Abstract
The internet allows politicians to deliver controlled messages via campaign websites, but citizens often turn to other sources (e.g., news websites) for political information. Using the cognitive response model as a framework, this article considers how varying contexts may affect the way citizens process candidate messages. We experimentally test the interaction of media trust, media context, and message content on evaluation of and support for a candidate. A national adult sample viewed a candidate profile on either the candidate’s website or CNN.com, and the profile contained either subtle or vivid religious cues. Those high in media trust who saw the CNN profile rated the candidate highest, but only when it contained subtle religious cues. Our findings suggest that while message source and media trust can condition message effects, these factors have less influence in the presence of strong message cues.
Scholars have been attuned to the potential of new mediums to affect the way citizens engage with and process political messages since the first televised presidential debate between John F. Kennedy and Richard Nixon in 1960. The story goes that those who viewed the debate believed Kennedy to have won, while those who listened on radio believed Nixon had won. The difference was attributed to the new medium allowing people to view Kennedy’s physical appearance and demeanor. In this way, television helped usher in an era of personality-based campaigning (Druckman, 2003), demonstrating that technological innovations have the potential to change the landscape in which politicians deliver strategic messages (Agre, 2002).
It is well known that the internet has again transformed the manner in which political messages are delivered, received, and processed (Becker & Whener, 2001; Druckman, Kifer, & Parkin, 2007; Lievrouw, 2001). The internet offers a unique opportunity for politicians to deliver controlled messages to citizens, but it has also produced a fragmented media environment in which citizens turn to numerous sources to obtain political information (Bennett & Iyengar, 2008; Sunstein, 2009). More work is needed to understand how varying contexts for encountering political messages afforded by the new media environment may affect the way citizens process those messages.
A wide variety of factors can moderate how political messages affect a citizen’s evaluation of a political candidate. One important consideration is where the information is encountered, whether on a mainstream news site, a politician’s home page, a blog, or through interaction with peers via social media. Citizens’ assumptions about a source can play an important role in the way they process information, even if the content of the information remains consistent (Gunther & Liebhart, 2006). In addition, the impact of information sources likely depends on predispositional factors, such as a citizen’s trust in news media (Tsfati, 2003a). Furthermore, these factors likely have varying degrees of impact depending on the type of message encountered (Brewer, 2003; Schemer, Wirth, & Matthes, 2012). Specifically, as political information becomes more vivid or extreme, the role of message source in candidate evaluation may be attenuated.
We test the interplay of these three factors (message source, media trust, and message content) on political candidate evaluations by employing an experimental design in which we construct a fictitious congressional candidate’s profile and host it on a mock-up of either the Cable News Network (CNN) website or the candidate’s home page. In addition, we vary the religious language contained on the profile page, using a subtle and a vivid condition. These factors are fundamentally tied to political information and message processing. By testing the degree to which these factors interact to affect political evaluations, we hope to contribute to the emerging body of literature that seeks to unpack the manner in which the internet continues to shape political campaigns in the United States.
Mediated Messages: The Use of the Internet for Politics
The internet has become a central medium for both citizens and politicians to communicate about politics (Smith, 2011; Trammel, Williams, Postelnicu, & Landreville, 2006; Williams, Trammell, Postelnicu, Landreville, & Martin, 2005; Xenos & Bennett, 2007). While politicians have harnessed the internet and social media to help with organizing and fundraising, a key motive for politicians’ use of such media remains to deliver unfiltered messages to potential voters and promote their own narratives (Lilleker, Pack, & Jackson, 2010; Tedesco, 2007; Williams et al., 2005).
An important venue for delivering a clear and unfiltered message is a candidate’s personal website. Here, candidates have control over how they present themselves and the messages they deliver to potential voters (Bimber & Davis, 2003). A home page allows politicians to clearly lay out their issue stances and provide information that presents a favorable picture of their character and fitness for office. Indeed, exposure to a candidate’s home page typically results in more favorable evaluations of that candidate (Hansen & Benoit, 2005; Kaid & Postelnicu, 2005).
At the same time, mainstream news organizations are increasingly providing political information online. This includes a constant stream of breaking news, political analysis, and online candidate profiles. For example, American Broadcasting Company, CBS, CNN, Fox News, and MSNBC all had candidate profiles on their websites for the 2012 presidential election. These portrayals range from relatively sparse to comprehensive. These profiles will likely only become more extensive with each passing election. Voters can therefore access detailed candidate information on both candidates’ home pages and news websites. As candidates strive to get their own perspective across to viewers, the question becomes not just where viewers are going to get their information, but how the context in which they view this information impacts the way they receive and interpret these messages.
