Abstract
The study assesses the extent to which the various sources of political information may influence the beliefs and values held by three groups of potential voters in the United States (Democrats, Republicans, and undecided). It interprets findings regarding the efficacy of the various sources of political information in the context of communication theories to explain the emerged group differences and suggests political marketing strategies to influence potential voters. The study findings suggest the possible powerful effects of social media relative to other sources in affecting the political views of undecided potential voters.
MANAGEMENT SLANT
Social media likely will have the greatest effect on undecided potential voters.
The various media and personal sources of information that Republican and undecided potential voters find helpful in shaping their political views may be more effective in promoting a liberal than a conservative political ideology.
The media might be more effective in changing the importance of an issue that favors one candidate over another than in attempting to change a candidate’s position on that issue.
Undecided voters might more likely develop a favorable attitude for a candidate who presents facts about his or her previous record on a certain issue than one who promises a favorable course of actions on that particular issue.
INTRODUCTION
In recent years, one has witnessed America’s increasing division with respect to issues that affect the welfare of its citizens and their descendants. Americans seem to hold extreme positions on issues such as immigration, climate change, and abortion, and such issues likely determine the outcome of midterm and presidential elections in the United States. On the basis of the results of surveys frequently reported in the mainstream media, the outcome of presidential elections is not determined by supporters of the two main parties who tend to hold the most diverse views on key issues (Mayer, 2007). Rather, it is the undecided voters who likely determine the person who occupies the White House during the next four years (Liu, Ye, Sun, et al., 2020).
Although the mainstream media have been instrumental in communicating to the general public the philosophy that underscores the position of the two major parties on important issues, they are often criticized for their biased views and inaccurate information that they present to their audiences. Many critics of mass media’s credibility today have cited research that shows the media’s disproportional support of one political party, and they have frequently used terms such as “fake news” and “propaganda media” to express their discontent (e.g., Bennet and Livingston, 2018; Tucker, Guess, Barberá, Vaccari, et al., 2018). Some communication researchers, however, are likely to argue that the effects of mass media on most registered voters who plan to vote for a presidential candidate in that year’s election are likely to be trivial, as most have already decided who they are going to vote for. Information that is consistent with the views of a political party’s supporters likely will reinforce the views of those who already favor that political party; information favoring the opposing party or candidate’s views likely will be selectively perceived, distorted, or rejected (e.g., Chaffee and Miyo, 1983; Klapper, 1960; Murphy and Auter, 2012).
Parallel to the alleged bias of mainstream media, there are also serious allegations about social media. The main social-media platforms used for dissemination of political information, such as Facebook and Google, reportedly have been allowing the postings of messages by those who favor candidates and views of one political party while censoring messages and restricting information posted by supporters of another party (Barrett and Sims, 2021). Barrett and Sims, for example, note the positive effects of social media on the outcome of the 2016 presidential election: “…social media played a central role in Trump’s 2016 victory. Brad Parscale, his former digital campaign chief, told Wired just after the election: ‘Facebook and Twitter were the reason we won this thing. Twitter for Mr. Trump, and Facebook for fundraising.’ Targeted ads on Facebook helped generate the bulk of the $250 million the first Trump campaign raised online” (Lapowsky, 2016, p. 8).
By contrast, conservative media and Internet commentators contend that conservative views are censored online (for a sample of allegations, see Barrett and Sims, 2021, p. 4). These allegations are supported by the results of two recent polls: One by the Pew Research Center reports that 90 percent of Republicans and Republican-leaning independents say that it is at least somewhat likely that social-media companies intentionally censor political viewpoints that they find objectionable; another by HuffPost/YouGov finds that the vast majority (72 percent) of Republicans think that social-media platforms have a liberal bias, compared with Democrats who think the platforms are either neutral (39 percent) or biased in favor of conservatives (16 percent; Barrett and Sims, 2021). The evidence in support of the anticonservative narrative, however, has been viewed as inadequate and unreliable, as additional data have been presented (Barrett and Sims, 2021) that supports the belief that conservatives dominate social media. That report concludes: “There are no credible studies showing that Twitter removes Tweets for ideological reasons or that Google manipulates search results to impede conservative candidates” (2021, p. 4). Even the conclusion reached by Barrett and Sims in their 2021 report, based on analyses and interpretation of existing data—i.e., that social media do not censor conservative views—is of questionable validity, in part because, as the authors admit, there was no sufficient or reliable evidence (access to Twitter data) to support their conclusions and because their study has been criticized on the grounds that it was funded by an anticonservative source (Alexander, 2021).
To summarize, the present state of knowledge on the effects of media on voters is not sufficient to describe conclusively a phenomenon that would lend itself to theoretical explanation. The contradictory allegations and the lack of sufficient evidence on the effects of media in general, and social media in particular, suggest the use of the inductive rather than deductive approach to knowledge development for the purpose of producing evidence of empirical regularities and conclusively describing the nature of the phenomenon that should be explained by theory, a research process consistent with Hunt’s (1983) notions of “logic of discovery” and “logic of justification,” respectively. To this end, research that answers empirical questions for revealing relationships between variables are essential in knowledge development and policy, even in the absence of theoretical explanation for the observed relationships (e.g., Alford, 1998). The contradictory views and limited empirical evidence regarding the effects of mainstream and social media on voters in general, and undecided voters in particular, beg the following research questions:
RQ1: Can the various sources of political information influence undecided voters’ political views? If so, what is the relative influence of these sources on undecided voters’ political cognitions and preferences?
RQ2: Are the present mainstream and social media more likely to foster conservative or liberal views among undecided voters? Are they likely to suppress such views?
RQ3: Can social media affect the development and change of opinions of political party supporters? If so, what is the relative influence of social media on the political orientations of supporters of the two main parties, in comparison with the influence of traditional media and personal sources of political information?
