Strategic Platform Differentiation on Facebook and Twitter in the 2016 Election

• In Event: Don't Be Cruel: Incivility in Media and Campaigns

Abstract

Political communication scholars increasingly highlight the need to understand elite discourse across social media platforms (Bossetta, 2018; Kreiss, Lawrence, & McGregor, 2018; Stier, Bleier, Lietz, & Strohmeir, 2018). Synthesizing the results of this early empirical work, we argue that two aspects of political campaigns’ social media messaging remain understudied. The first concerns the extent to which campaigns strategically differentiate their messages by social media platform, which may signal that campaigns communicate distinct messages to segments of the electorate. Second, scholars should also strive to uncover when and why the same messages are recycled across multiple platforms.

To advance this research agenda forward, we systematically test the extent to which campaigns engage in what we refer to as strategic platform differentiation: the practice of broadcasting political campaign messages on different social media platforms. We ask two overarching research questions:

RQ1: To what extent do political campaigns differentiate their messaging across social media platforms?

RQ2: What explains the similarity or difference in political campaigning across social media platforms?

To answer these questions, we compare the public Facebook and Twitter messages of the Hillary Clinton and Donald Trump campaigns during the 2016 US general election. Our analysis first combines computational string matching to quantify the extent of cross-platform versus platform-specific content. From this analysis, we find a relatively high degree of cross-platform broadcasting. Approximately one-quarter of the Trump campaign’s messages were largely identical across Facebook and Twitter, and 60% of Clinton’s Facebook messages were also present on Twitter. By analyzing the time differential between the cross-posts, we find that campaigns preferred posting to Twitter first; however, the majority of cross-platform messages were issued to both platforms within one hour. This short time difference suggests that campaigns may not be beta-testing messages on Twitter before posting to Facebook, a notion posited by studies that interview campaign staff but do not verify campaigners’ claims with trace data (Stromer-Galley, 2019; Kreiss et al., 2018).

To answer the second research question and better understand why campaigns differentiate their messages by platform, we developed lexicons and machine learning algorithms to categorize posts (n=4,434) along two dimensions: policy issues and strategic message function. Once the posts were classified by 8 policy issues and 7 message functions, we then performed automated content analyses, chi-square tests, and binomial logistic regressions to test which of these variables explains strategic platform differentiation.

Our results suggest that both cross-platform posting and strategic platform differentiation are influenced by the intended function of campaigns’ messages -- not policy issues. For both campaigns, our regressions show that calls to action were twice as likely to be cross-platform, whereas issues (particularly economic ones) were negatively associated with cross-platform posting. When removing cross-posts and testing Facebook- versus Twitter-specific messages, we found no crystal-clear pattern in how each campaign applied their strategizing. Clinton leaned on Twitter more heavily for posting about policy issues) and updating supporters about campaign information, whereas Trump used Twitter for attacks, calls for action, and campaign-related information. The most prominent example of strategic differentiation, obtained through the word frequency analyses, is the Clinton campaign’s reaching out to Hispanic audiences by posting in Spanish exclusively on Facebook. The Trump campaign, meanwhile, reserved Twitter for promoting mass media experiences on Fox News.

Overall, we interpret the main finding of our analysis to be this: campaigns differentiate their social media posts more along message strategy than policy topics. This finding aligns with those of Bode et al. (2016), who find that politicians’ Twitter posts have significantly less issue content than television ads. We observe that for campaigns, strategic messaging ultimately guides the differences in social media messaging across platforms, but differences in messaging have little to do with the substance of the candidates’ policies.

Authors

• Michael Bossetta, Lund University, Sweden

• Jennifer Stromer-Galley, Syracuse University

• Jeff Hemsley, Syracuse University