.97

Ideological Self- Placement (Conservatism High)

1.04*** 1.02**

Ideological Extremity 1.01 1.00 Political Knowledge 0.97* 0.97* Political News

Surveillance 1.04** 1.03***

News Trust 1.04*** 1.08*** Facebook Usage Intensity 1.04** 1.03* Facebook Self-Efficacy 0.92*** 0.92*** Political Facebook Usage

Intensity 1.13*** 1.14***

FNSE 1.00 1.00 Number of Correctly

Classified Articles 0.95*** 1.02**

Model Likelihood Ratio χ2(19) = 206.71*** χ2(19) = 168.82*** McFadden R2 .20 .14

Notes: * = p <.05; ** = p < .01; *** p < .001; Exponentiated coefficients reported; FNSE = fake news self-efficacy

JOURNAL OF INFORMATION TECHNOLOGY & POLITICS 9

important basis for future interventions design to limit the spread of fake news. Addressing fake news on a topic-by-topic or issue-by-issue basis is ineffi- cient and open to politicization processes (e.g., Brummette et al., 2018; Shin & Thorson, 2017). And, research has shown that even under optimal conditions, flagging false information has only a modest corrective effect (e.g., Clayton et al., 2019). Educational approaches centered on enhancement of critical thinking and media literacy skills are promising but are also embedded within a broader context of elongated developmental and socialization processes that are, themselves, shaped by any number of structural and cultural factors. Research across education, health, and leadership contexts has shown that it is possible to implement effective and dynamic self-efficacy interventions (e.g., Flora, Maibach, & Maccoby, 1989). In the present context, such interventions could empha- size individual capabilities to diagnose potentially problematic information, provide information allowing for the simulation of mastery experiences, and encourage various forms of analytic reasoning important to the diagnosis of fake news. FNSE- based interventions could also accompany or enhance literacy interventions (e.g., Guess et al., 2020), potentially resulting in long-lasting and additive ameliorative effects.

While the present work’s design was not amenable to a formal test of casual mediation, the pattern of observed results is consistent with a relational struc- ture wherein FNSE is indirectly associated with online information sharing (i.e., FNSE → fake news identification → fake news dissemination). This implied relational rendering might help offer some clarification on a recent study conducted by Khan and Idris (2019). Here, the authors employed a concept they labeled “perceived self-efficacy of recognizing misinformation” (PSERM). Using a sample of Indonesian Internet users, the authors failed to find a statistically significant relationship between PSREM levels and self-reported online information sharing behaviors. This finding, consid- ered in light of current results, suggests that informa- tion sharing outcomes are complex, and are perhaps centrally derived from individual identification, cate- gorization, and classification capabilities.

Notably, the vast majority of users do not share fake news on social media (Guess et al., 2018), and,

therein, most fake news content is spread by so- called “supersharers” (Grinberg et al., 2019). While recent research has increasingly identified demo- graphic and political correlates of false news shar- ing online (e.g., Grinberg et al., 2019; Guess et al., 2019; Hopp et al., 2020), this literature on fake news dissemination has not yet taken a focused look at how the confluence of personal attributes, informa- tion processing capabilities and motivations, infor- mational attributes, and contextual considerations might together result in the development of habits that support the repeated spread of fake news. To some extent, this study’s results are a first step in this regard and are suggestive of a structurally com- plex and potentially situationally idiosyncratic spe- cification of influential variables.

