Comparing Romney and Obama's Facebook Messages & Audience Interests in 2012 Election, Schemes and Mind Maps of Communication

Download Comparing Romney and Obama's Facebook Messages & Audience Interests in 2012 Election and more Schemes and Mind Maps Communication in PDF only on Docsity! Agenda-setting in the one-step flow: Evidence from Facebook in the 2012 election Deen Freelon School of Communication, American University Abstract: Agenda-setting is one of the best-substantiated theories in communication research. But as the digital age has reoriented the media landscape, it has called into question some of the theory’s most basic underlying assumptions. The mass media, once by far the most powerful and prominent agenda- setting force at work, must now compete with other interests wishing to set their own agendas. Politicians, once relegated to communicating with citizens through mass media, can now connect with them directly via social media. Such a direct, targeted connection is known as the one-step flow of communication. This study examines how agenda-setting works under one-step flow conditions by applying a lexicon analysis to Barack Obama’s and Mitt Romney’s Facebook posts and the comments appended to them during the 2012 US presidential campaign (N = 858,307). Results indicate that Romney’s agenda is more similar to his audience than Obama’s is to his, both audiences are more interested in religion than either candidate, and the audiences emphasize the same attributes as the candidates for some issues but different attributes for others. Social media and the prospects for expanded democratic participation in national policy-setting workshop, Boston University April 9, 2015 Agenda-setting in the one-step flow: Evidence from Facebook in the 2012 election Politicians can communicate with citizens in two ways. They can take the indirect route by funneling their thoughts to citizens through the news media, which may alter them in unanticipated ways. Alternatively, politicians are increasingly turning to digital media to communicate directly with citizens, cutting the news media out of the equation entirely. The rise in popularity of such disintermediated or “one-step flow” communication pathways (Bennett & Iyengar, 2008; Blumler & Kavanagh, 1999; Katz, 1988) introduces intriguing new possibilities for well-known political communication theories. Agenda-setting is one of the best-substantiated theories in communication research (McCombs, 2005). Its original formulation assumed the existence of a more or less monolithic news media whose decisions about story salience would be reflected in perceptions of issue importance among audience members. Recent work on agenda-setting in the digital age has continued to treat news media as central (Neuman, Guggenheim, Jang, & Bae, 2014; Vargo, Guo, McCombs, & Shaw, 2014). Less explored has been the potential for politicians to set their constituents’ agendas without the participation of the media. Audience members may to varying degrees follow along the agenda path set by their leaders, or attempt to inject their own pet concerns onto the agenda. Digital media have given them unprecedented opportunities for such agenda injection, but few studies have investigated how often they actually do so. The current study provides evidence in answer to these questions by examining over 850,000 messages posted to Barack Obama’s and Mitt Romney’s official Facebook pages during the 2012 US presidential campaign. It uses a lexicon-based approach to compare the issue agendas of both candidates to those of their followers. Results indicate substantial agenda overlap between all parties with several consequential exceptions. Agenda-setting, past and present digitally-mediated communication pathways now exist between politicians and their supporters. And unlike prior direct pathways, social media platforms afford both two-way communication and convenient archiving of message content. Thus, they allow for communications between politicians and citizens to be analyzed in greater detail than ever before. Few existing studies have examined how politicians set citizens’ agendas directly, in part because of the overwhelming preeminence of the mass media in the 20th-century agenda-setting process. It should be noted that the concept of agenda-setting undergoes a slight transformation as it transitions to a one-step flow/social media context. Critically, the “agenda” of which we speak must be reconstructed based on participant conversations rather than survey self-reports. Instead of asking individuals directly about the issues they believe to be most important, the current analysis assumes that the issues they discuss most often are most important to them. This method has the advantage of avoiding observer- or social desirability biases that might accompany survey questions about issue importance. Its main disadvantage is a lack of external validity stemming from the unrepresentativeness of social media users, a point that will be explored further in the discussion. The prospect of agenda-setting in a one-step flow context raises a new