October 15, 1994. Susan Lewis is doing what all politically interested and motivated citizens do in October of a contentious midterm election: everything possible to persuade the people she disagrees with to change their opinions and to encourage the people she agrees with to vote. Lately, she’s been glued to CNN, grateful for the round-the-clock coverage that still seems like a novelty. She devours any information she can get her hands on and feels compelled to share her opinions with others. There is no way for Susan to easily contact everyone she knows, but she does her best. She’ll sing the praises of her preferred candidate to anyone who will listen. It is too expensive to make long-distance phone calls to friends and family who live in more competitive congressional districts, but she mobilizes the people in her local community. Susan photocopies flyers with information about voter registration and posts them at her church and in the break room at work. She volunteers for the state party’s phone bank and canvasses door-to-door in her community with a clipboard in hand.
Susan’s sister, Janet, couldn’t be more different in terms of her level of political engagement. Although Janet trusts Susan’s opinions and typically agrees with her political views, she has always avoided politics, perhaps as a legacy of her sister’s diatribes and rapt attention to the news of the day. For the most part, Janet is successful in her efforts to completely ignore political news and policy debates. She doesn’t subscribe to a newspaper and has plenty of other channels to watch when the news comes on the television in the evening. Janet occasionally overhears her co-workers talking about the latest clash between the political parties, but it’s easy enough to avoid the conversation. Her sister inevitably makes discussions at holiday meals awkward when she attacks the views of their uncle, but it only happens a couple of times a year. Her close friends seem similarly politically disinterested, perhaps part of the reason they originally became friends. Janet doesn’t give much thought to what other people think about politics; not only does partisan identity seem unimportant or irrelevant, but Janet isn’t even sure how she would learn that information, since it isn’t visible in her interactions with them.
This depiction of political interaction is outdated. Interpersonal political communication has undergone major changes in the last twenty-five years. While traditional forms of interaction will always persist – face-to-face political conversation with colleagues, friends, and family is a staple of democratic citizenship – changing technologies have lowered the costs for certain behaviors, while rendering obsolete others.
While the details of their communication patterns are antiquated, the roles these characters play are still familiar. Most of us know a Susan, someone who cares deeply about politics. Political scientists call people like Susan opinion leaders or members of the engaged public (Abramowitz Reference Abramowitz2010), and in dozens of studies across multiple decades, we’ve found that people like Susan are in a minority. For some readers, those sufficiently politically engaged and old enough to remember the costs of political communication before the rise of the Internet and social media, Susan’s efforts may inspire nostalgia. But the average American, reminiscing back to the pre-Internet era, can probably better relate to Janet. Most people, most of the time, don’t think too much about politics (Converse Reference Converse and Apter1964). While many Americans report that they have political discussions at least occasionally, the majority of people prefer to seek out entertainment over news, given the choice.
Technology in and of itself does not cause societal change (Starr Reference Starr2005), but how people adopt and integrate technology has enormous implications for the ways in which we engage with one another. This can be readily observed in the domain of political interaction. Fictional Susan Lewis, circa 2018, has dozens of outlets to share her political views and persuade her social network, allowing her to reach more people, more quickly and dynamically than ever before. She follows candidates and elected officials on the social media site Facebook, and, paired with push notifications on her phone from her favorite news sites, she knows about current events almost instantaneously. She regularly posts links to important stories in her News FeedFootnote 1 and goads people into engaging in political discussions on Facebook by sharing funny political cartoons she stumbles across. In the last election, her presidential candidate’s Facebook app used an algorithm to identify all the people in her social network who lived in competitive “battleground” states, and Susan was able to easily and personally reach out to them to remind them to vote.
The advent of social media, particularly the popular social networking site Facebook, has had a profound effect on the “Susans” of the world, but as the anecdote at the beginning of the chapter shows, Susan has always been politically engaged, even when it was very costly in terms of time and effort to do so.
While alterations to the media environment have made political communication easier for Susan, they’ve had a much more dramatic effect on Janet. Social media has rendered changes to political communication that fundamentally alter the way the disengaged public experiences politics. Writing a vignette for Janet in the age of social media looks something like this:
October 15, 2018. Janet Lewis now knows more about the political opinions of her family, friends, and acquaintances than she ever has before, but not because she has sought out that information. When she scrolls through her Facebook News Feed – which she does every day in order to share pictures of her grandkids and keep tabs on what people from her past are doing – she inevitably encounters political content in the form of news stories, memes, and videos, with commentary written by her social connections accompanying it all. It seems in the months leading up to an election, she simply can’t escape the onslaught. She’s always felt this way about her sister, but now she’s connected to at least a dozen people who post about politics all the time.
While her friends who post this content share views with which she tends to agree, a substantial minority post content that so obviously signals divergent political views that even Janet can recognize the disagreement, although she cares little and knows less about current political issues. This seems especially to be the case with those Facebook friends with whom she would never talk about politics in person, such as long-lost friends from high school, her grandson’s piano teacher, and the other volunteers at the food bank. Based on what they post, she’s begun to make associations about which kinds of people tend to believe what kinds of things. It seems as if there is a huge gulf between the members of the two political parties.