Future research could be taken in a number of directions. Most obviously, the results of this study suggest that FNSE should – at the very least – be controlled for in studies involving fake news iden- tification outcomes. A high-definition exploration of the demographic and psychological antecedents to FNSE is also appropriate. By better the under- standing factors that support or are otherwise asso- ciated with the development of FNSE beliefs, scholars will be better positioned to propose demo- graphic and modality-specific intervention strate- gies. Moreover, researchers have recently begun melding self-report and social media trace data (e.g., Guess et al., 2019; Hopp et al., 2020). Future work should explore the degree to which self- reported FNSE levels are associated with observable on-site behavioral outcomes pertaining to fake news. In line with the conversation above, future work should design and test various intervention self-efficacy-related strategies. While many pro- posed solutions to the so-called fake news epidemic deal with literacy efforts among youth and emer- ging/young adults, research has consistently shown that it is older Americans that are most responsible for the dissemination of fake news (Grinberg et al., 2019; Guess et al., 2019; Hopp et al., 2020). As such, it seems appropriate to develop self-efficacy inter- ventions aimed specifically at older age cohorts. Moreover, the FNSE concept articulated in this manuscript dealt only with self-efficacy beliefs related to the identification of fake news on Facebook. Because self-efficacy beliefs manifest in domain-specific ways (Bandura, 1982), future

10 T. HOPP

research should develop measures pertinent to other social media platforms. Moreover, it may be valuable to develop and assess a measure that speaks to self-efficacy beliefs related specifically to information sharing. An additional line of inquiry could involve establishing the conditions necessary for formal and explicit causal assessment of the set of relations that plausibly emanate from the ana- lyses reported in Tables 1 and 2 (i.e., FNSE → fake news identification → fake news dissemination). Finally, for the purpose of ecological validity, this study provided respondents with source brand details. While such approach is broadly in-line with prior scholarship (e.g., Fazio, 2020; Guess et al., 2020; Van Duyn & Collier, 2019), the precise degree to which organizational details influence informational evaluation cannot be meaningfully isolated or ascertained in the current study. Future research should explore the extent to which brand and content details uniquely contri- bute to Facebook users’ ability to identify and clas- sify fake news.

This study has a number of limitations. First, this work employed a convenience sample. This obviously limits the degree to which the current findings can be used to derive direct inferences about the U.S.-based Facebook using population as a whole. Second, behavioral outcomes were assessed in only in prospective terms. Future work, as dis- cussed above, should assess observable sharing beha- viors. Third, this work specifically focused on Facebook and Facebook users, and, in light of the domain specificity of the self-efficacy construct, its results may not be applicable to other social media or computing contexts. Fourth, all of the articles eval- uated in this study pertained to a specific phenom- enon (the COVID-19 pandemic) and may not be immediately transferable to situations governed by, for example, stronger or different partisan heuristics. Fourth, this study was cross-sectional, and, as such, the observed effects (and the implications that flow from these effects) cannot be treated as causal in nature. Lastly, the control variables used in this study should be considered non-exhaustive. As social scientists continue to learn about the diffusion of so-called fake news on social media, it may well be the case that variables unaccounted for in this study

are shown to play a central (and therefore confound- ing) role in the evaluation and dissemination of low- quality news-like information.

Notes

1. Broadly speaking, self-efficacy theory suggests that related- but-distal forms of self-efficacy should be predictive of more particularized forms of self-efficacy (Agarwal et al., 2000; Bandura, 1986, 1997; Hasan, 2006; Marakas et al., 1998). For instance, general levels of computing self-efficacy have been shown to predictive of self-efficacy beliefs relating to usage of a particular operating system (Agarwal et al., 2000). See the supplemental materials for further discussion on and explora- tion of this topic.

2. Additional analyses provided strong evidence that the FNSE construct is unidimensional in nature. See the supple- mental materials for more information.

3. Analyses described in the supplemental materials docu- ment show that the proposed FNSE construct is empirically distinct from other, similar self-efficacy measures such as epistemic self-efficacy and social media self-efficacy.

4. The fake news and news identification outcomes were also assessed in separate Poisson models. These models used log link functions and accounted for all of the covariates included in the PLH model. In both models, FNSE was posi- tively and statistically significantly associated with the number of articles correctly classified.