set of empirical questions. One might first ask about the extent to which different politicians actually set the agenda(s) of their audiences. Much has been made of ordinary citizens’ technologically-enhanced abilities to set their own agendas with less intervention from mass media outlets (Castells, 2007; Shaw et al., 1999). Politically-interested citizens may be even more empowered to do so than the median citizen given their subject-matter expertise and strong motivations to participate in politics. It is therefore not certain that all politicians will be able to set agendas equally effectively in online one-step flow contexts. Individual politicians may differ in their agenda-setting abilities just as distinct media outlets do (Golan, 2006; Walgrave & Van Aelst, 2006).  RQ1: To what extent do Barack Obama and Mitt Romney set the agendas for their respective social media followers? An important related question addresses what might be called the “agenda gaps” between politicians and audiences. These are the differential foci of the two respective interests on each issue: for example, Romney might be more concerned with the economy than his audience, who might care more about religion (which happens to be the case). On issues for which the gap is small, candidate and audience might be said to concur on issue importance; but major differences show us exactly where their respective agendas diverge. This phenomenon has been noted in previous research (e.g. in Neuman et al., 2014) but its theoretical relevance has not been adequately considered. For example, issues that are popular among audiences but not candidates may evince “agenda melding” or agenda- setting between nonelites (Ragas & Roberts, 2009; Shaw et al., 1999).  RQ2: Which issues do the candidates and their followers emphasize proportionately and disproportionately? Third, there is the question of how second-level agenda attributes differ between candidates and audiences within a single first-level issue. It is quite possible that candidates may discuss issues in ways that differ fundamentally from how audiences discuss them. This may be one reason certain candidates at times find themselves accused of being “out of touch” with the electorate. The second level of agenda-setting is another domain within which audience members may exercise the power to resist elite agenda-setting or even to set the terms of the debate among themselves. Neuman et al. (2014) briefly broach this possibility but do not specifically address the elite/audience distinction, nor is their lexicon sufficiently comprehensive to capture a broad range of audience concerns. Hence the current extension of their work.  RQ3: For the most prominent issues, to what extent do the candidates discuss the same issue attributes as their followers? Methods Lexicon-based approaches have a long history in the agenda-setting literature (Neuman et al., 2014; Tedesco, 2001, 2005). It is a fitting approach to examine one-step agenda-setting in social media because of the ready availability of an extensive textual record of communication between citizens and politicians. While previous studies have inferred agenda-setting processes on the basis of time-lagged correlations between source and target texts (e.g. Neuman et al., 2014; Tedesco, 2001, 2005), social media offer key affordances that render such inferences unnecessary. When sources (politicians in this case) post to their social media accounts, followers can append comments directly to the original messages. Therefore, any correspondence in subject matter between original messages and replies is assumed to be the result of a top-down agenda-setting process. Of course, citizens who follow politicians on social media may also attempt to steer the agenda toward their own concerns, especially because they know that attaching their comments to messages posted by well-known individuals is an effective means of attracting attention. Agenda salience is measured by the prevalence of keywords related to political issues. As such, we can measure not only the extent to which candidates set the agendas of their followers, but also which issues are relatively more and less relevant to each candidate compared to his audience. Moreover, we can compare the specific terms used by each candidate and his audience to examine qualitatively which issue attributes each discusses most commonly. This expands the analysis onto the second level of agenda-setting (Ghanem, 1997; Weaver, 2007), which is concerned with the specific attributes of broader issues (e.g., if the issue is the economy, specific aspects could include taxes, GDP, debt, and jobs). This study’s methods thus afford an examination of not only which issues candidates and their followers discuss, but also of how they discuss them. The data to be analyzed were collected from Facebook. They include all messages posted to Barack Obama’s and Mitt Romney’s official Facebook pages between April 25, 2012 (the day the After the data were preprocessed in the above manner, the lexicon analysis could begin. Upon completion, it yielded a set of per-post