ECPR

Install the app

Install this application on your home screen for quick and easy access when you’re on the go.

Just tap Share then “Add to Home Screen”

ECPR

Install the app

Install this application on your home screen for quick and easy access when you’re on the go.

Just tap Share then “Add to Home Screen”

Polling Panic, Twitch on Twitter ꟷ Immediate Social Media Reactions to Polling Results

USA
Campaign
Candidate
Quantitative
Social Media
Public Opinion
Big Data
Marius Sältzer
Carl Von Ossietzky Universität Oldenburg
Marius Sältzer
Carl Von Ossietzky Universität Oldenburg

Abstract

Do polls influence politics? While this question has been posed for voters in terms of strategic voting, polls also create a connection to the voters from the perspective of political elites. Electoral closeness has been demonstrated to be an important driver of campaign activity, spending, responsiveness and negative campaigning. Does the constant updating of this knowledge influence the behavior of candidates in elections? Due to data limitations, analyzing instant reactions by politicians has not been feasible yet. The active use of the microblogging platform Twitter by US politicians on the other hand allows us to create complete time series of daily activity. But how would politicians react? Following the literature on negative campaigning, I argue that if election is endangered, politicians become more negative. Campaigning on Twitter should become more intense, engaged and possibly more aggressive. Instant Twitter data allows to generalize the effect presented in the literature for static races to the continuous updating function of polls. Continuous updates of reelection probabilities, as indicated by polls, should induce behavioral responses. Reactions should be driven by the influence of changes in polling results on reelection chances, being high for large relative change. The higher this marginal poll effect, the stronger reaction will be. I test reaction to polling results on a new data set of all Twitter contributions by House of Representative candidates during the 2018 midterm elections. I use time series cross sectional analysis of tweet content sentiment to isolate the effect of published polls. I find that candidates react to polling results by becoming more negative when the changes actually affect reelection chances. These findings sheds a new light on the constant availability of polling results and reflects the normative consequences of the service political science provides for society, as well as the incentive structures candidates face in elections.