5. While the distributional forms taken on by the variables used in this work limited the ability to directly statistically probe and assess indirect effects, parallel OLS-based analyses indicated that article classification capabilities facilitated a statistically significant negative indirect effect between FNSE and fake news sharing likelihood and a statistically significant positive effect between FNSE and news sharing likelihood. These results are consistent with the contention made in this manuscript that article classification capabilities may play a facilitating role in the relationship between FNSE and article sharing outcomes. See the supplemental materials document for more information on the above-mentioned OLS analyses.

6. The results reported here were replicated in two indepen- dent samples. See the supplemental materials for more information.

7. Additional analyses conducted using an independent sample indicated that the FNSE was significantly associated with fake news identification capabilities even after accounting for the effects or related constructs such as epistemic self- efficacy and social media self-efficacy. See the supplemental materials for more details.

8. For researchers in need of very brief FNSE measures, the reported results were closely matched in terms of point esti- mate magnitude, sign, and significance by models using any single FNSE item.

JOURNAL OF INFORMATION TECHNOLOGY & POLITICS 11

Notes on contributor

Toby Hopp’s published academic work focuses on issues such as online mis-and-disinformation, political incivility on social media, political knowledge, and organizational transparency.

References

Agarwal, R., Sambamurthy, R. M., & Stair, R. M. (2000). Research report: The evolving relationship between general and specific computer self-efficacy—An empirical assess- ment. Information Systems Research, 11(4), 418–430. doi:10.1287/isre.11.4.418.11876

Aldwairi, M., & Alwahedi, A. (2018). Detecting fake news in social media networks. Procedia Computer Science, 141, 215–222. 2018. doi:10.1016/j.procs.2018.10.171

Allcott, H., & Gentzkow, M. (2017). Social media and fake news in the 2016 election. Journal of Economic Perspectives, 31(2), 211–236. doi:10.1257/jep.31.2.211

Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84(2), 191–215. doi:10.1037/0033-295X.84.2.191

Bandura, A. (1982). Self-efficacy mechanism in human agency. American Psychologist, 37(2), 122–147. doi:10.1037/0003- 066X.37.2.122

Bandura, A. (1986). Social foundations of thought and action. Englewood Cliffs, NJ: Prentice Hall.

Bandura, A. (1989). Regulation of cognitive processes through perceived self-efficacy. Developmental Psychology, 25(5), 729–735. doi:10.1037/0012-1649.25.5.729

Bandura, A. (1994). Self-efficacy. In V. S. Ramachaudran (Ed.), Encyclopedia of human behavior (pp. 71–81). New York, NY: Academic Press.

Bandura, A. (1997). Self-efficacy: The exercise of control. New York, NY: W.H. Freeman and Company.

Bandura, A. (2006). Guide for constructing self-efficacy scales. In F. Pajares & T. Urdan (Eds.), Self-efficacy beliefs of ado- lescents (pp. 307–337). Greenwich, CT: Information Age Publishing.

Bandura, A., & Locke, E. A. (2003). Negative self-efficacy and goal effects revisited. Journal of Applied Psychology, 88(1), 87–89. doi:10.1037/0021-9010.88.1.87

Barthel, M. A., Mitchell, A., & Holcomb, J. (2016). Many Americans believe fake news is sowing confusion. Pew Research. Retrieved from https://www.journalism.org/ 2016/12/15/many-americans-believe-fake-news-is-sowing- confusion/

Brummette, J., DiStasio, M., Vafeiadis, M., & Messner, M. (2018). Read all about it: The politicization of “fake news” on Twitter. Journalism & Mass Communication Quarterly, 95(2), 497–517. doi:10.1177/1077699018769906

Bucy, E. P., & Tao, -C.-C. (2007). The mediated moderation model of interactivity. Media Psychology, 9(3), 647–672. doi:10.1080/15213260701283269

Caprara, G. V., Vecchione, M., Capanna, C., & Mebane, M. (2009). Perceived political self-efficacy: Theory, assessment,

and applications. European Journal of Social Psychology, 39 (6), 1002–1020. doi:10.1002/ejsp.604