counts for each issue category and term. That is, the lexicon output contains the number of posts in which each candidate and his commenters 1) mentioned each individual term and 2) mentioned at least one term from each issue category. These counts form the basis of the empirical analysis that follows. Results Before we address the research questions, let us first consider some of the more relevant descriptive statistics. In total, 81 of Obama’s 268 posts (30.2%) contained at least one issue term while 292 (50.0%) of Romney’s 584 did. Similarly, Obama’s commenters discussed policy issues proportionally less often (80,795 posts of 233,129, 34.6%) than Romney’s did (266,335 posts of 624,326, 42.7%). Thus both Romney and his followers were more focused on policy than their Democratic counterparts, at least on Facebook. Of course, many posts mentioned more than one term or issue. Figures 1a-d reveal each subset’s relative issue emphasis. Each chart’s y-axis indicates the proportion of all issue category mentions. Figure 1a, for example, shows that Obama’s two most- mentioned issues were civil rights and finance/economics/labor (FEL), followed by healthcare, with no other issue exceeding 10% of all issue mentions. In Figure 1b we see that Romney’s messages were dominated by economic concerns and that the next most popular issue, healthcare, accounted for only 7% of issue mentions. Unsurprisingly given their much larger post volumes, the commenters’ issue mentions were more evenly distributed than the candidates’, although this was more the case for Obama’s commenters than Romney’s (Figures 1c-d). Both were most interested economics with religion and civil rights following close behind, as the remaining issues gently tapered off. RQ1 The first research question asks about how effectively each candidate set his commenters’ respective agenda. To test this, each group of commenters’ issue proportions were entered as a dependent Figure 1a: Obama’s issue proportions Figure 1b: Romney’s issue proportions Figure 1c: Obama commenters’ issue proportions Figure 1d: Romney commenters’ issue proportions 0 0.1 0.2 0.3 0.4 0.5 0.6 Is su e p ro p o rt io n 0 0.1 0.2 0.3 0.4 0.5 0.6 Is su e p ro p o rt io n 0 0.1 0.2 0.3 0.4 0.5 0.6 Is su e p ro p o rt io n 0 0.1 0.2 0.3 0.4 0.5 0.6 Is su e p ro p o rt io n variable into two OLS regressions: one using only the corresponding candidates’ issue proportions as the sole predictor, and the other adding the remaining two issue proportion subsets as predictors. This hierarchical approach includes the focal test of each candidate’s agenda-setting power over his audience in two separate regressions to lower the likelihood that the results are spurious. The unit of analysis in all four regressions is the issue category (n = 18). Table 1a contains the output of the regressions of Obama’s commenters, while Table 1b contains those for Romney’s commenters. Each table’s Model 1 includes the candidate’s issue proportions as the sole predictor, while Model 2 adds the remaining two subsets as controls. For the current purposes the standardized regression coefficients are the appropriate quantities to compare between models and candidates; unstandardized coefficients are included for the sake of completeness. Model 1 Model 2 Predictor B Beta SE B Beta SE Obama issues 0.515*** 0.762 0.109 0.12** 0.178 0.039 Romney issues . . . -0.303*** -0.708 0.047 Romney commenters’ issues . . . 1.161*** 1.474 0.094 Constant 0.026 . 0.01 0.001 . 0.003 R2 0.581*** 0.976*** ΔR2 . 0.395*** Table 1a: OLS regressions of Obama commenters’ issue proportions. * = p < 0.05, ** = p < 0.01, *** = p< 0.001 Model 1 Model 2 Predictor B Beta SE B Beta SE Romney issues 0.505*** 0.928 0.051 0.276*** 0.508 0.023 Obama issues . . . -0.082* -0.095 0.036 Obama commenters’ issues . . . 0.790*** 0.622 0.064 Constant 0.0276 . 0.006 0.001 . 0.003 R2 0.861*** 0.99*** ΔR2 . 0.129*** Table 1b: OLS regressions of Romney commenters’ issue proportions. * = p < 0.05, ** = p < 0.01, *** = p< 0.001 RQ3 The final research question concerns term usage differences within issue categories between the subsets. Because space limitations prohibit an exhaustive exploration of all issues, this section will examine the four most-frequently-mentioned issues across all four subsets: FEL, civil rights, healthcare, and religion. For each of these issues, Table 2 presents each subset’s top five most-used terms. Perhaps the most readily apparent pattern in this table is the strong similarity of terms within most issue categories. This is probably most clear for FEL and healthcare, within which many of the same terms are repeated across subsets (“job,” “tax,” “obamacare,” “medicare”). Those terms that are not repeated do not seem to indicate a consistent focus on fundamentally different issue attributes than those that are. Issue Rank Obama O comments Romney R comments Civil rights 1 woman woman woman woman 2 daughter rights mother rights 3 mother daughter rights racis 4 latino abortion latino daughter 5 voter_registration racis voting_right abortion Finance/ economics/ labor (FEL) 1 job job job job 2 econom