Answers to these research questions should be of interest to practitioners, policy makers, students of mass media, and society in general. Evidence suggesting that social-media platforms suppress liberal or conservative political views, for example, could serve as bases for repealing Section 230 of the Communications Decency Act, which protects platforms from liability related to the content of their postings. Evidence of mainstream and social media’s possible effects on party supporters and undecided voters, furthermore, could have implications for theory development. Findings that the various media likely would affect only the views of undecided voters, rather than the views of the supporters of the main parties, for example, would support the limited effects model (Klapper, 1960); and they would help extend this theory of media effects to social-media platforms. The present study has three main objectives. First, it attempts to answer the research questions posed by assessing the extent to which the various sources of political information may influence the beliefs and values of the potential voters in three groups—Democrats, Republicans, and undecided. Second, the study attempts to explain emerged relationships in the context of existing and emerging communication theories that could provide directions for theory development and further research. Third, it shows how political campaign managers could use the study findings to develop effective marketing campaigns for the three groups of potential voters. Although the authors do not expect the results of this cross-sectional exploratory study to allow causal assertions, the emerged relationships between variables in the present study’s cross-sectional data provide bases for theoretical interpretations that can help guide future research (Popper, 1959). To inform the reader and further justify the exploratory value of this study and its intended contribution, the section that follows presents a brief discussion of previous research on media effects and the existing landscape of relevant mass communication theories.
BACKGROUND
Research on Media Effects
The study of the effects of mass media on target audiences has long been of interest to practitioners, legislators, and scholars in several fields. The vast majority of the studies of media effects have examined the efficacy of a specific medium (television and print, in particular), message, and its source (for a review of such studies, see, e.g., Wilson and Sherrell, 1993). Of particular research interest has been the study of credibility in media, with much of the evidence of media effects attributed to characteristics of the source and medium (e.g., Murphy and Auter, 2012).
In contrast to studies that focus on media characteristics, another stream of research has attempted to explain the effects of media on the basis of audience characteristics. Central to this type of research has been the audience’s disposition to the media and its contents, with explanation for media’s differential effects stemming from cognitive balance theories, especially theories of selective exposure and selective perception (Murphy and Auter, 2012). The former theory maintains that people attempt to avoid cognitive dissonance by selectively attending messages that support their views on issues over those that support other views. Selective perception states that people interpret messages on the basis of their previously held beliefs, which affect the way the message will be processed (e.g., Taylor, Franke, and Bang, 2006). Because the credibility of traditional news media has been eroding in recent decades (e.g., Murphy and Auter, 2012)—in part, because media have become increasingly partisan (e.g., Watts, Domke, Shah, and Fan, 1999)—audience selective exposure and processing has been extended to media as well and likely has affected media credibility. Some research suggests that online media are thought to be more believable, accurate, in depth, and fairer than traditional media (Johnson and Kaye, 2000), especially television, with perception of its credibility affecting its use (Kiousis, 2001).
Because the credibility of traditional news media has been eroding in recent decades—in part, because media have become increasingly partisan—audience selective exposure and processing has been extended to media as well and likely has affected media credibility.
In the political arena, media can affect the audience’s perceptions of candidates, political views, and voting decisions (e.g., Baum and Groeling, 2008; McLeod, Glynn, and McDonald, 1983; Prete, 2007; Wanzenried, Smith-Howell, and Powell, 1992), but it is not clear whether, and to what extent, the content and impact of traditional media actually differ from those of Internet news reporting (Baum and Groeling, 2008). Although selective exposure and selective perception may affect party supporters, as use of online media and cable networks is a major indicator of selective exposure (Stroud, 2007), it is not clear whether the various types of media, especially social media, contribute to political polarization (Tucker et al., 2018) or how they may affect undecided voters who are not likely to hold extreme political views. The current study addresses these gaps in knowledge and interprets emerged findings within the context of mass communication theories.
Models of Communication Effects
Scientists have had a long-standing interest in understanding the effects of media on people. Early views on how the media affects audiences began to appear nearly a century ago, when publications stressed the direct and powerful impact of the media (e.g., Lasswell, 1927). This model of communication, known as the “hypodermic needle model” or “bullet theory,” places emphasis on the message; it suggests that a properly designed and delivered message could change people’s beliefs and behaviors. This view of mass media effects was reformulated during the 1940s (Lazarsfeld, Berelson, and Gauget, 1948), when it was argued that there are a host of other factors that weaken the effects of media, such as selective processes (e.g., selective attention, selective perception) and interpersonal mediating factors (e.g., “gatekeepers” of information in the two-step flow hypothesis). This “limited effects model” assumed an active audience that critically evaluated message content, developed counterarguments, or even ignored information inconsistent with the audience’s existing views. According to this view, the media more likely will be successful at reinforcing than changing existing attitudes and beliefs (Klapper, 1960).
Although the limited effects model led many researchers to believe that media messages could only reinforce existing attitudes rather than change them, several researchers felt that the effects of media were underestimated and sought to examine other potential effects of media exposure. A major outcome of these efforts was the development of the “agenda-setting” hypothesis, the assertion that the media may not be successful at telling the audience what to think but more successful at telling those exposed what to think about (Cohen, 1963). The agenda-setting hypothesis has received substantial support and is relevant in political elections, as the media can increase the salience of issues that are promoted by the candidates, especially unobtrusive issues—i.e., those for which people have no direct experience or knowledge (O’Guinn and Faber, 1991).
The view of an active audience was challenged by Krugman’s 1965 study, which argued that the media’s limited effects apply only to an audience that is highly involved with the media’s content. It asserted that, for trivial issues, people do not bother to erect cognitive defense barriers and critically evaluate message content and that, by letting their defense mechanisms down, the message “gets through.” The assertion that media effects depend on audience level of involvement was reinforced almost concurrently by the work of others who held views of either active or passive audiences. The “transactional model” (Bauer, 1964), for example, assumes an active audience, whereas the “mere exposure” theory (Zajonc, 1968) assumes a passive audience. This active-to-passive audience continuum, as articulated by that 1965 study’s concept of involvement and constructs related to the transactional and mere exposure paradigms, has become quite prominent and has influenced the development of models in the fields of marketing and mass communication.
In the field of marketing, two studies (Ray, 1973; Robertson, 1976) built on the 1965 Krugman study’s notion that media effects on passive audiences might not lead to change in attitude but result in subtle shift in cognitive structure. This view differs from the “mere exposure” theory, which suggests that familiarity with a novel stimulus creates liking for that stimulus object and that “time spent” through repeat or continuous exposure to the stimulus increases familiarity and results in attitude change. One of the two studies (Ray, 1973) contrasted the stages in the hierarchy-of-effects model under high involvement and low involvement conditions. In a rather similar vein, another study (Petty and Cacioppo, 1981) developed an elaboration likelihood model, which conceptualizes two routes to persuasion depending on the level of cognitive elaboration (a concept analogous to level of involvement).