Carey, M., & Forsyth, A. (2012). Teaching tip sheet: Self-efficacy. American Psychological Association. Retrieved from http://www.apa.org/pi/aids/resources/education/self- efficacy.aspx

Chadwick, A., Vaccari, C., & O’Loughlin, B. (2018). Do tabloids poison the well of social media? Explaining demo- cratically dysfunctional news sharing. New Media & Society, 20(11), 4255–4274. doi:10.1177/1461444818769689

Clayton, K., Blair, S., Busam, J. A., Forstner, S., Glance, J., Green, G., . . . Nyhan, B. (2019). Real solutions for fake news? Measuring the effectiveness of general warnings and fact-check tags in reducing belief in false stories on social media. Political Behavior.

Duffy, A., Tandoc, E., & Ling, R. (2020). Too good to be true, too good not to share: The social utility of fake news. Information, Communication & Society. doi:10.1080/ 1369118X.2019.1623904

Ellison, N., Heino, R., & Gibbs, J. (2006). Managing impres- sions online: Self-presentation processes in the online dat- ing environment. Journal of Computer-Mediated Communication, 11(2), 415–441. doi:10.1111/j.1083- 6101.2006.00020.x

Fazio, L. (2020). Pausing to consider why a headline is true or false can help reduce the sharing of false news. The Harvard Kennedy School Misinformation Review, 1(2), 1–8.

Flanagin, A., Winter, S., & Metzger, M. J. (2018). Making sense of credibility in complex information environments: The role of message sidedness, information source, and thinking styles in credibility evaluation online. Information, Communication, & Society. doi:10.1080/ 1369118X.2018.1547411

Flora, J. A., Maibach, E. W., & Macoby, N. (1989). The role of media across four levels of health promotion intervention. Annual Review of Public Health, 10(1), 181–201. doi:10.1146/annurev.pu.10.050189.001145

Flynn, D. J., Nyhan, B., & Reifler, J. (2017). The nature and origins of misperceptions: Understanding false and unsup- ported beliefs about politics. Advances in Political Psychology, 38, 127–150. doi:10.1111/pops.12394

Fourney, A., Racz, M. Z., Ranade, G., Mobius, M., & Horvitz, E. (2017, November). Geographic and temporal trends in fake news consumption during the 2016 presiden- tial election. Presented at the ACM International Conference on Information and Knowledge Management, Singapore. From: http://erichorvitz.com/CIKM2017_fake_news_ study.pdf

Gil de Zúñiga, H., Diehl, T., & Ardévol-Abreu, A. (2017). Internal, external, and government political efficacy: Effects on news use, discussion, and political participation. Journal of Broadcasting & Electronic Media, 61(3), 574–596. doi:10.1080/08838151.2017.1344672

Gil-Or, O., Levi-Belz, Y., & Turel, O. (2015). The “Facebook- self”: Characteristics and psychological predictors of false self-presentation on Facebook. Frontiers in Psychology, 6, 1–10. doi:10.3389/fpsyg.2015.00099

12 T. HOPP

Grinberg, N., Kenneth, J., Friedland, L., Swire-Thompson, B., & Lazer, D. (2019). Fake news on Twitter during the 2016 U.S. presidential election. Science, 363(6425), 374–378. doi:10.1126/science.aau2706

Guess, A., Lerner, M., Lyons, B., Montgomery, J. M., Nyhan, B., Reifler, J., & Sircar, N. (2020). A digital media literacy intervention increases discernment between main- stream and false news in the United States and India. Proceedings of the National Academy of Sciences of the United States of America, 117(27), 15536–15545. doi:10.1073/pnas.1920498117

Guess, A., Nagler, J., & Tucker, J. (2019). Less than you think: Prevalence and predictors of fake news dissemination on Facebook. Science Advances, 5(1), 1–8. doi:10.1126/sciadv. aau4586

