Evans: Tweeting during the 2014 Mid-term Elections

Background

Most candidates running for political office use Twitter to disseminate information to the public. Like newsletters and campaign advertisements, Twitter offers a way for candidates to talk about what they have been doing to help their communities, discuss their plans for government if elected, and post their beliefs about important political topics. Over the past few years, scholars have turned to Twitter to understand the mood of the electorate. Some have investigated how politicians use the social networking site, while others have tried to test whether messages on Twitter can actually affect the political participation of citizens.

In this project, we investigated both the tone and content of tweets sent by candidates for the U.S. House and the U.S. Senate during the 2014 election. In particular, we were seeking answers to the following questions:

1. Who is tweeting and how much are they tweeting? Previous research has revealed that competitiveness and partisanship matters.
2. What important political topics are being discussed on Twitter by political candidates, and do these topics coincide with public opinion data? Some research has suggested that Twitter exposure can affect public opinion. Do the topics discussed by politicians follow the Gallup data on the “Most Important Problem” facing the country questionnaire? If so, which comes first: more discussion on Twitter or a move in public opinion?
3. Have female candidates changed in their use of Twitter since 2012? In research conducted on the 2012 election, Evans, Cordova, and Sipole (2014) show that women sent significantly more negative, campaign-specific, and mobilization tweets.
4. Are female candidates more likely to talk about specifically defined “female issues” (healthcare, education) on Twitter? Previous work has shown that news coverage of female candidates is more likely to stress “female issues” than news coverage about male candidates. Do female candidates discuss “female issues” more often than men on Twitter?
5. How do House candidate tweets differ from Senate candidate tweets?

Method

Each of the students in this project coded a dataset of approximately 20,000 tweets.
With some help from graduate students in my department, as well as my own coding, we have a dataset of approximately 140,000 tweets for candidates for the U.S. House and U.S. Senate.

We just submitted a proposal for the Midwest Political Science national conference on some of our results. We show that particular individual-level and campaign specific characteristics are related to whether and how often candidates go negative on Twitter. In particular, we show that there are differences based on competitiveness, incumbency, gender, and partisanship. The timing of the negative tweets is also explored. The title of our upcoming presentation is “#Mudslinging on Twitter: When Twitter Fingers Become Trigger Fingers.”