Janet has always avoided political conversations. But for the first time, she has a window into the conversations of groups of people with whom she disagrees strongly. They all seem to egg each other on, liking and writing encouraging comments on each other’s posts, and posting information Janet thinks must be biased or factually incorrect based on what Susan posts. As a consequence of seeing this, Janet has become more judgmental about those people with whom she disagrees, and she assumes the worst about the people who appear to have such extreme opinions. She’s never much cared for politicians, but she now has developed negative feelings for the people who believe what those politicians spout.
We would say that Janet has become psychologically polarized. She’s not alone. Using a variety of different measures, political scientists have shown that a large and likely growing proportion of Americans have very negative attitudes toward people and candidates who identify with the opposing political party.
Susan would have been an opinion leader in any era of American politics. However, Facebook offers her an unprecedented platform to express her political opinions, disseminate political information, and draw others into political discussions. People like Susan create an atmosphere within the Facebook News Feed that has the potential to psychologically polarize people like Janet, those who don’t know or care that much about politics. Susan has several hundred friends on Facebook who may see the political content she circulates, and the vast majority of them are like Janet, choosing not to actively engage with it. But Facebook users are not able to easily escape exposure to the content opinion leaders post. Because of the motivations most people have for using the site – to learn about the social lives of their friends and family – that exposure has consequences for the connections they make between people’s political and social identities.
The findings of this book suggest that psychological polarization can result from the political communication environment on social media, for both the politically engaged and disengaged publics. While there are many reasons that Susan Lewis may be more polarized now than she was two decades ago, previous explanations cannot fully account for why people like Janet have become more negative about their fellow citizens. Exposure to political interactions on Facebook could play an important role in that process. The changes to the way in which opinion leaders interact with and influence those in their networks, and the result of exposure to the vitriolic ecosystem of political communication online, have facilitated Americans in forming derogatory judgments about people who disagree with them.
We begin with a puzzle that has consumed political scientists studying American politics for the past twenty years: political polarization. Scholars almost uniformly believe that political elites are more polarized today than they have been since the American Civil War, and there is ample empirical support documenting and explaining the rise of this phenomenon. There is less consensus about the extent to which the average American has developed more extreme political attitudes and opinions. But in the process of exploring the trajectory of policy preferences of the mass public, we discovered something else. Distinct from the extremity of the policy views Americans hold, they have developed stronger in-group preferences for their own political party alongside more negative effect for those across the aisle and they perceive more distance – socially and politically – between the two groups. How this psychological polarization has occurred is a question that remains an unresolved puzzle.
Psychological Polarization
Scholars have written prolifically on the many facets of polarization that affect the American political system. Among political elites, polarization is characterized by the disappearance of moderates in Congress and the two parties pulling apart ideologically, whether measured by interest group ratings (Poole and Rosenthal Reference Poole and Rosenthal1984), scaled measures of roll call votes (Poole and Rosenthal Reference Poole and Rosenthal2001; Clinton et al. Reference Clinton, Jackman and Rivers2004; McCarty et al. Reference McCarty, Poole and Rosenthal2006), or measures of party vote percentages and party unity scores (Bond and Fleisher Reference Bond and Fleisher2000; Stonecash et al. Reference Stonecash, Brewer and Mariani2003; Jacobson Reference Jacobson2004). Although there have been eras of deep polarization in the country’s history, the increase in polarization that has developed since 1970 has resulted in the most polarized Congress seen in modern times. There are few indications that this trend will reverse itself and many more signs of the existence of feedback mechanisms that may perpetuate polarization. Ideological moderates are less likely to run for Congress (Thomsen Reference Thomsen2014) and elected officials are less likely to hear from their constituents who belong to the other political party (Broockman and Ryan Reference Broockman and Kalla2016).
What is less apparent is the degree to which the mass public has become polarized. The most prominent debate – between Alan Abramowitz and Kyle Saunders on one side, and Morris Fiorina and Samuel Abrams and colleagues on the other – went in circles, largely because of disagreements in the definition and measurement of polarization (Saunders and Abramowitz Reference Saunders and Abramowitz2004; Abramowitz and Saunders Reference Abramowitz and Saunders2005, Reference Abramowitz and Saunders2008; Abramowitz and Stone Reference Abramowitz and Stone2006; Fiorina and Abrams Reference Fiorina and Abrams2008; Abramowitz Reference Abramowitz2010; Abrams and Fiorina Reference Abrams and Fiorina2012; Fiorina Reference Fiorina2013). At the core of this initial debate over mass polarization in the early 2000s was the question of whether Americans had become more extreme in their viewpoints or had simply sorted their partisanship with their ideological viewpoints, what Lelkes (Reference Lelkes2016) differentiates as ideological divergence and ideological consistency.Footnote 2 In the process of debating the evidence for these two forms of attitudinal polarization, scholars uncovered evidence for two coexisting phenomena that are more related to Americans’ attitudes about their political identities and the relationship between groups in the political system.Footnote 3
Leaving aside the question of whether Americans hold more extreme or sorted policy preferences, there is substantial evidence that they see large differences between the parties, develop political identities that overlap considerably with their social identities, and disparage the out-party. These psychological forms of polarization matter as much or more than the empirical reality of preference polarization for understanding the way in which Americans view themselves and view others in the political landscape.