Guess, A., Nyhan, B., & Reifler, J. (2018). Selective exposure to misinformation: Evidence from the consumption of fake news during the 2016 U.S. presidential campaign. Retrieved from http://www.ask-force.org/web/Fundamentalists/Guess- Selective-Exposure-to-Misinformation-Evidence- Presidential-Campaign-2018.pdf

Hasan, B. (2006). Delineating the effects of general and system-specific computer self-efficacy beliefs on IS acceptance. Information & Management, 43(5), 565–571. doi:10.1016/j.im.2005.11.005

Hocevar, K. P., Flanagin, A. J., & Metzger, M. J. (2014). Social media self-efficacy and information evaluation online. Computers in Human Behavior, 39, 2014. doi:10.1016/j. chb.2014.07.020

Hoffmann, C. P., Lutz, C., & Meckel, M. (2015). Content creation on the Internet: A social cognitive perspective on the participation divide. Information, Communication & Society, 18(6), 696–716. doi:10.1080/1369118X.2014.991343

Hofstetter, C. R., Zuniga, S., & Dozier, D. M. (2001). Media self-efficacy: Validation of a new concept. Mass Communication & Society, 4(1), 61–76. doi:10.1207/ S15327825MCS0401_05

Holbert, R. L., Lambe, J. L., Dudo, A. D., & Carlton, K. A. (2007). Primacy effects of the daily show and National TV news viewing: Young viewers, political gratifications, and internal political self-efficacy. Journal of Broadcasting & Electronic Media, 51(1), 20–38. doi:10.1080/ 08838150701308002

Hong, J.-C., Hwang, M.-Y., Szeto, E., Tsai, C.-R., Kuo, Y.-C., & Hsu, W.-Y. (2016). Internet cognitive failure relevant to self-efficacy, learning interest, and satisfaction with social media learning. Computers in Human Behavior, 55, 214–222. doi:10.1016/j.chb.2015.09.010

Hopp, T., Ferrucci, P., & Vargo, C. J. (2020). Why do people share ideologically extreme, false, and misleading content on social media? A self-report and trace data–based analysis of countermedia content dissemination on Facebook and Twitter. Human Communication Research, 46(4), 357–384. doi:10.1093/hcr/hqz022

Hopp, T., & Gangadharbatla, H. (2016). Novelty effects in augmented reality advertising environments: The influence of exposure time and self-efficacy. Journal of Current Issues

& Research in Advertising, 37(2), 113–130. doi:10.1080/ 10641734.2016.1171179

Hovland, C. I., Janis, I. L., & Kelley, J. J. (1953). Communication and persuasion. New Haven, CT: Yale University Press.

Huang, H. (2016). Media use, environmental beliefs, self-efficacy, and pro-environmental behavior. Journal of Business Research, 69(6), 2206–2212. doi:10.1016/j. jbusres.2015.12.031

Khan, M. L., & Idris, I. K. (2019). Recognise misinformation and verify before sharing: A reasoned action and informa- tion literacy perspective. Behaviour & Information Technology, 38(12), 1194–1212. doi:10.1080/ 0144929X.2019.1578828

Kim, D. H., Jones-Jang, S. M., & Kenski, K. (2020). Why do people share political information on social media? Digital Journalism, 1–18. Online before print. doi:10.1080/ 21670811.2020.1827966

Kim, H., Kim, J., & Lee, D. (2020). Understanding the role of social media in political participation: Integrating political knowledge and bridging social capital from the social cog- nitive approach. International Journal of Communication, 14, 1–22.