More Media, Less Trust
The extent to which people trust the mainstream news media to properly fulfill its role in democratic governance often plays a large role in how they process political messages encountered through news consumption. Trust in the news media has been steadily declining in recent years. Ratings of confidence in the press, a measure employed in the General Social Survey, have dropped considerably since the 1970s, though local news sources have largely been spared this decline (Gronke & Cook, 2007). Pew data have corroborated these claims, also showing a downward trend (Smith, 2011).
It has been well documented that media trust plays a moderating role for several types of media effects (Bandura, 2008; Miller & Krosnik, 2000; Tsfati, 2003a; Tsfati & Cohen, 2005). This includes the evaluation of the climate of public opinion (Tsfati, 2003a), agenda setting (Miller & Krosnik, 2000; Tsfati, 2003b), and priming (Miller & Krosnik, 2000). In all these cases, media effects tend to be more pronounced among those high in media trust compared to those low in media trust. The role of media trust in processing mainstream news media content would have limited practical significance if those who do not trust mainstream media do not attend to it, but this is not the case. Tsfati and Cappella (2003) found that while those skeptical of mass media tend to consume mainstream news less than those who are not as skeptical, they are still exposed to a substantial portion of mainstream news. Even those who are extremely skeptical tend to retrieve a lot of news from the very sources they do not trust (Tsfait & Cappella, 2003).
Media trust has been operationalized in many ways, with measures focusing on the media’s “selectivity of topics,” “selectivity of facts,” “accuracy of depictions,” and “journalistic assessment” (Kohring & Matthes, 2007); credibility (Gaziano & McGrath, 1986); and confidence in the press (Gronke & Cook, 2007). Pew’s measures of media trust, which we employ in this study, focus on whether citizens believe that the media is fair, accurate, and has proper motivations (Pew Research Center for the People and the Press, 2011).
Heuristics and Political Evaluations
According to the cognitive response model, attitudes concerning an object are the result of the tally of positive and negative thoughts ascribed to that object (Eagly & Chaiken, 1993; Fiske & Taylor, 1984). But this process does not always proceed in a straightforward calculation of all relevant pieces of information. A number of factors can bias these processes when people engage in this form of evaluation. Because people have limited capacity to process information and instead use cues, they often rely on a small subset of information to make judgments (Fiske & Taylor, 1984; see also Shrum, 2002).
The degree to which people use a more careful processing strategy or, conversely, rely more on information shortcuts to arrive at decisions is also often dependent on recipient predispositions, such as issue involvement (Eagly & Chaiken, 1993; Fiske & Taylor, 1984). Source credibility serves as a heuristic in making judgments for those who encounter a message with low personal relevance (leading to low involvement), such that high communicator credibility or likeability can result in more positive thoughts, and thus, positive judgments of the attitude object (Chaiken, 1980; Eagly & Chaiken, 1993; Fiske & Taylor, 1984). Under conditions of low involvement, source cues tend to play a larger role than message content (Chaiken, 1980).
Because media effects tend to be heightened for those high in media trust (Miller & Krosnik, 2000; Tsfati, 2003b), source heuristics will likely be especially powerful for those high in media trust. We expect that when individuals are high in media trust, they will be less skeptical of political information when it is delivered from a mainstream media outlet compared to other sources. Conversely, those low in media trust will likely be more skeptical when a message is delivered from a mainstream media outlet compared to other sources. The degree to which a citizen trusts a message source should, in turn, influence how favorably they react to a political candidate’s message. Thus, we predict:
The Medium Is Not the Only Message
These effects, however, will likely be contingent on another factor—the message itself. Research on social cognition suggests the role of predispositions is often contingent on several factors occurring concurrently (Fiske & Taylor, 1984). One factor that is likely to interact with the source of political messages is the actual content of these messages. Political campaigns inevitably flood the public with a wide array of messages. Some of these messages are explicit, while others are conveyed through subtle cues designed to activate desired reactions among voters (Mendelberg, 2001).