Communication researchers also have used early theories to build additional models to explain media effects on audiences, recognizing individual differences that exist in media and message interest. The uses and gratification perspective represents such an important development, because it attempts to link motivations for media use with the effects of media exposure, moving “research attention to the individual, and away from the field’s singular obsession with the message” (O’Guinn and Faber, 1991, p. 354). This perspective has gone through several evolutionary stages over the years, and although it has suffered a barrage of criticisms, it has been widely used in different fields, including politics (Lariscy, Tinkham, and Sweetser, 2011; O’Guinn and Faber, 1991); also, it has been shown to be relevant to the audience’s media exposure for surveillance reasons in the political field—i.e., desire to acquire news and information about candidates and issues (O’Guinn and Faber, 1991).
The powerful effects of mass media on audiences have resurfaced in recent years in the form of a new perspective. Unlike the bullet theory, which assumes effects as a result of a single exposure, this perspective places emphasis on information that people receive from several sources over time, including media and other people. Proponents of this perspective argue that much of the information disseminated by mass media and personal sources that is not subject to immediate verification tends to be forgotten or absorbed into our cognitive structure. Also, over time, it is treated as factual, because this unverified information is shared by others around us who have been exposed to the same information and ideas; it is reinforced and becomes reality, as one thinks that everyone sees or “‘ought’ to see things the way we do” (McLeod and Chaffee, 1972, p. 51). This perspective seems relevant in assessing the audience’s knowledge about facts that potentially could affect attitudes and voting behavior. It appears to be particularly relevant in today’s rich digital environment, where voters use Facebook and other social network sites to exchange information about political candidates with those who hold similar views (Fernandes, Giurcanu, Bowers, and Neely, 2010; Kushin and Yamamoto, 2010).
To summarize, theories of mass communication can provide bases for understanding empirical findings of the current study. Because those who have already expressed their political views and voting intentions are the least likely to be affected by the mass media, the authors’ main interest is on the undecided or uncommitted potential voters. These voters have maintained their independence by not committing to any political ideology, and many may not register with any political party; however, they represent a very influential group that potentially could determine the outcome of presidential elections. In view of this, it is important to study their political orientations; preferences for sources of political information; and, most important, the effects of various sources of information on their present mental state and views—a term used in this study to subsume a wide variety of cognitive, affective, and conative orientations present in tricomponent models of attitude and hierarchy of effects (e.g., Ray, 1973; Robertson, 1976). Although it can be assumed that undecided voters might be in the middle of the views expressed by two major groups, they may reveal their independent position in the form of different levels of issue importance, beliefs, or attitudes from those held by the strong two-party supporters.
Undecided voters have maintained their independence by not committing to any political ideology, and many may not register with any political party; however, they represent a very influential group that potentially could determine the outcome of presidential elections.
THE STUDY
Data Collection
The data for the study were collected from a national sample of adult Americans who are U.S. citizens and registered to vote. Qualtrics, a commercial survey company, was contracted for panel participants and data collection. An online survey was developed on Qualtrics’s platform, and the company reached out to its panel participants and invited them to participate in the survey. It invited potential participants from all age groups to achieve a preset quota, a minimum of 100 participants from each of the six age groups of adult U.S. citizens registered to vote (18–29 years, 30–39 years, 40–49 years, 50–59 years, 60–69 years, and 70 years or older). Also, the quota was set to achieve equal representation of male and female participants in each age group. Participants took an average of 16.45 minutes to complete the survey and were compensated in cash or in kind for their participation in accordance with the company’s agreement with its panel members. Additionally, a prescreening question and additional quality check procedures were used by Qualtrics to ensure quality of collected data (complete responses, no rushed responses, etc.). Responses that did not meet quality screening or checks were eliminated from consideration. Actual age of participants was provided by Qualtrics from their panel database. Qualtrics charged their normal fee for the use of their panel and for their data collection services. A total of 680 responses were obtained, of which 629 had complete responses to questions in this study and were used for analyses. Participants represented almost all states and Washington, D.C., except South Dakota, West Virginia, and Wyoming. Although the sample cannot be considered to be representative of the entire U.S. population, it provided sufficient variance in responses to be useful for examining relationships between variables. The data were collected from March 25 to March 28, 2020.
Party loyalty was measured by asking participants two questions. The first question asked them to indicate whether their views on proposed legislation and policies are more in favor of liberals or conservatives on a 5-point scale (1 = much more in favor of liberals, 2 = more in favor of liberals, 3 = neither or no opinion, 4 = more in favor of conservatives, and 5 = much more in favor of conservatives). The second question asked them to indicate how they likely would vote, “if the presidential election were to be held today,” on a 5-point scale (1 = definitely will vote Democrat, 2 = likely will vote Democrat, 3 = neither/not sure, 4 = likely will vote Republican, and 5 = definitely will vote Republican). If a participant answered “more (or much more) in favor of liberals” and “definitely (or likely) will vote Democrat,” the person was considered a loyal Democrat. Similarly, if a participant answered “more (or much more) in favor of conservatives” and “likely (or definitely) will vote Republican,” he or she was considered a loyal Republican. On the basis of this classification, the sample of 629 registered voters contained 207 loyal Democrats and 157 loyal Republicans. The remaining 265 participants were those who indicated that they are neither liberal nor conservative or had not decided whether to vote Democrat, Republican, or Independent. The authors label these potential voters as “undecided,” in line with previous research that defines these individuals as “swing voters”—i.e., those “who could go either way” or are “not so solidly committed to one candidate or the other” and “those whose final allegiance is in some doubt all the way up until Election Day” (Mayer, 2007, p. 359). The authors present demographic profiles of the participants as loyal Democrats, loyal Republicans, and undecided potential voters (See Table 1). The percentage of participants in the “retired or unemployed” category was high across all three groups, indicating the possible impact of the COVID-19 pandemic on the employment status of the present study’s participants.
Analysis and Results
The initial phase of analysis involved the assessment of the participants’ state of mind on topics that were covered frequently in the mainstream media in the early months of 2020. The authors assessed their perceptions of issues, political views, knowledge about current public affairs, and perceptions of the role of various sources of political information in shaping their political views. Responses were measured on scales and were cross-tabulated or averaged across the three groups of participants.