tax econom tax 3 tax econom middle_class econom 4 middle_class dollar tax middle_class 5 private_sector debt unemploy dollar Healthcare 1 health health obamacare obamacare 2 obamacare obamacare medicare health 3 medicare medicare health medicare 4 copay medical nurs medical 5 preventive doctor medicin doctor Religion 1 faith god prayer god 2 - muslim faith muslim 3 - islam religio religio 4 - religio god islam 5 - christian catholic mormon Table 2: Ranked terms for four issue topics across candidate and audience data subsets Civil rights represents a partial departure from this pattern. While all three subsets are substantially focused on women’s issues—almost certainly a result of candidate agenda-setting—the two candidates are in near-total lexical lockstep with one another, with nearly every term having an obvious analogue on the opposing side. But while the commenters share this concern, they also both decisively inject abortion and racism onto the agenda item’s second level, two issue attributes neither candidate mentions at all. The current methods do not permit any claims about which sides of each issue are most popular, but it is interesting to see the commenters raising issues the candidates would prefer to avoid. Nowhere is this agenda injection phenomenon more evident than on the issue of religion. Obama makes only one passing reference to it, in an exhortation to his supporters to “keep the faith,” which carries at best only a vague religious connotation. Romney mentions religion more often, but usually with nonspecific terms like “prayer” and “god,” with the exception of “catholic” which he uses only once. Meanwhile, the commenters do not hesitate to discuss the specifics of religions that played controversial roles in the campaign, especially Islam. Romney’s religion of Mormonism was a popular topic among his commenters, albeit less popular than Islam. Overall, these differences in how and how much audiences and candidates discuss religion speak volumes about which aspects of the topic they feel are most consequential. Discussion This study offers a summary impression of the issue agendas expressed by two presidential candidates and their social media audiences during the 2012 US presidential campaign. Its main findings are as follows: first, the fit between Romney’s agenda and that of his audience is closer proportionally than for Obama. This outcome, combined with the pride of place and time precedence of candidate messages as compared to audience comments, offers strong evidence that Romney set his Facebook audience’s agenda more effectively than Obama set his. Of course, the outcome of the 2012 election demonstrates that agenda-setting power on Facebook is neither necessary nor sufficient to win. Part of the disparity may be explained by the fact that Romney’s audience was simply more engaged on policy issues overall, as their greater proportion of issue mentions compared to Obama’s audience attests. The second major finding is that each candidate’s agenda generally resembled his audiences’ in terms of proportional emphasis with a few conspicuous exceptions. Obama’s strong interest in civil rights, and specifically women’s issues, far outstripped that of his audience. Their top concern was FEL issues, which also topped the lists of the other two subsets. While Obama spent almost as much time discussing both issues, his audience felt that FEL was almost twice as important as civil rights. For his part, Romney’s concern with the economy exceeded his audience’s, which also prioritized it over all other issues. Both audiences raise the issue of religion in greater proportions than their respective candidates. This is a clear example of what might be called “agenda injection”—an attempt to introduce an unsanctioned topic into a conversation ostensibly about something else. And while it is unlikely that this injection attempt noticeably affected the candidates’ agendas, it may have played some role in shifting the agendas of audience members and their Facebook followers. We see further evidence of agenda injection within some issues in the results of the RQ3 analysis. On the issues of FEL and healthcare, the difference between the four subsets is minimal—the same terms appear time and again, and unique terms do not indicate a fundamental shift in focus onto a different attribute of the issue. This is also true to some extent for civil rights; however, the presence of “racis” and “abortion” on both audience lists shows a willingness of nonelites to court controversies the candidates would prefer to ignore. Specifically on the topic of women’s issues, it is telling that the candidates speak in nebulous terms of “women’s health” and “women’s issues,” while the audience directly references the specific procedure that most concerns them: abortion. On religion the candidates again remain either vague (Romney) or silent (Obama), in contrast to the audience, which invokes specific religious identities they feel are germane to the campaign. In recent American campaign history, References Aday, S., Farrell, H., Freelon, D., Lynch, M., Sides, J., & Dewar, M. (2013). Watching