Affective polarization can be conceptualized as partisans’ increasingly negative feelings and negative trait attribution toward identifiers of the opposing party (Iyengar et al. Reference Iyengar, Sood and Lelkes2012; Lelkes Reference Lelkes2016). A contributory, but distinct, pattern is the rise of perceived polarization, or the extent to which the mass public identifies the political parties and their adherents to be polarized (Lelkes Reference Lelkes2016). These perceptions have important influences on emotional evaluations of the parties (Levendusky and Malhotra Reference Levendusky and Malhotra2016a) and on people’s motivatation to exaggerate their perception of party difference for reasons related to identity affirmation (Bullock et al. Reference Bullock, Gerber, Hill and Huber2015; Prior et al. Reference Prior, Sood and Khanna2015).
Evidence is mounting for these two forms of psychological polarization. Feeling thermometer survey questions asked in the American National Election Study (ANES) over time demonstrate that while positive feelings toward one’s own party have stayed consistently high, feelings toward the opposition party have dropped considerably in the last thirty years (Iyengar et al. Reference Iyengar, Sood and Lelkes2012; Kimball et al. Reference Kimball, Summary, Vorst, Green, Coffey and Cohen2014). These ratings are closely tied to emotional evaluations: in 2012, almost half of partisans reported being afraid of and almost two-thirds reported being angry at the presidential candidate of the opposition party (Kimball et al. Reference Kimball, Summary, Vorst, Green, Coffey and Cohen2014). On average, the political parties are perceived to be about 20 percent farther apart on a broad set of issues than they are in reality (Levendusky and Malhotra Reference Levendusky and Malhotra2016a).
Taken to the extreme, both of these forms of polarization suggest that partisans could perceive greater social distance between the parties in addition to recognizing greater policy difference. Perhaps disturbingly, but most relevant for the argument of this book, Americans do appear to have become more negative about each other. Large proportions of Americans would be “displeased” if their child married someone from another political party (Iyengar et al. Reference Iyengar, Sood and Lelkes2012). Fewer than 15 percent of people view the out-party – including both candidates and voters – as possessing core moral traits (Miller and Conover Reference Miller and Conover2015). Findings presented in this book reveal that majorities of people agree that voters of the out-party are ignorant, narrow minded, and ideologically driven.
Furthermore, there is evidence that partisan identity affects evaluations that have nothing to do with politics, and may even be involuntary, such as our evaluations of physical attractiveness (Klar and Krupnikov Reference Klar and Krupnikov2016; Nicholson et al. Reference Nicholson, Coe and Song2016), worthiness of academic scholarships (Iyengar and Westwood Reference Iyengar and Westwood2015), and job interview callbacks (Gift and Gift Reference Gift and Gift2015). Even more distressing is that social norms appear to encourage people to overstate their antipathy toward the political out-group. Explicit measures of bias – asking people directly to self-report their attitudes – typically underestimate the extent of negative sentiment because of social desirability concerns. But when assessing partisan bias, explicit measures reveal as much or more bias than implicit measures designed to more subtly capture biased attitudes. This is the exact opposite of what we observe when we measure bias against different racial or religious groups (Lelkes Reference Lelkes2016).
Collectively, this evidence demonstrates that with or without true movement to the ideological poles in the policy opinions of the average American, the contemporary political climate has activated Americans’ political identities and affective orientations. Many Americans have processed the polarizing messages from political elites and the media in a way that alters their evaluations of the parties, candidates, and, most strikingly, their fellow citizens. The classic explanations about why Americans’ policy opinions have moved to the poles cannot entirely account for the identity-based aspects of polarization. People translate their attitudes toward the abstract notion of the political parties to concrete feelings about Americans who do not share the same partisan identity. How did this come to be? The answer to this question is of the utmost importance. These sentiments have implications for the trust Americans have in each other, our leaders, and our political institutions, as well as our perception of the legitimacy of our opponents’ viewpoints.
There are a number of existing explanations for the rise of affective polarization. Americans’ social identities have become increasingly aligned with their political identities, raising the stakes of partisan competition (Mason Reference Mason2018). In a more polarized political environment, partisan cues become important drivers of opinion, especially out-party cues (Nicholson Reference Nicholson2012), and the negativity of the modern campaign environment likely contributes to negative attitudes (Sood, Iyengar, and Dropp Reference Iyengar, Sood and Lelkes2012). The fragmented and partisan media, including both cable news and Internet news, furthers this divisive messaging (Stroud Reference Stroud2010; Levendusky Reference Levendusky2013; Garrett et al. Reference Garrett, Gvirsman and Johnson2014; Levendusky and Malhotra Reference Levendusky and Malhotra2016b; Lelkes et al. Reference Lelkes, Sood and Iyengar2017), and the homogeneity of people’s social networks likely reinforces these other mechanisms of polarization (Parsons Reference Parsons2015).