Issue Importance. First, the authors wanted to know how important several current issues were to the participants. The authors asked them to indicate the importance of the following 10 factors on a 4-point scale (1 = not important, 4 = very important) in their decision to vote for a presidential candidate: environment and climate control, immigration, economy, abortion, the Second Amendment (or gun control), foreign policy, health care, crime, terrorism, and the candidate’s credibility or likability. The authors present the percentage of the survey participants who consider each of the issues “very important” in each of the three groups, along with chi-square values and levels of statistical significance of differences in responses (See Table 2).
Opinions. Next, the authors asked the study participants to say how they felt about the current economic, political, social, and global state of affairs. The authors presented them with eight statements relevant to key issues, asking them to indicate the extent to which they disagree or agree with each statement on a 5-point scale (1 = strongly disagree, 5 = strongly agree). The authors summed their responses across all eight statements to form a measure of political ideology (reverse coding where appropriate). Reliability (alpha) of the composite scale was 0.83. The range for this scale was 8–40 and had a mean value of 23.96 (SD = 7.07), with smaller scale values indicating conservative political ideology and larger values indicating liberal ideology. The scale was negatively related to the statement, “I am generally satisfied with my life” (r = –.16, p < .001), a finding consistent with the results of previous studies over decades that show that conservatives are happier than liberals (e.g., Taylor, 2008), providing concurrent validity support. The average scale score for loyal Democrats was 29.08 (SD = 4.43), for loyal Republicans it was 16.07 (SD = 5.82), and for the undecided participants it was 24.64 (SD = 5.03). All between-groups differences were statistically significant (p < 0.001), providing strong evidence of face validity. The authors present the percentage of respondents in each group who “somewhat agree” or “strongly agree” with each statement (See Table 3). Loyal Democrats and loyal Republicans differed in terms of their political ideology as reflected in their opinions on current economic, political, social, and global state of affairs (See Table 3).
Public Affairs Knowledge. Participants’ knowledge about the current economic, political, social, and global state of affairs related to the main current issues was measured by asking participants to indicate whether eight statements about such affairs were “true” or “false”, or “not sure/don’t know.” Correct responses were coded as 1, and “not sure/don’t know” and incorrect responses were coded as 0. A composite measure of knowledge of public affairs related to these issues was computed by summing responses to all correct responses to form an index from 0 to 8 points, with a mean score of 2.80 (SD = 1.70). The authors present the eight statements, as well as the correct responses given by participants in each of the three groups (See Table 4). Participants could not answer most of the questions correctly, with loyal Democrats and undecided participants scoring the lowest (M = 2.51, SD = 1.5; and M = 2.48, SD = 1.7, respectively), compared with loyal Republicans (M = 3.74, SD = 1.61), who scored significantly higher than the other two groups of registered voters (both p < 0.001) (See Table 4). Although the authors cannot assume that the responses given by the participants are representative of the general U.S. population, the finding regarding the low level of knowledge of current state of public affairs is consistent with those of earlier studies (e.g., Delli and Keeter, 1996). Also, because these issues are not necessarily related, the composite index of these responses is not expected to exhibit internal consistency (e.g., Herbert and Cohen, 1996).
Perhaps of greater interest are the differences in public affairs knowledge among participants in the three groups of registered voters. These results suggest that, although registered voters, in general, are very concerned about issues that affect them and have strong political orientations, a large portion of them are likely to cast their vote having relatively little knowledge about the issues promoted by the presidential candidates.
Value of Political Information Sources. Next, the authors wanted to find out how much influence the various sources of information have on each group of voters. The value of political sources of information was measured by asking participants to indicate the extent to which nine types of sources of political information were helpful to them, “…whether the place or individuals have not been helpful, somewhat helpful, helpful, or very helpful in shaping your present political views” on a 4-point scale (1 = not helpful at all, 4 = very helpful). The sources were: parents or close relatives, close friends or “significant others,” other peers or coworkers, newspapers or magazines, television or radio programs, social media (Facebook, Instagram, Twitter, WhatsApp, etc.), information shown on the Internet, formal or informal meetings, and credibility or likability of politicians. The authors present the responses given by participants in each of the three groups (See Table 5).
It is interesting to note the lack of significant differences among the three groups regarding the value of information they receive from individuals in their immediate social environment, as well as the significant differences in the value of the information they receive from impersonal sources. Sources such as one’s parents or close relatives, close friends or “significant others,” other peers or coworkers, and credibility or likability of politicians are equally valued by all three groups of participants (See Table 5). There are, however, differences in the percentages of loyal Democrats, loyal Republicans, and undecided potential voters who find the traditional media (print and broadcast) and online sources (social media and Internet) helpful.
The authors also analyzed responses across three age groups— younger (18–39 years), middle-aged (40–59 years), and older (60 years and over)—as they expected an age divide in the use of the various media sources (Mitchell, Jurkowitz, Oliphant, and Shearer, 2020). This analysis reveals that the value of information that the participants receive from personal sources, online media, and formal or informal meetings declines with age, whereas the value that participants receive from the remaining sources remains constant with increasing age.
Collectively, these findings suggest that the traditional sources of information are the most preferred sources of political information among loyal Democrats, as major print media and most major television networks largely report negative information about Republicans that reinforces liberal views (e.g., Patterson, 2017). Social media likely have the greatest effect on undecided potential voters’ opinions about political affairs, suggesting that this group may see these sources as more effectual than traditional media. Loyal Republicans and older voters are the most “disadvantaged/deprived” groups in getting valuable information from various forms of online media. To further examine the possible effects of the value of political information sources on the three groups of voters, the authors correlated each of the nine types of sources of political information with the summated measures of public affairs knowledge and political ideology (See Table 6).
Effects of Value of Political Information Sources. Correlation analyses showed that, among loyal Democrats, there were no significant relationships between participants’ knowledge about public affairs and the nine sources of political information these voters consider helpful in having shaped their present political views. Loyal Democrats appear to get very little factual information from sources they rely on (See Table 6). Among loyal Republicans, the public affairs knowledge measure was negatively correlated with the value of political information received from parents or close relatives (r = –0.170, p < 0.05), other peers or coworkers (r = –0.174, p < 0.05), and social media (r = –0.171, p < 0.05). These findings suggest that those Republicans who rely on these sources are more likely to receive incorrect rather than accurate information about issues that may affect their political views. Among the undecided potential voters, the measure of knowledge about public affairs also was negatively related to the value of information that these potential voters receive from their parents or close relatives (r = –0.155, p < 0.05). Collectively, these findings suggest that there is very little factual information about public affairs issues that is disseminated through the sources of political information that loyal Democrats rely on, whereas Republicans are more prone to getting incorrect information from parents or close relatives, peers or coworkers, and social media (e.g., Facebook, Instagram, Twitter, WhatsApp); and the various sources of political information do not help increase the undecided voters’ public affairs knowledge. One could also make alternative interpretations of the inverse relationships between the value of the various information sources and political knowledge. The lower a person’s level of knowledge, for example, the more he or she appreciates social media and others’ opinions on politics; or the higher their political knowledge, the less they value political information disseminated through personal sources and social media.