These cues are important because, as noted above, people are only capable of sorting through the abundance of information they are exposed to by relying on cues. By relying on mental shortcuts, citizens can make assessments without fully considering all the available information (Popkin, 1991). For this reason, information that is more salient is more likely to be noticed and applied to a judgment. Because more extreme or vivid information is more salient, it will be more accessible and have greater weight in a judgment (Fiske & Taylor, 1984; Shrum, 2002).
One prominent example of political cues is the use of religiously coded language. Religion plays a prominent role in American politics (McTague & Layman, 2009; Smidt, Kellstedt, & Guth, 2009; Wald & Calhoun-Brown, 2011). Religious cues allow citizens to assess politicians without investing the extensive time required to understand all of a politician’s issue stances and record. “Religious signals, therefore, have the ability to trump many other traditionally relevant considerations” (Domke & Coe, 2008, p. 20).
Domke and Coe (2008) assert that modern politicians have increasingly relied on religious cues as a means of securing votes. This strategy, however, has been known to backfire (McLaughlin & Wise, 2014). Domke and Coe (2008) refer to the “calculated, deliberate, and partisan use of faith” as the “God Strategy” (p. 7). In order to successfully implement the God Strategy, politicians must exhibit faith, while avoiding going so far that they alienate less religious voters. Therefore, religious language often relies on subtle religious cues rather than overt discussion of religion (Domke & Coe, 2008; Wald & Calhoun-Brown, 2011).
We therefore expect that as religious messages become more overt, they are more likely to become salient in the minds of potential voters and alter the effect of source cues. Thus, we expect that the role source cues play in the evaluation of political candidates will depend not only on individuals’ trust in media but also on the content of the message a politician delivers. Specifically, we expect that the role of source cues and media trust could be particularly important when there are not strong message cues upon which to form judgments. We therefore predict:
Method
Participants
Participants were recruited through Survey Sampling International (SSI). SSI uses multiple levels of randomization in their recruitment process and strives to achieve a nationally representative sample of U.S. adults. A total of 203 participants completed the approximately 15-min online experiment. Participants were distributed across 42 states. Ages ranged from 19 to 87 (mean = 47, SD = 17). The sample was 47.8% female, 81.8% White, 8.4% Black, 3% Asian, 5.4% Hispanic, and 4.9% other. Party identification was 33.5% Democrat, 22.2% Republican, 36.5% independent, and 3% Tea Party. Many of these demographics track closely with national averages.
Procedure
In order to test the interaction of source cues, media trust, and content extremity, we employ a 2 (candidate home page vs. CNN webpage) × 2 (vivid religious cues vs. subtle religious cues) between-subjects experimental design.
Data were collected using a web-based experiment in the spring of 2011. After providing informed consent, participants completed a pretest questionnaire in which, in addition to questions regarding political predispositions, attitudes, and participation, they were asked to rate the mass media’s fairness and accuracy and to answer questions about their own religious beliefs. Participants were then randomly assigned to view 1 of the 4 versions of a fictitious congressional candidate’s profile. Participants were assigned to a profile on the candidate’s website or a profile on a page designed to look like CNN’s website. In addition, both conditions included a subtle religious and a vivid religious language condition. Following exposure to the profile, participants completed a posttest questionnaire asking about their evaluations of the candidate and were then debriefed.
Stimulus Material
Participants were told that they were randomly selected to view the profile of a political candidate. Aside from the changes to religious language and context, the profiles used identical wording throughout. Care was taken to ensure the content reflected what is typically found on a real political candidate’s website. Alan Cole was presented as a moderate Democrat who had a history of working “across the aisle.” The issue positions were chosen to be consistent with what a moderate Democrat would be expected to hold. The CNN manipulation was created by copying the design and layout of the CNN website. The body text was exactly as it appeared on the candidate website and placed under the headline “Alan Cole, In his own words.” This was done to make it clear that the text, which was written to cast Cole in a favorable light, was provided from the candidate, and not the result of editorializing from CNN, which could suggest endorsement.