The explanation on which I focus is one that encompasses many of the ideas proposed previously. Where do all of these factors intersect, creating a maelstrom of political interaction to which many Americans are exposed on a daily basis? On social media, especially the Facebook News Feed, the most ubiquitous forum for online interaction. The News Feed serves as the home page for each Facebook user’s personally tailored experience on the site, delivering a distillation of the information posted and shared by the people and groups to whom a user is connected. The framework of social identity theory (Tajfel et al. Reference Tajfel, Billig, Bundy and Flament1971) predicts that – given social and political sorting in American society, the resultant political discussions within largely homogeneous social networks, and the unprecedented levels of exposure to partisan media and negative campaigning – the effects of intergroup interactions should be to reinforce in-group identity and foster the perception of difference between groups.
Most scholars have conceptualized social media as a new form of communication technology whereby media and political elites can transmit information directly to the interested public in a dynamic, interactive, social way. Others have focused on the receiving end of this relationship, studying why people click on and read particular types of political news. However, this theorizing has not extended to assess the way in which the change in technology has altered the way in which citizens communicate with each other about politics, and the consequences of that change.
This book is written to fill that void.Footnote 4 I argue that the defining characteristics of political communication on the Facebook News Feed are uniquely suited to facilitate psychological processes of polarization: identity formation and reinforcement, biased information processing, and social inference and judgment. The confluence of features and norms on Facebook affects the interactions people have with each other, creating a communication ecosystem that facilitates negative and stereotyped evaluations of the Americans with whom people disagree. We will more fully explore these ideas in Chapters 2 to 4 and test the arguments in Chapters 5 to 8. But we must first address a pressing question about the focus of the book. Why should we care about political interaction on Facebook specifically?
Politics on Social Media
The far-reaching impact of the Internet in revolutionizing the way in which people access political information has been thoroughly chronicled (Bimber Reference Bimber2001; Norris Reference Norris2001; Jennings and Zeitner Reference Jennings and Zeitner2003; Kaye and Johnson Reference Kaye and Johnson2004; Kenski and Stroud Reference Kenski and Stroud2006). But the potential of the Internet in changing the way in which people interact with each other was not fully realized until the advent of Web 2.0, a series of technologies that facilitated social connections and interactivity between users. The term “social media” refers to an ever-changing and growing set of web-based interactive applications such as Facebook, Twitter, Instagram, Snapchat, YouTube, Tumblr, Flickr, Reddit, or Vine; any list constructed while writing this book will inevitably be incomplete or outdated by the time it is published.Footnote 5
However, not all of these social media technologies are created equally in terms of their utility for political interaction. The Pew Internet and American Life Project has studied extensively the way in which Americans incorporate social media into their political communication. And while candidates, elected officials, and the media have adopted a wide variety of social media and social networking sites, there is one platform that stands above all others in terms of its general popularity among American adults and its ubiquity for political engagement: Facebook. More than three times as many Americans have Facebook accounts as have Twitter accounts; five times as many Americans get political news in a given week from Facebook compared to those who get political news on Twitter (Pew Research Center 2014b).
Why Study Facebook?
The effects of using any particular social media platform result from the combination of the features of the site, called affordances, and the norms that develop around the way in which users employ those affordances. In addition to its widespread adoption for political communication, the book’s focus on one particular social media platform, Facebook, is justified in two primary ways. First, Facebook is the social media site with the highest concentration of affordances that could foster polarization. My argument is predicated on the way in which certain technological features foster particular psychological phenomena, and Facebook is an interesting test case because the confluence of these features permits interesting empirical tests of their interaction and reinforcement. Political communication on Facebook is conceptually related to more traditional forms of political behavior, but the ways in which the behaviors are different are pivotal to understanding the polarizing aspects of Facebook interaction. Identifying and describing these patterns within the Facebook ecosystem provides a baseline for future scholarship testing the scope conditions of the argument. This future research can build on the present findings by extending the work into different social media platforms containing different configurations of these key affordances – platforms that may not yet exist.
Second, Facebook has cornered enough of the market on social media communication in early twenty-first-century America that it merits scholarly attention. In this sense, Facebook is a unique social phenomenon worthy of study, even if the behavioral patterns detected on the site cannot be generalized to other forms of social media. As we’ll see in the next chapter, Facebook has become so integrated in the daily lives of the majority of Americans that the consequences of these interactions are important in their own right.
The Origins of Facebook
Mark Zuckerberg founded Facebook in 2004 while he was a student at Harvard University. The company’s legendary and heavily disputed origin story was the basis for the 2010 movie The Social Network. The site grew rapidly, and since 2006 anyone age 13 or older can become a registered user of Facebook. Facebook’s user base is unfathomably large. By mid-2016, the company reported that they had 1.65 billion users.