The study’s main research questions concern the effects of the value of the sources of political information, especially social media, on voters’ political views, particularly those of undecided voters. The authors reasoned that, if the various sources did not disseminate, suppress, or censor information that favored a party or its candidate, these sources would have no significant impact on voters’ political views and preferences. By contrast, if they favored liberal or conservative views, the relationships between the value of information sources and voters’ political ideology would be positive or negative, respectively. Correlation analyses among loyal Democrats suggest that the nine sources of political information may not have a significant impact on the development of the political ideology of these potential voters to a liberal or conservative direction (See Table 6). Among loyal Republicans, however, seven out of nine sources of information were positively related to the current study’s ideology measure, suggesting that information disseminated by these sources may foster the development of these potential voters’ political ideology to a liberal direction. The strongest positive correlations between source helpfulness and this group’s tendency to favor a liberal political ideology emerged for social media (e.g., Facebook, Instagram, Twitter, WhatsApp) (r = 0.475, p < 0.001) and the Internet (r = 0.408, p < 0.001). A strong positive relationship was also found for other sources, including peers or coworkers (r = 0.353, p < 0.001), newspapers and magazines (r = 0.329, p < 0.001), formal and informal meetings (r = 0.342, p < 0.001), television and radio programs (r = 0.213, p < 0.01), and parents or close relatives (r = 0.276, p < 0.001).
Similarly, for undecided potential voters, all nine sources were also positively and significantly correlated with the political ideology measure, with correlations ranging between 0.22 and 0.36 (all significant at p < 0.001). The strongest correlate of this measure was social media (r = 0.365), followed by formal and informal meetings (r = 0.320), and information shown on the Internet (r = 0.276). The correlations of the political ideology measure with parents or close relatives, peers or coworkers, and television or radio programs were 0.26; and the correlations for the remaining sources were 0.22. Although causality cannot be established from correlational data (Popper, 1959), these relationships suggest that the various sources of political information may foster the development of a liberal political ideology among loyal Republicans, most strongly influenced by information available on social media and the Internet. Similarly, personal and non-personal sources, including mass media, social media, and the Internet, may help shape liberal political views of undecided potential voters.
Multiple regression analyses also were performed to assess the relative influence of the sources of political information and their possible overall joint effects on political affairs knowledge and ideology of the three groups of registered voters. Because six recent surveys conducted by Pew Research Center revealed that U.S. adults’ reliance on various information sources, especially social media, for political information could be affected by their age and income (Mitchell et al., 2020), these two variables were included in the regression models. The authors present the results of these analyses (See Table 6). In interpreting regression results, the reader should keep in mind that the various sources are expected to be intercorrelated, as suggested by theories (e.g., between media and personal sources; O’Guinn and Faber, 1991) and recent research (e.g., relationships between mainstream and social media; Tucker et al., 2018), which could have affected the size and stability of regression coefficients. This analysis suggests the significant impact of social media on the political orientations of undecided registered voters. Potential voters in this group become increasingly knowledgeable and conservative with age and income, whereas the strong links of online media use to a liberal political ideology among Republican registered voters are marginalized in the presence of the remaining factors used in the regression analyses. The R2 values underscore the likely significant impact of the various sources of political information on the development of political views among Republican and undecided voters.
To further validate these findings, the authors performed bootstrap analyses using 2,000 samples for each regression (See Table 6). For all significant parameters reported in Table 6, bootstrap analyses confirmed that the 95 percent confidence interval did not include zero and did not change the direction (positive or negative) of the parameter. This provides evidence of the robustness of the parameter estimates obtained through these regression analyses (IBM, n.d.)
Collectively, these findings suggest that the various personal and media sources of political information may provide little factual information, and possibly disinformation, about public affairs and political issues. Also, they may be more instrumental in fostering liberal rather than conservative political ideologies among Republicans, whereas social media alone may have similar effects on undecided or Independent voters.
Changes in Political Ideology. Because the summated bipolar measure of political ideological orientation was not correlated with any of the information sources that the loyal Democrats find helpful, the authors considered two possibilities for these results: The various sources were merely helpful in reinforcing existing attitudes rather than changing them; and in light of evidence that political ideology is not stable even among political leaders (Mixon, Sankaran, and Upadhyaya, 2019), the information sources may be effective in shifting this group’s political ideology to a stronger position for some of its members and to a more conservative position for others, leading to insignificant source effects for loyal Democrats. To address these possibilities, the authors developed two measures of change in political ideology: one showing change to a liberal ideology and another showing change to a conservative position. Because the authors did not want to frame the question in the context of any president’s performance, they asked participants to indicate whether their political views had changed in the past five to ten years, “…whether they have become more liberal, more conservative, or have changed very little or have not changed.” The authors considered “no or little change” responses as “no effect” (assigning 0) and used two dummy variables to measure change: one to measure change from no effect (0) to a more liberal position (1) and another to measure change from no effect (0) to a more conservative position (1). Thus, although the political ideology measure serves to locate each political group on a bipolar continuum in relation to the other two at the time of the survey, the measures of change in political ideology tap shifts in ideology relative to one’s earlier position during the recent past. The authors then correlated these two variables with each of the nine sources of information, separately for each of the three groups of participants (See Table 7).
Among loyal Democrats, the significant positive relationships between the current study’s measure of change to a more liberal political ideology and perceived helpfulness of peers or coworkers (r = 0.149, p < 0.05), social media (r = 0.188, p < 0. 01), and politicians (r = 0.164, p < 0.05) suggest that either these sources have been effective in helping strengthen the political views of some supporters of the Democratic party, or these party supporters are selectively exposed to these sources of political information (See Table 7). Change in the political orientation of this group to a more conservative position may have been the result of information in social media (r = 0.433, p < 0.001) and formal or informal meetings (r = 0.35, p < 0.001). By contrast, the conservative ideology of loyal Republicans may not have been affected by the sources of political information in recent years, with the only possible influence of broadcast media (television or radio programs) showing change to a more liberal position (r = 0.182, p < 0.01).