Religious language was designed to be subtle or vivid. This distinction follows Domke and Coe’s (2008) conceptualization of “two related yet distinct types of religious communication: invocations of God and invocations of faith” (p. 31). For example, in the subtle condition, he describes the United States as a “blessed nation” says he “remains optimistic due to his faith,” while in the vivid condition, he states that the United States is a “Divine nation” and that he “remains optimistic due to his faith in God.” Similar manipulations were placed throughout the text, but all other language were held constant. The stimulus materials can be viewed at http://mediatruststim.wordpress.com
Measures
Independent Variables
Media trust was measured by asking respondents to rate their agreement with the following statements adapted from Pew Research Center for the People and the Press (2011) on a 7-point scale: “most of the information provided by the news media is accurate” and “the media usually deal fairly with both sides.” The measures were combined to create media trust (Cronbach’s α = .80, M = 3.45, SD = 1.43). A median split was used to form two roughly equal-sized groups. The low group had scores ranging from 1 to 3.5 (55.3%), while the high group had scores ranging from 4 to 7 (44.7%).
Dependent Variables
Candidate traits
This measure asked respondents to rate the candidate on 7-point bipolar scales on the attributes of trustworthiness, credibility, likeability, informativeness, persuasiveness, and altruism. The answers were merged to create candidate traits (Cronbach’s α = .94, M = 4.84, SD = 1.27).
Willingness to vote
This measure asked participants whether they would be willing to vote for Cole if he were to run for Congress in their district. Responses were measured using a 100-point feeling thermometer, with 0 being very unlikely, 50 being unsure, and 100 being very likely (M = 57.59, SD = 26.59).
Control variable
Because voters’ assessments of candidates and their willingness to vote for them can be affected by whether a voter’s party affiliation matches up with that of the candidate, we controlled for respondent party identification, using dummy variables to identify Republicans and Democrats. To measure party identification, respondents were asked, “Which of the following best describes your party identification?” Those identifying as Democrats made up 33.5% of the sample, Republicans made up 22.2%, and those identifying as independents or some other party made up the remainder.
Results
We used a full-factorial analysis of covariance (ANCOVA) to analyze the data, using source, message vividness, and media trust as factors. We used the Bonferroni correction to control Type 1 error for all pairwise comparisons. In cases where findings relate to directional hypotheses, we used one-tailed tests of significance (McNeil, Newman, & Kelly, 1996).
Hypothesis 1a predicted that those high in media trust who saw the profile on a news media website would evaluate a candidate more favorably than those high in media trust who viewed the same information on the candidate’s home page. Hypothesis 1b predicted that those low in media trust would rate the candidate lower when profile information was encountered on a news media site.
We found a significant interaction between media trust and source on evaluation of candidate traits, F(1, 190) = 3.06, partial η2 = .02, p = .043. Pairwise comparisons showed those high in media trust who saw the CNN profile (M = 5.1) rated the candidate higher than those who saw the home page (M = 4.64), t(190) = 1.82, p = .035. This supports Hypothesis 1a. We did not, however, find evidence that those low in media trust who saw it on CNN (M = 4.69) rated the candidate significantly lower than those who saw it on the home page (M = 4.84), t(190) = 0.6, ns. Thus, Hypothesis 1b was not supported.
For willingness to vote, the interaction between media trust and source was not significant, F(1, 193) = 1.06, partial η2 = .01, ns. Pairwise comparisons failed to show that those high in media trust who viewed the CNN profile (M = 63.51) expressed significantly more willingness to vote for the candidate than those who saw it on the home page (M = 56.55), t(193) = 1.2, ns. They also failed to demonstrate that those low in media trust who saw the CNN profile (M = 54.54) expressed significantly less willingness to vote than those who saw it on the home page (M = 55.4), t(193) = 0.17, ns. Thus, these data support neither Hypothesis 1a nor Hypothesis 1b.
The findings above, however, must be considered in the light of the three-way interactions we predicted in Hypothesis 2, which posited that the effect of media trust and source cues on candidate evaluation would be sensitive to message extremity.
In support of this hypothesis, there was a significant three-way interaction between media trust, source, and vividness of religious cues on evaluation of candidate traits, F(1, 190) = 4.09, η2 = .02, p = .022. This was such that those high in media trust who saw the CNN profile with subtle religious cues rated the candidate higher than those who viewed the home page profile that used subtle religious cues, t(190) = 2.51, p = .006. There were no significant differences between those who saw the CNN or home page for those with low media trust in the subtle cue condition, nor for either group in the vivid religious cue conditions.
