Replication materials: A Bad Workman Blames His Tweets

This github repository contains the replication materials for the paper "A Bad Workman Blames His Tweets: The Consequences of Citizens' Uncivil Twitter Use When Interacting With Party Candidates," forthcoming in the Journal of Communication, authored by Yannis Theocharis, Pablo Barberá, Zoltan Fazekas, Sebastian Adrian Popa, and Olivier Parnet.

Abstract: Existing studies focusing on politicians' adoption of Twitter have found that they use it primarily as a broadcasting tool. We argue that citizens' impolite and/or uncivil behavior is one possible explanation for such decisions. Social media conversations are rife with harassment and politicians are a prime target. This alters the incentive structure of engaging in dialogue on social media. We use Spanish, Greek, German, and U.K. candidates' tweets sent during the run-up to the recent European Parliament elections, and rely on automated text analysis and machine learning methods to measure their level of civility. Our contribution is an actor-oriented theory of political dialogue that incorporates Twitter's specific affordances, clarifying how and why Twitter's democratic promise may be limited.

You can read our Monkey Cage blog post at the The Washington Post for a summary of our paper.

This README file provides an overview of the replications files for the article.

• replication.r generates all the tables and figures in the text, with the exception of Table 3
• replication.r.log is the log file from running the previous R script, reproducing the results in the paper.
• replication-ts.do generates Table 3.
• replication-ts.log is the log file from running the previous STATA do-file, reproducing the results in the paper.
• monkeycage-figure.r replicates the Figure from the Monkey Cage blog post.
• COUNTRY.dta contains the candidate-levle datasets for each country, in .dta format.
• COUNTRY-time-series.dta contains the time-series datasets for each country, in .dta format.
• candidate_tweet_responses.csv contains the tweet-level dataset in .csv format.
• figureX.pdf is the set of figures from the paper, as generated by the R script.
• figure-MC.pdf and figure-MC.png are the figures from our Monkey Cage blog post.
• tabl3.tex is Table 3 from the paper, as generated by the STATA do-file.

Note: in compliance with Twitter's terms of service, we cannot provide the full text of the tweets. Instead, we provide the tweet IDs, as well as the variables we generated in our analysis.