Finally, for the undecided voters, the number of significant correlations of political information sources with changes to a more liberal and more conservative positions suggest that this group, which is the one most likely to determine the outcome of presidential elections, is the one most likely to be influenced by the various sources of political information before the election. Potential voters in this group may be receptive to information that promotes a liberal ideology through a variety of sources, including peers or coworkers (r = 0.22, p < 0.01), print media (r = 0.216, p < 0.01), television or radio programs (r = 0.283, p < 0.001), social media (r = 0.273, p < 0.001), and formal or informal meetings (r = 0.172, p < 0.01). They are most receptive to information that promotes conservative views coming from peers or coworkers (r = 0.197, p < 0. 01), social media (r = 0.138, p < 0.138), and formal or informal meetings (r = 0.172, p < 0.01).
The results of partial correlation analysis (shown in parentheses; see Table 7) nullify the possible effects of political information sources on some Democrats’ recent shift to a more liberal ideology but confirm the possible effects of social media and formal or informal meetings on those who became more conservative in the recent past. They also marginalize the possible role of broadcast media in strengthening conservative values among Republicans. Among the undecided or Independents, partial correlation analysis validates only the role that social media may play in shifting voter political views to a liberal direction and only the emerged relationship between change to a more conservative ideology and certain types of personal sources (other peers or coworkers). Finally, this analysis suggests that Democrats who value the Internet as a source of political information less likely would change their political views to a more conservative direction (r = –0.319, p <. 01), whereas undecided voters who rely on this source are also less likely to change their political views to a more liberal direction (r = –0.171, p <. 05).
Changes in Beliefs. The authors also considered the effects of mass media on voters’ beliefs. Because an important function of the media is to inform people, the authors reasoned that media effects could be in the form of changes in beliefs about the main issues that affect them. Therefore, the authors measured the extent to which the participants had noticed changes in political, cultural, economic, and social environments by asking them to indicate whether in the previous five to 10 years they had become aware of any major change(s) in nine factors that had been a reason for their present political views and preferences. The factors were: environment and climate control, immigration, economy, proposed changes in abortion laws, proposed changes in gun control laws, foreign policy, health care legislation, crime, and terrorism. Responses were measured on a 3-point scale (1 = change for the better, 2 = change for the worse, 3 = no change). A measure of perceived changes for the better or the worse was constructed by summing responses to a change (coded as 1, else coded as 0) to form an index of 0 to 9 points. The authors correlated responses to each of the four types of media sources with the index of perceived changes in factors that pertain to the main political issues, separately for each of the three groups of study participants.
For loyal Democrats, the index of perceived changes correlated significantly with newspapers and magazines (r = 0.227, p < 0.01) and television and radio programs (r = 0.145, p < 0.05), suggesting that loyal Democrats are more sensitive to information in traditional media. For loyal Republicans, the index of perceived change correlated significantly only with newspapers and magazines (r = 0.251, p < 0.01). Finally, for the undecided potential voters, the index of perceived change correlated significantly with newspapers and magazines (r = 0.232, p < 0.001), television and radio programs (r = 0.251, p < 0.001), social media (e.g., Facebook, Instagram, Twitter, WhatsApp) (r = 0.151, p < 0.05), and information shown on the Internet (r = 0. 174, p < 0.01). These findings suggest that print media and television and radio programs are the most effective vehicles for informing the public about changes in important issues.
Agenda-Setting Effects. The last phase of the analysis was to determine the extent to which the different types of media influence voters’ perceptions of the importance of various issues that people consider when they evaluate presidential candidates. Although previous research has supported the agenda-setting function of traditional media (e.g., O’Guinn and Faber, 1991), relatively less is known about the extent to which the various forms of more recent technologies and constantly evolving forms of the Internet and social media shape people’s views about the importance of various issues. The authors assessed the effects of these two forms of electronic media on the measures of issue importance and included also for comparison purposes the two types of traditional media. Because responses to the measures of issue importance and source helpfulness may have been affected by social desirability—i.e., “the tendency on the part of individuals to present themselves in a favorable light, regardless of their true feelings about an issue or topic” (Podsakoff, Mackenzie, Lee, and Podsakoff, 2003, p. 881)—the authors used the short version of the Marlowe–Crowne Social Desirability Scale (Strahan and Gerbasi, 1972) to partial out any such effects (See Table 8).
Consistent with previous findings (e.g., O’Guinn and Faber, 1991), the traditional mass media (print and broadcast) affect potential voters by suggesting the wide variety of issues they should think about, especially issues related to the environment and foreign policy (See Table 8). Messages that people receive or share online tend to focus on issues concerning the environment, the Second Amendment, crime, and—to a lesser extent— abortion. Messages received on the Internet appear to focus on gun control and crime. When the effects of traditional media are taken into account, however, social media and the Internet appear to have no impact on issue importance.
DISCUSSION
Before discussing the findings of this study, the reader should keep in mind several limitations of the present investigation. First, as in any cross-sectional study, the emerged relationships between variables do not suggest causality but merely associations between variables. Second, as variables are measured at the same time, they are subject to common method bias (Podsakoff et al., 2003), despite authors’ awareness and efforts to reduce them (e.g., using stringent data collection methods). Third, although the sample used produced adequate variance in the variables of interest, some of the emerged relationships might not hold in samples representative of the U.S. population, because it might still be largely idiosyncratic because of self-selection in panel participation. Fourth, the list of issues and questions tapping public affairs knowledge were not as elaborate and relevant to today’s environment as they could have been, in part because the instrument was developed before the COVID-19 pandemic and its publicized consequences. Finally, measures of the value of media were crude, combining social-media source, exposure, and benefits in a single measure (i.e., “helpful”) that does not take into account media impact beyond participants’ awareness (e.g., long-term familiarity with persons or issues that creates liking; e.g., Zajonc, 1968). These limitations notwithstanding, some of the findings of this exploratory study are of interest and warrant further investigation and explication.