At the time of its initial public offering (IPO) in early 2012, Zuckerberg articulated a clear vision for the company. In a letter filed with the Securities and Exchange Commission (SEC) as part of the IPO, Zuckerberg wrote that Facebook “was built to accomplish a social mission – to make the world more open and connected … At Facebook, we’re inspired by technologies that have revolutionized how people spread and consume information.”Footnote 6 The constantly evolving features on the site are explicitly designed to accomplish these aims.
The letter in the SEC filing articulates that “personal relationships are the fundamental unit of our society” and emphasizes the extent to which the site is designed to facilitate interpersonal communication. The company’s primary aims are to enhance social connectivity among the mass public, but it aspires to do so with the goal of building a service to “give people the power to share and help them … transform many of our core institutions and industries.” From early in its development, Facebook incorporated features that fostered communication about one of the most prominent societal institutions: politics. It was not long after the company’s expansion beyond college campuses that candidates, elected officials, and opinion leaders alike adopted the site for political purposes. In addition to the “Political Affiliation” aspect of the profile page, every new iteration of the site – the introduction of the News Feed in 2006, the “like” button in 2009, and the “groups” feature in 2010 – was embraced quickly by those looking to capitalize on social connectivity to alter the operations of our political institutions.
Other sources have documented the rapidly growing use of social media by politicians (Golbeck et al. Reference Golbeck, Grimes and Rogers2010; Lassen and Brown Reference 270Lassen and Brown2010; Gainous and Wagner Reference Gainous and Wagner2013; Glassman et al. Reference Glassman, Straus and Shogan2013; Davis et al. Reference Davis, Bacha and Just2016; Evans and Clark Reference Evans and Clark2016; Jungherr Reference Jungherr2016; Kreiss Reference Kreiss2016). In 2006, 32 percent of US Senate candidates and 13 percent of US House candidates had profiles on Facebook (Gulati and Williams Reference Gulati and Williams2013); by 2011, 90 percent of representatives and senators were registered with the site. Every major news network has an active presence on the site. News sites that arose in the Web 1.0 era of blogs and niche news have readily taken to Facebook to expand their readership. Beyond these traditional players in the political communication environment, social media in particular has spawned the development of organizations that appear to be devoted primarily to generating and circulating content to the users who like their Facebook page.
In the next chapter, I characterize the aspects of the Facebook platform that contain the affordances and functionalities that are most crucial to facilitating Americans in forming negative judgments about their fellow citizens based on their political opinions. The argument is about the way in which the mass public interacts with each other; however, it is important not to lose sight of the fact that political elites, mass media, and political advocacy organizations are key players in generating the content that opinion leaders help circulate to the wider public.
Who Uses Facebook?
Previous research about political behavior on social media has largely focused on people who are active participants in creating and circulating political content, the minority of Facebook users like Susan in the vignette opening this chapter. While these individuals are critical in fostering a potentially polarizing information environment, this book focuses more on the audience: people who are “listening” and are exposed to political content, in most cases unintentionally. The book’s findings should thus be applicable to the population of American Facebook users, not just the most politically active among them. While this population is different in some ways from the American public at large, it is not subject to the same kinds of selection biases that structure the composition of the group of people most likely to encounter political information in other ways, such as by reading a daily newspaper or watching cable television.
As of mid-2016, 79 percent of all Internet-using American adults report using at least one social networking site (Greenwood et al. Reference Greenwood, Perrin and Duggan2016). While Pew’s data includes usage of a range of sites – including Twitter, Google Plus, Linked In, and Pinterest – in each of the surveys on which these data are based, the overwhelmingly most ubiquitous site is Facebook (Pew Research Center 2014b). Only 24 percent of Pew’s panel reports using Google Plus, and only 22 percent reports using Twitter. Statista reports that in 2016, there were 191.3 million Americans with Facebook accounts and only 66 million Americans with Twitter accounts. Figure 1.1 above reports the total percentage of Americans who have an account on a social networking site over time.
Growth in social media usage in America, 2005–2016
The lightest gray line shows the growth in the number of daily active users, in millions, of the Facebook site in North America (Canada and the United States) using data from Statista. The two dashed lines show the growth in the percentage of Americans who report that they have a social media account, and the percentage of Internet users who say they have a Facebook account, using data from Pew.
I build on the time series data Pew has generated about the growing use of Facebook with data from two original surveysFootnote 7 conducted in the spring of 2016 to assess the extent to which Facebook usage has permeated the daily lives of most Americans. The first survey included both Facebook users and non-Facebook users. The second survey followed up with the Facebook users and asked respondents both how frequently they logged into the site as well as how many times a day they scrolled through their News Feeds.Footnote 8
In a series of regression models shown in online Appendix C, the strongest correlates of whether a person has a Facebook account are age and gender. Consistent with the results found by Pew and others, women are more likely to have Facebook accounts, to spend time on Facebook, and to scroll through their News Feed more regularly. Age is inversely associated with Facebook account ownership and usage, although sizable proportions of Americans of all ages use Facebook regularly. Income is negatively associated with whether someone reports having a Facebook account, but neither race nor education are. While race is not associated with the decision to maintain a Facebook account, there is suggestive evidence that white Americans scroll through their News Feeds less frequently than do non-white Americans. Neither education level nor income is correlated with the frequency of Facebook usage or News Feed scrolling.