Although the three-way interaction between media trust, source, and vividness of religious cues on respondents’ willingness to vote for the candidate if they lived in his district was not significant, F(1, 193) = 2.38, η2 = .01, ns, pairwise comparisons showed a pattern similar to candidate evaluation. This was such that those high in media trust who saw the CNN profile with subtle religious cues expressed more willingness to vote for the candidate than those who viewed the candidate website profile, t(193) = 1.75, p = .04. There were no significant differences between source conditions for those low in media trust who saw the subtle cues profile, nor for either of the groups in the vivid religious cues conditions (see Table 1 for a summary of the results).
Mean Trait Evaluation Scores and Willingness to Vote by Religious Cue Vividness and Participant Media Trust Level.
Note. Estimated marginal means shown. Standard errors in parentheses.
*p < .05 (one-tailed tests).
Supplemental Analyses
In order to obtain a finer-grained picture of the way news media trust interacts with source and message vividness, we replicated the analysis above using multiple regression with media trust as a continuous variable, as opposed to a dichotomous variable. We centered the media trust variable at its mean, and the resultant variable ranged from −2.46 to 3.54. For each dependent variable, we ran two models. The first included the party identification covariates, the main effects, and all two-way interactions. The second included the three-way interaction term. We then used the Johnson-Neyman technique to probe the nature of the hypothesized interactions (Preacher, Curran, & Bauer, 2006). Because we posed directional hypotheses, we used an α of .10 to compute the Johnson-Neyman regions of significance (see Table 2 for the regression results).
Multiple Regression Analysis of the Effect of Religious Cue Vividness, Source, and Participant Media Trust on Trait Evaluation and Willingness to Vote.
Note. ID = identification; SE = standard error; SEE = standard error of the estimate.
aIndependent is the reference group. bLow vividness is the reference group. cCandidate website is the reference group.
*p < .05. **p < .01. ***p < .001 (one-tailed tests for variables relating to directional hypotheses).
For candidate evaluation, the results tracked closely with the ANCOVA analysis. Notably, in our first model, we found a significant interaction between media trust and source, β = .311, t(191) = 2.5, p = .007. The Johnson-Neyman procedure showed that near the lowest values for media trust (below −2.3), evaluations for the candidate when the profile appeared on CNN were lower than when it appeared on the candidate website, lending support to Hypothesis 1b. At media trust values above .61, increasing media trust resulted in the candidate being rated higher among those who saw the profile on the CNN page, lending support to Hypothesis 1a.
In our second model, we found a three-way interaction between vividness of religious cues, source, and media trust, β = −.492, t(190) = −1.996, p = .023. In order to probe this interaction, we ran Johnson-Neyman tests separately for those in the low vividness and high vividness conditions. Among those in the low vividness condition, we found that at media trust values below −1.74, those who viewed the profile on the candidate’s home page rated the candidate higher than those who saw the CNN profile. Conversely, at media trust levels above .19, those who saw the CNN profile rated the candidate higher. For those in the high vividness condition, the Johnson-Neyman test showed no evidence that source affected candidate evaluation differently at any level of media trust. Together, these findings lend support to Hypothesis 2.
For willingness to vote for the candidate, the results were also similar to those found with ANCOVA. In the first model, we found a significant interaction between media trust and source, β = 5.13, t(196) = −1.96, p = .025. The Johnson-Neyman technique showed media trust influenced the effect of source on willingness to vote at media trust levels above 1.5. Among those who viewed the CNN profile, as media trust increased, there was increasing willingness to vote for the candidate. The lower bound for which media trust would impact the effect of context was −5.8, well below values we observed in our data. Thus, these findings lend support to Hypothesis 1a, but only marginal support to Hypothesis 1b.
Our second model failed to find a significant three-way interaction between vividness of religious cues, source, and media trust on willingness to vote, β = −5.53, t(193) = 1.053, ns. We nonetheless ran Johnson-Neyman tests to see whether we would observe similar effects of media trust and context at both levels of message vividness. For those in the low vividness condition, as media trust increased above .85, those in the CNN condition were increasingly more likely to say they would vote for the candidate compared to those in the candidate website condition. The lower bound, where this relationship would be inverted, was −3.45, which is below values for media trust we observed. For those in the high vividness condition, there was no evidence that source affected candidate evaluation differently at any level of media trust.
Discussion
Given the rapidly changing media environment, we sought out to examine how message source may interact with other variables to affect voters’ evaluation of political content. Specifically, we tested the interplay of message source, message content, and media trust on political candidate evaluations by employing an experimental design. By testing the degree to which these factors interact to affect political evaluations, we contribute to the emerging body of literature that seeks to unpack the manner in which the internet continues to shape political campaigns in the United States.