The emerged results help answer the study’s main research questions. First, the various sources of political information are not related to the undecided voters’ political knowledge, but they are positively associated with liberal political ideology. The latter finding suggests that the various sources of political information that undecided voters find helpful in forming their political views may be dispensers of partisan information or misinformation. The strong positive correlations between these voters’ liberal political orientation and social media, as well as formal and informal meetings (e.g., classes, hometown meetings) relative to other sources, suggest that these sources may dispense or censor partisan information, and they may provide partisan disinformation—a term used to include “‘fake news,’ rumors, deliberately factually incorrect information, inadvertently factually incorrect information, politically slanted information, and ‘hyperpartisan’ news” (Tucker et al., 2018, p. 2)—that lead to the formation of liberal political ideology.
These findings also help answer the research question about the possible effects of mainstream and social media on undecided voters. These sources of information that undecided voters consider helpful in shaping their political views more likely will promote a liberal than a conservative ideology, although the mechanism(s) of development of such an orientation cannot be determined from the cross-sectional data. Additionally, the findings about undecided voters also apply to Republican party supporters, and they help answer the research question regarding the effects of social media on the development of political orientations of political party supporters and their relative influence, in comparison with the influence of traditional media and personal sources of political information. Specifically, the findings suggest that Republican party supporters likely receive inaccurate information from social media and personal sources, and the various sources examined may play a role in fostering liberal political views, even among conservative voters. The various sources appear, however, to provide little factual information about public affairs and have no significant influence on the development of liberal views among Democrats.
Although the authors cannot infer causality from these findings, the emerged positive relationships between liberal political ideology and the several sources of political information are of concern, as these relationships apply to both Republicans and undecided or Independent potential voters. These relationships deserve further exploration and explication, as well as theoretical explanations. The relationships between mainstream media and liberal political orientation, for example, may reflect disproportional audience exposure to a larger number of media sources that promote liberal political views. The relationship between social media and liberal ideology may, in part, reflect user characteristics, besides age and income, and the relationship between personal sources and liberal ideology may reflect greater willingness of liberals, or lower likelihood of conservatives, to express or promote their political views during interpersonal encounters. To the extent that partisan information, disinformation, or censoring is present in social media, and these media directly or indirectly (through other sources of political information) are contributing to political polarization that undermines democracy, they would provide grounds for repealing Section 230.
Also, the relationship between the perceived value of social media as a source of political information and changes in the beliefs of many Democrats to both liberal and conservative positions in part contradicts the selective exposure theory. It underscores the power of social media not only in strengthening liberal ideological views but also in shifting liberal ideology to the opposite direction. Thus, social media may promote political polarization and disinformation or misperception, both of which are reciprocally linked and may undermine democracy (Tucker et al., 2018). Although selective exposure may explain these relationships, especially among party supporters who tend to be selectively exposed to online media and cable networks (Stroud, 2007), the possibility of causal influence of the Internet and social media on potential voters (undecided voters in particular) would violate Section 230 of Communications Decency Act, which gives social platforms such as Facebook and Google immunity from liability for partisan disinformation.
It is well established that, although Americans are largely misinformed on political issues (Delli and Keeter, 1996), they tend to express their opinions on important issues and respond to changing political and social environments (Gilens and Murakawa, 2002). Although the relative low level of the average voter’s knowledge was expected to emerge in the current study, of greater interest are the differences in knowledge among the three groups of registered voters. The “helpfulness” of the various sources of political information does not appear to increase public affairs knowledge among voters in all three groups. Perhaps this is due to misinformation or contradictory information disseminated through the various sources and voters’ inability to validate information they receive from various sources. At least two possible mechanisms may account for this phenomenon that can be explained by theories of mass communication.
A mechanism that people use to compensate for their ignorance and convince themselves and others that they are “rational” in their political choices is by using the opinions of those they perceive in the mass media to be experts in politics, such as politicians, journalists, and leaders in various organizations or institutions (Gilens and Murakawa, 2002). Such a strategy places fewer demands on their ability to make sound decisions, because it helps them to “be knowledgeable in their reasoning about political choices without necessarily possessing a large body of knowledge about politics” (Sniderman, Brody, and Tetlock, 1991, p. 19). This view is consistent with the original two-step flow hypothesis, and its revision to include “multistep” flow of information and influence from the media to gatekeepers and larger audiences (e.g., Weimann, 1982); and it could explain the ineffectiveness of various mass media in communicating accurate information and the greater impact of social media on the development of liberal political views among undecided registered voters.
A second mechanism by which many uninformed voters may develop and hold on to their beliefs is suggested by McLeod and Chaffee’s (1972) power effects model regarding the role of media in constructing “social reality” rather than objective reality. According to this view, media have powerful effects on people’s political ideology. Because of their inability to verify information to which they are exposed, this information over time becomes factual, because it is shared with others in various contexts or platforms such as Facebook (Kushin and Yamamoto, 2010). Social reality, rather than objective reality, likely will become the basis for the development of extreme political ideologies (McLeod and Chaffee, 1972). The current study’s findings largely support this theory. They suggest that liberal and conservative ideologies develop and change in absence or suppression of factual information about public and political affairs by the media; they are largely influenced by traditional and social media, especially among Republican and undecided voters, with such impacts largely favoring a liberal ideology over a conservative one. The stronger influence of traditional and social media on shaping liberal views may be a reflection of a lack of factual information and recent negative coverage of conservative views by the main media (Patterson, 2017) that further determines the information contained in the social media (Taylor, 2008).
These notions also are supported with the strong negative correlation between public affairs knowledge and the bipolar measure of political ideology (r = –.402, p < 0.001), which suggests that those who have the strongest liberal views are the least knowledgeable about public and political affairs. These notions also are suggested by the negative link of age to both public affairs knowledge and liberal political ideology among the undecided—i.e., as voters become more knowledge able with age, they might more likely develop a conservative rather than liberal political ideology. Thus, to the extent that potential voters, especially Democrats, do not develop their political orientations based on facts, as the present study’s findings suggest, further research is needed on the development of liberal political ideology and the causal direction between the two variables. Recent research, for example, suggests that the negative relationship between political knowledge and liberal ideology may, in part, reflect unwillingness to accept what the scientific community has proven as a fact (Scheufele and Krause, 2019).
Reliance on the various sources of political information, predicting stronger liberal over conservative political views among the undecided and even among Republican party supporters, is a phenomenon.