More consequential for an assessment of the potentially polarizing consequences of using social media are selection effects related to political characteristics. Partisan strength and political interest are correlated with whether someone has a Facebook account. But even supermajorities of political independents (70.5 percent) and the politically uninterested (70.7 percent) report that they have Facebook accounts compared to about 80 percent of the most partisan and most interested Americans. Ideological strength, past voting history, and political knowledge are not associated with opening a Facebook account.
While this might indicate a slight selection effect into who creates Facebook accounts,Footnote 9 none of these political variables is associated with the frequency or intensity with which someone uses Facebook. The factors that typically explain what kinds of people are most politically engaged are not the factors that best explain regular Facebook usage or News Feed scrolling, suggesting few selection biases shaping the composition of people who are active on the site on a daily basis. The only variables that are associated with heightened Facebook usage are those related to news engagement: whether people report that the Internet is their main source for news as well as the frequency with which people seek out news. Thus, the traditional correlates of political engagement and political information seeking – interest in politics, political knowledge, partisan strength, partisan leaning, or past voting behavior – are not associated with the frequency or intensity of a person’s daily exposure to the Facebook News Feed.
The survey was not exhaustive of every contribution to the decision to use Facebook, and there are almost certainly differences between people that are correlated with creating and using a Facebook account. But in terms of explaining what kinds of people “opt in” to regular usage of Facebook and repeated exposure to the News Feed, these results show that a broader swath of Americans than just the politically engaged public use Facebook. Unlike the findings of Arceneaux and Johnson (Reference Arceneaux and Johnson2013), who show that systematically different kinds of people are exposed to cable news programming, a similar pattern does not emerge for exposure to Facebook more broadly. As we’ll see in Chapter 5, there are important differences between those users who post political content and those who do not, but a very large majority of the American public is exposed to political content on Facebook in the course of their normal usage of the site.
Although some may argue that focusing on the hallmark characteristics of a single communication platform limits the generalizability of the argument and findings of this book, I disagree. A majority of Americans spend time on Facebook on a daily basis, and among those who do, the majority scrolls through their News Feed at least once a day. While not perfectly representative of the American public at large, a far larger group of Americans – with far fewer systematic biases shaping its composition – report being regularly engaged with their News Feed than report watching cable news or reading a newspaper regularly. And when they do, they cannot avoid encountering political content.
Political Engagement on Facebook
The Pew Internet and American Life Project has conducted the most thorough tracking of the growth of social media for political purposes, largely focused on the role of social media as a source for political information or a way to communicate about politics with others.
Large minorities of people report that they use social media in some capacity for political engagement, but few users report regular and sustained usage for these purposes. In other words, a much greater proportion of people report seeking or encountering political content on Facebook than report creating or circulating it. Most people don’t generate political content frequently – only 14 percent report that some, most, or all of the content they generate is about politics. Yet, a substantial proportion have generated content at some point, ranging from the 28 percent of users who have posted stories or articles to the 47 percent who have liked content other users have posted, with other forms of engagement falling somewhere in between.
Pew’s finding about the proportion of Facebook users who use the site for news (66 percent) translates to an estimate that 44 percent of the overall American adult population uses Facebook to get news (Gottfried and Shearer Reference Gottfried, Barthel and Shearer2016). That figure is equivalent to or greater than the proportion of Americans who report using cable news, local news, or newspapers to access information,Footnote 10 but the intensity of exposure appears to be higher on Facebook. In my survey, on average, people report seeing political information half the time they scroll through their News Feeds, even when they are not looking explicitly to find it. This high rate of exposure to political content might not matter if people see political information and communication on social media, but don’t pay any attention. However, Pew reports that of those users who report that they see political posts, 52 percent report paying some or a lot of attention, a figure that jumps to between 60 and 70 percent for the most ideologically consistent users in the sample (Pew Research Center 2014b).
Novel Behavior, Novel Questions
To date, political scientists have tested hypotheses about the antecedents and consequences of political engagement on social media within the framework used to study traditional political behaviors. There is a large body of literature trying to explain online engagement with factors such as political interest, efficacy, and knowledge and to disentangle the causal relationship between online and offline political behavior. Researchers have assessed exposure to political information primarily from a news-seeking and agenda-setting framework, applying the theories of selective exposure to study which information on social media users are most likely to click and the consequences of doing so on learning and issue salience.