Kiousis, S. (2001). Public trust or mistrust? Perceptions of media credibility in the information age. Mass Communication & Society, 4(4), 381–403. doi:10.1207/ S15327825MCS0404_4

Kohring, M., & Matthes, J. (2007). Trust in news media: Development and validation of a multidimensional scale. Communication Research, 34(2), 231–252. doi:10.1177/ 0093650206298071

Kushin, M. J., & Yamamoto, M. (2010). Did social media really matter? College students‘ use of online media and political decision making in the 2008 election. Mass Communication & Society, 13(5), 608–630. doi:10.1080/ 15205436.2010.516863

Lazer, D. M., Baum, M., Grinberg, N., Friedland, L., Joseph, K., Hobbs, W., & Mattsson, C. (2017). Combating fake news: An agenda for research and action. Retrieved from https:// shorensteincenter.org/wp-content/uploads/2017/05/ Combating-Fake-News-Agenda-for-Research-1.pdf

Lazer, D. M., Baum, M. A., Benkler, Y., Berinsky, A. J., Greenhill, K. M., Menczer, F. M., . . . Zittrain, J. L. (2018). The science of fake news. Science, 359(6380), 1094–1096. doi:10.1126/science.aao2998

Lent, R. W., Brown, S. D., & Gore, P. A., Jr. (1997). Discriminant and predictive validity of academic self-concept, academic self-efficacy, and mathematics-specific self-efficacy. Journal of Counseling Psychology, 44(3), 307–315. doi:10.1037/0022-0167.44.3.307

Marakas, G. M., Yi, M. Y., & Johnson, R. D. (1998). The multilevel and multifaceted character of computer self-efficacy: Toward clarification of the construct and an integrative framework for research. Information Systems Research, 9(2), 126–163. doi:10.1287/isre.9.2.126

Metzger, M., Flanagin, A. J., & Medders, R. B. (2010). Social and heuristic approaches to credibility evaluation online.

JOURNAL OF INFORMATION TECHNOLOGY & POLITICS 13

Journal of Communication, 60(3), 413–439. doi:10.1111/ j.1460-2466.2010.01488.x

Mourão, R. R., & Robertson, C. T. (2019). Fake news as discursive integration: An analysis of sites that publish false, misleading, hyperpartisan and sensational information. Journalism Studies, 20(14), 2077–2095. doi:10.1080/1461670X.2019.1566871

Newhagen, J., & Nass, C. (1989). Differential criteria for evaluating credibility of newspapers and TV news. Journalism Quarterly, 66(2), 277–284. doi:10.1177/107769908906600202

Newhagen, J. E. (1994). Self-efficacy and call-in political tele- vision show use. Communication Research, 21(3), 366–379. doi:10.1177/009365094021003007

Ng, V., & Cribbie, R. A. (2016). Using the gamma generalized linear model for modeling continuous, skewed and hetero- scedastic outcomes in psychology. Current Psychology, 36 (2), 225–235. doi:10.1007/s12144-015-9404-0

Niemi, R. G., Craig, S. C., & Mattei, F. (1991). Measuring internal political efficacy in the 1988 National Election Study. The American Political Science Review, 85(4), 1407–1413. doi:10.2307/1963953

Pajares, F. (1996, April). Assessing self-efficacy beliefs and aca- demic outcomes: The case for specificity and correspondence. Paper presented at the annual meeting of the American Educational Research Association (New York, NY).

Park, C. S. (2019). Does too much news on social media discourage news seeking? Mediating role of news efficacy between per- ceived news overload and news avoidance on social media. Social Media+ Society, 5(3), 2056305119872956.

Pennycook, G., Cannon, T. D., & Rand, D. G. (2018). Prior exposure increases perceived accuracy of fake news. Journal of Experimental Psychology. General, 147(12), 1865–1880. doi:10.1037/xge0000465

Pennycook, G., & Rand, D. G. (2019). Lazy, not biased: Susceptibility to partisan fake news is better explained by lack of reasoning than by motivated reasoning. Cognition, 188, 39–50. doi:10.1016/j.cognition.2018.06.011

Pew (2016). Liberal Democrats most likely to have learned about election from Facebook. Retrieved from http://www.pewre search.org/fact-tank/2016/05/12/liberal-democrats-most-likely- to-have-learned-about-election-from-facebook/