In summary, the findings suggest that the various sources of political information generally appear to provide little factual or bipartisan information; the various media appear to be more powerful in shaping and shifting public opinion and, therefore, may be contributing to political polarization by disseminating partisan disinformation, especially among undecided voters who are not likely to hold extreme political views. These notions are consistent with earlier views on the effects of social media (Tucker et al., 2018). Because it is unlikely that voters will be able to find and consult unbiased sources of political information in today’s media landscape (e.g., Baum and Groeling, 2008; Li and Su, 2020), repealing Section 230 and changing the status of powerful social media to “publishers” could help reduce the dissemination of political disinformation.
Additional findings that have implications for political campaign development are the strong relationships between the value of mainstream media (broadcast and print) and issue importance that underscore the importance of the agenda-setting function of these media in campaign strategy. The links between the value of online media (social media and Internet) and issue importance help extend the agenda-setting theory of mass media effects by suggesting that topics covered by the mainstream media are received or shared not only in face-to-face settings but also online. The absence of significant relationships between the value of the online media and issue importance in the presence of mainstream media suggests the powerful effects of these traditional sources and help clarify the presumed reciprocal influence between traditional and social media (Tucker et al., 2018).
By contrast, these media sources have relatively weaker associations with measures of changes in the political ideologies of the three groups of registered voters, especially among the two main party supporters, that highlight selective perception processes. The value-expectancy theory of attitude change, which states that “a persuasive message will be effective to the extent that it produces a change in the likelihood or the desirability component of an attribute” of an object (Petty, Unnava, and Strathman, 1991, p. 254), provides guidance for promotional strategy using these findings. It suggests that it might be easier to change the political ideology of a party’s supporters by changing their beliefs on certain issues by, for example, increasing the importance of an issue that favors a candidate, than attempting to persuade them to change their beliefs about the candidate’s position on that issue. In a similar vein, undecided voters might more likely judge a candidate’s credibility on the basis of facts about his or her previous record on a certain issue—i.e., a higher likelihood of attribute belief—than on promises of actions on that particular issue.
If the study findings on the possible effects of the various sources of political information by subscribing to the life course paradigm—which assumes that the formation and change of people’s values and habits take place over one’s entire life— are accepted, then the powerful effects of social and media sources that serve as socialization agents must be acknowledged (Moschis, 2019). The findings of this study, albeit exploratory and lacking adequate theoretical explanations, suggest that these sources more likely would promote liberal than conservative political views. The finding that reliance on the various sources of political information predicts stronger liberal over conservative political views among the undecided and even among Republican party supporters is a phenomenon that requires theoretical explanation. It might be that these two groups are exposed to disproportionately greater amounts of liberal information disseminated by these sources, or that liberal-dominated sources are more effective than conservative sources in communicating partisan information. Alternatively, conservative views may be suppressed by the various media and personal sources, or partisan disinformation is disproportionally affecting conservatives. The latter explanation would be a concern that requires intervention, as disinformation, censoring, and suppressing a party’s political views undermines democracy (Tucker et al., 2018).
The relationships between voters’ reliance on social media and their lower level of knowledge and stronger liberal political ideology, especially among undecided and Republican potential voters, are even more disturbing when they are put in the context of previous research findings. Alternative explanations of these relationships notwithstanding, research studies suggest that misinformation that is introduced through a social format influences individuals drastically more than misinformation delivered nonsocially (Gabbert Memon, Allan, and Wright, 2004). Because diffusion of misinformation through social media is a potential threat to democracy and society (Allcott, Gentzkow, and Yu., 2019), the recent finding that 47 percent of Americans report social media as their main news source, as opposed to traditional news sources (Shearer and Gottfried, 2017), should be of concern. The recent finding that U.S. adults who mostly rely on social media for political news are often less knowledgeable about current events (Mitchell et al., 2020) might be particularly relevant to undecided potential voters, who are three times as likely as their Republican counterparts to say that they find social media helpful in shaping their political views.
Future research could attempt to replicate the findings of the current study and address the effects of disinformation in the various types of media on the development of political ideologies and political polarization. The study of the creation and flow of disinformation, which is present in the reciprocal relationship between social and traditional media (Tucker et al., 2018), deserves further examination. Additionally, little is known about the extent to which mainstream and social media suppress or censor information and the effects of these activities on voters’ opinions and preferences for political candidates and parties. In a similar vein, investigation of the relatively unexplored effects of social media, such as the tactical media that promote the ideology of protest and the mobilization of anarchy groups, would be particularly interesting (e.g., Vinogradova, Melnik, and Shaldenkova, 2020). Howard and Parks’s (2012) article, for example, documents the powerful effects of social media in mobilizing those who are uninformed and not involved in political activism. Further exploration into these phenomena is warranted, using multi-item measures of information sources (e.g., gratifications sought and received; O’Guinn and Faber, 1991), focusing on the effects of specific media content, and using longitudinal designs and qualitative research.
ABOUT THE AUTHORS
Anil Mathur is Brodlieb Distinguished Professor of Business and chairperson of the marketing and international business department at the Frank G. Zarb School of Business, Hofstra University, in Hempstead, NY. His research focuses on consumer behavior, research methodology, and neuromarketing, having appeared in the Journal of the Academy of Marketing Science, Psychology & Marketing, Journal of International Business Studies, and Journal of Business Research. In addition to publishing in these and other journals, Mathur is coauthor (with George P. Moschis) of Baby Boomers and Their Parents: Surprising Findings about Their Lifestyles and Well-Being (Paramount Publishing, 2007), among other books.
George P. Moschis is professor of management at Mahidol University, Bangkok, Thailand. He recently retired from Georgia State University, where he had begun this research. In addition to his academic scholarship, he has consulted for major corporations and has served as an expert witness on high-profile litigation cases. Moschis’s research can be found in the Journal of the Academy of Marketing Science, Journal of Marketing Research, Journal of Marketing, and Journal of Consumer Research, among others, as well as in several books, such as Consumer Behavior over the Life Course: Research Frontiers and New Directions (Springer, 2019).
- Received August 28, 2020.
- Received (in revised form) July 1, 2021.
- Accepted November 30, 2021.
- Copyright © 2022 ARF. All rights reserved.
REFERENCES
ARF MEMBERS
If you are a member of the Advertising Research Foundation, you can access the content by logging in here
Log In
Pay Per Article - You may access this article (from the computer you are currently using) for 30 days for US$20.00
Regain Access - You can regain access to a recent Pay per Article purchase if your access period has not yet expired.