Our study corroborates previous studies that have found media trust moderates media effects. We found this to be the case for the effect of message source on evaluation of candidate traits in the ANCOVA and regression analyses, and in some cases for willingness to vote in the regression analyses. Specifically, when those high in media trust saw the candidate on CNN, they rated him higher than those who saw it on the candidate home page. This suggests that among this group, coverage by a major news organization may signal that a candidate has attributes worthy of their attention. We also found evidence that among those with the far lowest levels of media trust, information appearing on a major media outlet can lead to lower ratings for a candidate than when it appears on that candidate’s own website.
More importantly, however, our study demonstrates that the interplay of source and media trust can be conditional on message features. When a message contains only subtle message cues, source can play an important role in evaluation depending on a person’s level of media trust. Specifically, we found that those who had high levels of media trust rated the candidate more favorably when his profile was hosted on CNN compared to a personal home page in the absence of vivid religious cues. When the message contains vivid religious cues, however, we fail to find source affects opinion, regardless of one’s level of media trust. Thus, overt cues may diminish the role of source.
This suggests that even as the manner in which political messages are constructed and disseminated is rapidly changing, the actual content of a message can still be of paramount importance. That is, even as the nature of the relationship between politicians, news organizations, and citizens changes, the content and quality of politicians’ campaigns may still be the most powerful factor in determining whether they get elected. It is easy to be so focused on how the internet has changed the nature of political campaigns that we overlook what has stayed the same—such as the importance of what politicians actually say and do.
Nevertheless, our study helps illuminate the conditions under which factors such as media trust and source cues will have important implications for political campaigns. Our results are consistent with the social cognition literature, which suggests that under conditions of low involvement, source cues may play a greater role than message content. Conversely, when people encounter unexpected content, such as extreme views or strong religious cues, these messages may override other cues, such as message source, and become the most important criteria in forming judgments.
There are some limitations to our study that we must acknowledge. Notably, our respondents were not exposed to the information in a naturalistic way. People tend to seek out and selectively expose themselves to media sources on the internet, whereas we circumvented this process in order to ensure exposure. But people are often exposed to media from sources they do not trust, and it is thus valuable to examine the impact of that exposure (Tsfati & Cappella, 2003). In addition, although some news organizations print candidates’ direct answers to questions on their sites, it is unlikely that a news website would post the same profile information as a candidate’s website. This sacrifice in ecological validity was essential, however, in order to test the effects of message source. These compromises allowed us to maintain control and establish causality.
Another limitation is that our cue strength manipulation was focused only on religious cues. Religion is a notoriously divisive political issue (Putnam & Campbell, 2010). Religious cues have been shown to be quite attractive to potential voters with religious orientations (Calfano & Djupe, 2009) but also quite a turnoff to those who are not so religious (McLaughlin & Wise, 2014). For this reason, overt religious cues may represent a specific type of vivid message and we should not assume these results would be generalizable to all other types of vivid messages (e.g., those relating to environmental concerns, international trade, etc.). The observed effects in this study, therefore, may only occur for certain types of issues, particularly those central to one’s worldview. That being said, because hot-button issues typically garner a lot of media coverage and subsequently command voter attention, this study highlights an important process that is relevant to the current political climate. Nevertheless, more research is needed to see if the patterns we observe here are evident with cues for other types of issues.
Similarly, respondents were only able to view information on one candidate. This may have particularly affected our willingness to vote measure, as voting decisions are often made in a competitive context. Future research should incorporate additional candidates to see whether the patterns hold.
Conclusion
Candidates continue to seek new ways to get their preferred messages across, and internet platforms allow news organizations the ability to present more in-depth information on candidates unconstrained by time and available column inches. At the same time, the public’s trust in mass media continues to decline while options for retrieving political information proliferate. It is therefore important to understand the ways message, message source, and people’s trust in news media work together to help form initial impressions of candidates. While the proportion of people who trust news media may be declining, we find that the media source in which candidate information is presented serves as an important cue for those who do trust news media. Our findings also highlight the importance of considering message content in addition to the platform on which it is presented. Indeed, we find evidence that in some cases, the message may trump the medium as an evaluative cue. Thus, even as the nature of the relationship between politicians, news organizations, and citizens changes, the content of a politician’s message can still be the most powerful factor in determining whether they get elected.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.