A Typological Approach to Big Questions in Political Social Media

• In Event: Social Media and Political Engagement

Abstract

Nearly two decades of research has examined social media as dynamic contexts for political communication. This work has found that social media sites not only deliver flows of news and political information, but also offer unique social environments in which users can engage in political discussion and expression. Yet researchers have often struggled to theorize and study how the diverse array of political behaviors on social media collectively influence broader engagement in political life. Instead, studies using survey data have tended to examine specific behaviors in isolation. For example, one study might examine the effects of reading news on social media, while another examines news article sharing. This tendency may ultimately hamper efforts to address “big questions” in digital political communication, including whether political social media use can help cultivate more knowledgeable and engaged citizens, and potentially impact political inequality.

This study attempts to address this challenge by empirically identifying types of social media users based on their patterns of political behavior. This typological approach allows for empirical tests of these “big questions” that account for how different combinations of political behaviors relate to political participation and knowledge. In doing so, we offer new insights into the role social media may play in shaping political inequality and demonstrate the utility of using a typological approach to study political communication on social media.

On a theoretical level, this study addresses three hypotheses that have been widely examined by previous research. First, we examine the “spillover hypothesis,” which predicts that using social media for political purposes will increase individuals’ participation in traditional offline forms of political participation (Boulianne, 2015). Second, we examine the “incidental learning hypothesis,” which predicts that social media use can increase political knowledge, even among those who do not actively seek political information (Lee & Kim, 2017). Finally, we examine the “stratification hypothesis,” which predicts that political use of social media will, in general, be stratified by socioeconomic status, such that those with higher status will have higher levels of political interest and therefore engage in politics more frequently on social media (Thorson et al., 2017). Rather than testing these hypotheses using indexes of political behavior on social media, as has been common in survey-based research, we classify users based on their political behavior on social media and use class membership as our independent variable. This offers a more nuanced test of each of the above hypotheses.

Using panel survey data from the 2016 U.S. election, this study assesses 8 political behaviors on Facebook, which broadly relate to either reception of political information (e.g., following a politician) or political expression (e.g., sharing a political video). We then use Latent Class Analysis (LCA) to identify classes of political Facebook users based on respondents’ probability of engaging in individual behaviors. We find that respondents in our dataset are best classified into four distinct classes: 1) the “all-engaged,” who have a high probability of engaging in all 8 political behaviors, 2) the “expressers” who have a high probability of commenting, posting and liking political content, 3) the “likers” who have a high probability of liking political content, and the “disengaged” who are not likely to engage in any of these behaviors.
We then use two-waves of panel data to test the three hypotheses outlined above. Examining the stratification hypothesis, we find that higher levels of education and political interest make respondents more likely to belong to the “all-engaged” or “liker” classes than the “disengaged” class. However, there is less evidence of stratification when it comes to “expressers,” who are more socioeconomically similar to the “disengaged.” Tests of the “incidental learning hypothesis” indicate that membership in any class of political Facebook use is associated with gains in political knowledge, even if that class does not indicate active efforts to consume political information. Finally, a test of the “spillover” hypothesis finds that, compared to respondents in the “disengaged” class, only membership in the “all-engaged” class is associated with gains in other forms of political participation.

Collectively these findings suggest that: a) political expression may be less stratified that other forms of political behavior on Facebook, b) even minimal political engagement on Facebook (e.g., liking a news article) may lead to political learning, and c) only users who both seek political information and express their political views have the potential for their social media use to “spillover” into other forms of political participation. This study lays the groundwork for further examination of political social media effects using a typological approach.

Authors

• Daniel Lane, UC Santa Barbara

• Brian E. Weeks, University of Michigan

• Nojin Kwak, University of Michigan