Pew (2018). Core trends survey. Retrieved from http://www. pewinternet.org/dataset/jan-3-10-2018-core-trends-survey/

Pew (2019). 10 facts about Americans and Facebook. Retrieved from http://www.pewresearch.org/fact-tank /2019/02/01/facts-about-americans-and-facebook/

Phan, H. P. (2009). Relations between goals, self-efficacy, critical thinking and deep processing strategies: A path analysis. Educational Psychology, 29(7), 777–799. doi:10.1080/01443410903289423

Pingree, R. J. (2011). Effects of unresolved factual disputes in the news on epistemic political efficacy. Journal of Communication, 61(1), 22–47. doi:10.1111/j.1460- 2466.2010.01525.x

Pingree, R. J., Brossard, D., & McLeod, D. M. (2014). Effects of journalistic adjudication on factual beliefs, news evalua- tions, information seeking, and epistemic political efficacy.

Mass Communication & Society, 17(5), 615–638. doi:10.1080/15205436.2013.821491

Schunk, D. H. (1991). Self-efficacy and academic motivation. Educational Psychologist, 26(3–4), 207–231.

Shin, J., & Thorson, K. (2017). Partisan selective sharing: The biased diffusion of fact-checking messages on social media. Journal of Communication, 67(2), 233–255. doi:10.1111/ jcom.12284

Slater, M., & Rouner, D. (1996). How message evaluation and source attributes may influence credibility assessment and belief change. Journalism & Mass Communication Quarterly, 73(4), 974–991. doi:10.1177/107769909607300415

Song, H., Jung, J., & Kim, Y. (2017). Perceived news overload and its cognitive and attitudinal consequences for news usage in South Korea. Journalism & Mass Communication Quarterly, 94 (4), 1172–1190. doi:10.1177/1077699016679975

Tandoc, E. C., Jr., Ling, R., Westlund, O., Duffy, A., Goh, D., & Wei, L. Z. (2018). Audiences’ acts of authentication in the age of fake news: A conceptual framework. New Media & Society, 20(8), 2745–2763. doi:10.1177/1461444817731756

Thorson, E. (2016). Belief echoes: The persistent effects of corrected misinformation. Political Communication, 33 (3), 460–480. doi:10.1080/10584609.2015.1102187

Van Duyn, E., & Collier, J. (2019). Priming and fake news: The effects of elite discourse on evaluations of news media. Mass Communication & Society, 22(1), 29–48. doi:10.1080/ 15205436.2018.1511807

Wardle, C. (2017). Fake news. It’s complicated. First Draft, 16, 1–11.

Yan, Y., Zhang, X., Zha, X., Jiang, T., Qin, L. and Li, Z. (2017). Decision quality and satisfaction: The effects of online information sources and self-efficacy. Internet Research, 27 (4), 885–904. doi:10.1108/IntR-04-2016-0089

Yilmaz, R. (2016). Knowledge sharing behaviors in e-learning community: Exploring the role of academic self-efficacy and sense of community. Computers in Human Behavior, 63, 373–382. doi:10.1016/j.chb.2016.05.055

Ylä-Anttila, T., Bauvois, G., & Pyrhönen, N. (2019). Politicization of migration in the countermedia style: A computational and qualitative analysis of populist discourse. Discourse, Context & Media, 32, 100326. doi:10.1016/j.dcm.2019.100326

Zorn, C. J. W. (1998). An analytic and empirical Examination of zero-inflated and hurdle poisson specifications. Sociological Methods & Research, 26(3), 368–400. doi:10.1177/0049124198026003004

Appendix A Stimuli

Fact-check Information: Poynter (2020). False: 5 G exposure/radiation pollution

compromises the structure and function of hemoglobin in cells and worsens COVID-19 pandemic. Retrieved from: https://www.poynter.org/?ifcn_misinformation=5g-exposure- radiation-pollution-compromises-the-structure-and-function -of-hemoglobin-in-cells-and-worsens-covid-19-pandemic