This accumulated knowledge is useful, but it misses a fundamental point about Facebook. People use the site to stay connected to and learn about other people. The functionalities of the site are optimized for social engagement more generally, subsuming its use for political purposes, and these aspects of the site should inform our conceptualization and measurement of political engagement on Facebook. For most people, most of the time, exposure to political information is largely incidental and occurs in a context in which they are actively seeking social information about their connections.
Thus, we should reorient our focus on the consequences of posting and encountering political content on Facebook by deriving hypotheses based on more accurate assumptions about how and why people actually use the site. I propose the END Framework for the study of social media political interaction. END refers to the characteristics of a subset of content that circulates in a social media ecosystem like the Facebook News Feed: a personalized, quantified blend of politically informative expression, news, and discussion that is seamlessly interwoven in a single interface with non-political content. Creating and consuming content are novel and distinct political behaviors, albeit ones that fuse together characteristics of more traditional behaviors. Differences in the content itself, the context in which interactions occur, and the composition of the people who interact with each other produce unique expectations about the consequences of regularly using the News Feed for politically relevant outcomes.
Reframing our study of political interaction on Facebook by considering it first and foremost a social behavior highlights the importance of particular avenues of inquiry. Instead of focusing solely on consequences for offline political behavior or downstream policy attitudinal changes, we should think about consequences more directly related to the site’s primary function: expressing identity and learning about the lives of other people. I consider the most important of these potential social consequences to be the notion of psychological polarization. How is it that regular usage of the Facebook News Feed could facilitate the formation of negative attitudes toward people who hold different political views than our own?
First, Facebook fosters the ready availability of signals about others’ political preferences. As we will see, people are able to glean cues about the partisan identities of other users even from information that is not overtly political. Thus, the volume of information that is politically informative is much larger than the volume of information that is explicitly political. Because users engage with the site for the purposes of social inference, they are highly attuned to recognizing indications of how other people think and behave. Frequent usage of Facebook gives users increased practice in their social inference skills, bolstering their confidence in their abilities to map the socio-political terrain of their social network.
Second, social identity theory suggests that once group identity is recognized and made salient, intergroup interactions should lead to polarized evaluations differentiating the in-group from the out-group. Thus, the very same affordances on Facebook that are designed to bring people together – the ability to share content within the site from all corners of the Internet, the functionalities to like and comment on other people’s posts, and the quantification of this social feedback – can actually push people farther apart. Facebook users watch each other navigate a complex and polarizing information environment and have a window into the not-so-deliberative processes through which others acquire information and form their opinions. Within the broader context of twenty-first-century American political polarization, the identity-reinforcing aspects of Facebook serve to bolster assessments of the political in-group at the expense of denigration of the out-group.
Data and Methodology
The analysis in this book is primarily focused on Facebook users, but it is useful to compare those who use Facebook to those who don’t for two main reasons: (1) to test for any politically relevant selection effects into site usage; and (2) to test the key argument of the book – that Facebook users are more psychologically polarized than are non-Facebook users. To do this, on a large survey (n=3,030) dubbed the Political Discussion Survey, I included a number of demographic, social, and political variables in addition to a survey question about whether the respondent had a Facebook account. As we saw above, it turns out that selection into regular Facebook use is largely unrelated to the factors that drive political behavior. On the same survey, respondents answered three batteries of questions related to psychological polarization. Later in the book, these questions will allow us to assess whether Facebook users are more polarized than are non-Facebook users (hint: they are).
The respondents in this initial survey who indicated that they had Facebook accounts were re-contacted several weeks later to participate in a follow-up study that focused on their Facebook behavior. Throughout the book, this study is referred to as the END Framework Survey.Footnote 11 Respondents reported on their Facebook usage behaviors, the information to which they were exposed on the site, and additional facets of polarization. The full survey is available in the online appendices. As part of the END Framework Survey, I included the Social Connections Battery. Modeled after a technique used by scholars of political discussion, this component of the survey allows us a window into Facebook users’ assessments of their most politically prolific Facebook contacts. By coding the nature of the offline relationship between these contacts and the survey respondents, we get traction on the role of social distance in the judgment process of Facebook ties.
All of the studies described above are observational, aimed at characterizing patterns among a set of Facebook users sampled to mirror the composition of the American public. To assess the mechanisms that could have created these patterns, we use different types of studies.
The Inference Studies are a set of three studies conducted in February to April 2016 that are a type of classification task. Respondents were randomly assigned to assess between six and ten stimuli that were formatted to look like typical Facebook content. After each stimulus, they answered a set of questions about the user who supposedly posted the content to Facebook. At the end of the study, respondents answered a set of questions about their own Facebook usage. These studies feature prominently in Chapters 5 to 8 and form the core of the argument that Facebook users not only recognize the partisan identity of other users based on what they post, but also that they evaluate and judge members of their out-group in a manner consistent with the processes suggested by social identity theory.
In the fall of 2016 in the weeks leading up the presidential election, I conducted two additional studies in conjunction with one another. The first of these, the Accuracy Study, is an extension of the approach used in the inference studies. However, instead of using publicly available content for the study stimuli, I had a group of subjects (the “posters”) generate political Facebook content and answer a set of questions about their partisanship, ideology, and levels of political knowledge. A different sample (described momentarily) then assessed those users based on the content they generated in an identical framework to the inference studies. This permits a test of the accuracy of the judges’ perception compared to the reality of the posters’ traits.