From Afar: Media Consumption Patterns Around the Arab Spring. American Behavioral Scientist, 57(7), 899–919. doi:10.1177/0002764213479373 Albaugh, Q., Sevenans, J., & Soroka, S. (2013). Lexicoder Topic Dictionaries. McGill University. Retrieved from http://www.lexicoder.com/download.html Albaugh, Q., Sevenans, J., Soroka, S., & Loewen, P. J. (2013). The Automated Coding of Policy Agendas: A Dictionary-Based Approach. Presented at the 6th annual Comparative Agendas Project (CAP) conference, Antwerp. Bennett, W. L., & Iyengar, S. (2008). A new era of minimal effects? The changing foundations of political communication. Journal of Communication, 58(4), 707–731. Bennett, W. L., Lawrence, R. G., & Livingston, S. (2007). When the press fails: Political power and the news media from Iraq to Katrina. Chicago: University of Chicago Press. Bennett, W. L., & Manheim, J. B. (2006). The One-Step Flow of Communication. The ANNALS of the American Academy of Political and Social Science, 608(1), 213–232. doi:10.1177/0002716206292266 Blumler, J. G., & Kavanagh, D. (1999). The third age of political communication: Influences and features. Political Communication, 16(3), 209–230. Bruns, A. (2005). Gatewatching: Collaborative online news production. New York: Peter Lang. Carey, J. W. (1997). The press, public opinion, and public discourse: On the edge of the postmodern. In E. S. Munson & C. A. Warren (Eds.), James Carey: A critical reader (pp. 228–260). Minneapolis: University of Minnesota Press. Castells, M. (2007). Communication, power and counter-power in the network society. International Journal of Communication, 1(1), 238–266. Center for Responsive Politics. (n.d.). The Money Behind the Elections. Retrieved February 21, 2015, from https://www.opensecrets.org/bigpicture/ Duggan, M., Ellison, N. B., Lampe, C., Lenhart, A., & Madden, M. (2015, January 9). Social Media Update 2014. Retrieved from http://www.pewinternet.org/2015/01/09/social-media-update-2014/ Eisenstein, J., O’Connor, B., Smith, N. A., & Xing, E. P. (2014). Diffusion of Lexical Change in Social Media. PLoS ONE, 9(11), e113114. doi:10.1371/journal.pone.0113114 Entman, R. M. (2009). Projections of Power: Framing News, Public Opinion, and U.S. Foreign Policy: Framing News, Public Opinion, and U.S. Foreign Policy. University of Chicago Press. Ghanem, S. (1997). Filling in the tapestry: The second level of agenda setting. Communication and Democracy: Exploring the Intellectual Frontiers in Agenda-Setting Theory, 3–14. Golan, G. (2006). Inter-media agenda setting and global news coverage. Journalism Studies, 7(2), 323– 333. doi:10.1080/14616700500533643 Hayes, D., & Guardino, M. (2010). Whose Views Made the News? Media Coverage and the March to War in Iraq. Political Communication, 27(1), 59–87. doi:10.1080/10584600903502615 Jenkins, H., Ford, S., & Green, J. (2013). Spreadable Media: Creating Value and Meaning in a Networked Culture. New York: NYU Press. Katz, E. (1988). Disintermediation: Cutting out the middle man. Intermedia, 16(2), 30. Katz, E., & Scannell, P. (Eds.). (2009). The end of television? Its impact on the world (so far). Thousand Oaks, CA: Sage. McCombs, M. (2005). A Look at Agenda-setting: past, present and future. Journalism Studies, 6(4), 543– 557. doi:10.1080/14616700500250438 McCombs, M., & Shaw, D. L. (1972). The Agenda-Setting Function of Mass Media. Public Opinion Quarterly, 36(2), 176–187. doi:10.1086/267990 Meraz, S., & Papacharissi, Z. (2013). Networked Gatekeeping and Networked Framing on #Egypt. The International Journal of Press/Politics. doi:10.1177/1940161212474472 Neuman, W. R., Guggenheim, L., Jang, S. M., & Bae, S. Y. (2014). The Dynamics of Public Attention: Agenda-Setting Theory Meets Big Data. Journal of Communication, 64(2), 193–214. doi:10.1111/jcom.12088 Petchler, R., & Gonzalez-Bailon, S. (forthcoming). Automated content analysis of online political communication. In S. Coleman & D. Freelon (Eds.), Handbook of Digital Politics. Northampton, MA: Edward Elgar. Ragas, M. W., & Roberts, M. S. (2009). Agenda Setting and Agenda Melding in an Age of Horizontal and Vertical Media: A New Theoretical Lens for Virtual Brand Communities. Journalism & Mass Communication Quarterly, 86(1), 45–64. doi:10.1177/107769900908600104 Shaw, D. L., McCombs, M., Weaver, D., & Hamm, B. J. (1999). Individuals, Groups, and Agenda Melding: A Theory of Social Dissonance. International Journal of Public Opinion Research, 11(1), 2–24. doi:10.1093/ijpor/11.1.2 Shehata, A., & Strömbäck, J. (2013). Not (Yet) a New Era of Minimal Effects A Study of Agenda Setting at the Aggregate and Individual Levels. The International Journal of Press/Politics, 18(2), 234–255. doi:10.1177/1940161212473831 Shpayher, J. (n.d.). United States House of Representatives - Congress Social Media Wiki. Retrieved from http://govsm.com/w/House Stromer-Galley, J. (2000). On-line interaction and why candidates avoid it. Journal of Communication, 50(4), 111–132. doi:10.1111/j.1460-2466.2000.tb02865.x Stromer-Galley, J. (2014). Presidential campaigning in the Internet age. New York: Oxford University Press.