14 T. HOPP

Fact-check Information: Snopes (2020). The origins and scientific failings of the

COVID-19 “bioweapon” conspiracy theory. Retrieved from https://www.snopes.com/news/2020/04/01/covid-19-bioweapon /

Fact-check Information:

PolitiFact (2020). No, Pelosi wasn’t caught trying to add abortion funding into coronavirus bill. Retrieved from https:// www.politifact.com/factchecks/2020/mar/16/blog-posting /no-pelosi-wasnt-caught-trying-add-abortion-funding/

Fact-check Information: PolitiFact (2020). No, Bill Gates didn’t say no church ser-

vices until everyone is vaccinated. Retrieved from https:// www.politifact.com/factchecks/2020/apr/13/blog-posting/no- bill-gates-didnt-say-no-church-services-until-e/

Fact-check Information: Factcheck.org (2020). False Claim of congressional pay

raises in stimulus bill. Retrieved from https://www.factcheck. org/2020/03/false-claim-of-congressional-pay-raises-in- stimulus-bill/

Fact-check Information: Snopes (2020). Did Michigan’s governor ban the sale of

American flags? Retrieved from https://www.snopes.com/fact- check/michigan-bans-buying-flags/

della Cava, M. (2020, April 18). Latinos disproportionately dying, losing jobs because of the coronavirus: “Something has to change.” USA Today. Retrieved from https://www.usatoday.

Figure A1. Fake news story #1.

Figure A2. Fake news story #2.

Figure A3. Fake news story #3.

JOURNAL OF INFORMATION TECHNOLOGY & POLITICS 15

com/story/news/nation/2020/04/18/coronavirus-latinos- disproportionately-dying-losing-jobs/5149044002/

Paul, D., & Lazo, A. (2020, April 1). Coronavirus tests aren’t hard to find everywhere. The Wall Street Journal. Retrieved from https://www.wsj.com/articles/coronavirus-tests-arent- hard-to-find-everywhere-11585738800

Kliff, S., & Bosman, J. (2020, April 5). Official counts under- state the US coronavirus death toll. The New York Times. Retrieved from: https://www.nytimes.com/2020/04/05/us/cor onavirus-deaths-undercount.html

Willman, D. (2020, April 18). Contamination at CDC lab delayed rollout of coronavirus tests. The Washington Post.

Retrieved from: https://www.washingtonpost.com/investiga tions/contamination-at-cdc-lab-delayed-rollout-of- coronavirus-tests/2020/04/18/fd7d3824-7139-11ea-aa80- c2470c6b2034_story.html

Lazarus, D. (2020, April 2020). Coronavirus has created opportunities for, shall we say, quirky cures. The Los Angeles Times. Retrieved from: Coronavirus has created opportunities for, shall we say, quirky cures

Mulvihill, G. (2020, April 18). Lacking US coordination, states team up on when to reopen. StarTribune. Retrieved from https://www.startribune.com/lacking-us-coordination- states-team-up-on-when-to-reopen/569757572/

Figure A4. Fake news story #4.

16 T. HOPP

Figure A5. Fake news story #5.

JOURNAL OF INFORMATION TECHNOLOGY & POLITICS 17

Figure A6. Fake news story #6.

18 T. HOPP

Figure A7. News story #1.

JOURNAL OF INFORMATION TECHNOLOGY & POLITICS 19

Figure A8. News story #2.

20 T. HOPP

Figure A9. News story #3.

JOURNAL OF INFORMATION TECHNOLOGY & POLITICS 21

Figure A10. News story #4.

22 T. HOPP

Figure A11. News story #5.

JOURNAL OF INFORMATION TECHNOLOGY & POLITICS 23

Figure A12. News story #6.

24 T. HOPP