The second additional study, the Generation Experiment, was an experiment conducted on a student sample. In the pre-survey, subjects first generated a piece of political content. Next, they completed the inference task where they were the “judges” in the Accuracy Study described above, except they were told they were evaluating content by other study participants. Then, before the subjects arrived in the lab for the second portion of the study, we formatted the content they’d generated to look like a Facebook post and randomly assigned them to receive high or low levels of social feedback on it. When they returned for the lab portion of the study, they received their “post” with the feedback that they were led to believe (temporarily, at least) had been generated by other study participants.
Appendix A includes an elaboration of all of these studies, as well as descriptive statistics of all the samples used. While the observational studies were conducted on subjects that can be described as approximating a nationally representative sample, the samples used in the other studies are decidedly convenience samples. A full consideration of the implications of this choice is in Appendix B.
Outline of the Book
In the chapters that follow, I build on the key points introduced in this chapter to offer compelling evidence that engaging with politically informative content on the Facebook News Feed has facilitated the processes that result in psychological polarization.
In the next chapter, we’ll take a step back and more fully consider the contemporary role of social media in the day-to-day lives of most Americans. We will explore how Facebook has affected other forms of social behavior to provide the background context necessary to appreciate why the END framework predicts particular consequences of News Feed engagement on attitudes toward other people. The chapter highlights six distinguishing features of the content on Facebook that contribute to social inference and the formation of affective attitudes, features that play a pivotal role in the studies later in the book. In Chapter 3, we’ll formalize the definition of News Feed political engagement – generation and consumption of politically informative content – at the intersection of political expression, information seeking, and discussion. In addition to articulating the key distinctions of the online manifestations of these behaviors, we’ll develop the theoretical groundwork about how to apply existing theories of political information processing and interpersonal interaction to these novel forms of behavior.
In Chapter 4, I fully outline the END Framework of communication, theorizing how generating and consuming politically informative News Feed content contributes to psychological polarization. After more thoroughly exploring what we know about these forms of polarization, we’ll consider how the defining characteristics of News Feed interaction facilitate distinct processes that contribute to the way in which people draw political inferences and make judgments about their social connections on the site. This theoretical chapter sets up the foundation for the empirical results in the second half of the book.
We will then turn to a puzzle that has emerged from previous research on social media. How is it possible that although most people report that they don’t post very much about politics, supermajorities of Facebook users report they have learned the political views of the people they are socially connected to on the site? Chapter 5 provides the first half of the answer: while a small minority of users generate explicitly political content that is partisan and opinionated, a much larger volume of content is implicitly political. Furthermore, there is a broad and varied definition of what kind of content is considered to be “political.” Certain topics are almost universally recognized as pertaining to politics, but a wider set of content can be considered politicized.
In Chapter 6, we’ll expand upon these notions to focus on informative signaling and inference on the News Feed. Facebook users – regardless of their level of political sophistication – are able to make inferences about the political inclinations of users who post political, as well as politicized and seemingly apolitical, content to the News Feed. The ability to infer underpins the assertion that the Facebook News Feed fosters people’s recognition of social and identity differences that align with political views. Inference forms the foundation for social evaluation and judgment, an important step in psychological polarization.
Chapter 7 reveals the extent of the cognitive biases that shape Facebook users’ inferences about others on the site. Our focus will be on three particular phenomena: the out-group homogeneity effect, perceived polarization, and the false consensus effect. Despite the fact that most Americans have moderate ideological opinions, Americans believe that there are vast differences between the two parties and their adherents. People attribute extreme and overly consistent ideological views to anonymous others based on the content of what they post, and they think the friends of people they disagree with are more extreme than their own friends. These perceptions of extremity are strongest among those who use Facebook most frequently. Additionally, the informational cues prolific on Facebook content facilitate biased estimates about the size of the political in-group, and generating political content reinforces users’ political identities when they receive social feedback from their network.
Social identity theory suggests that identity recognition and biased inference should go hand-in-hand with negative assessments of the out-group. In Chapter 8, we’ll explore the social evaluation and judgment that result from the way in which people process politically informative News Feed content. Users are considerably more negative about their friends with whom they disagree, judging harshly their knowledge levels and the credibility of the sources they use. These social inferences extend beyond the people to whom we are directly connected. People who generate and consume the most political content on Facebook hold the strongest negative stereotypes about Facebook users and members of the out-party more generally.
The final chapter of the book condenses its central arguments: the defining aspects of engagement with News Feed content – the behaviors at the intersection of political expression, information seeking, and political discussion on social media – are uniquely suited to foster the development and perpetuation of psychological polarization. We’ll consider the normative implications of these findings and whether changes to the affordances and norms on the site could reverse the trajectory on which we find ourselves.
