replication_dont_at_me
This repository contains replication materials for the article (Munger, 2020). Don't @ Me: Experimentally Reducing Partisan Incivility on Twitter.
All of the data provided has been calculated from word counts collected from subjects' Twitter accounts. In order to protect subjects' privacy, the raw text is not avaliable; even a single tweet can be enough to uniquely identify a user.
To replicate the figures in both the body of the text and the appendix, first run analyze_coded_tweets_total.R. This produces Table 1, Figure 9 as well as three datasets used in the rest of the analysis: subjects_70th.RData, subjects_90th.RData, and subjects_civil.RData.
The following list of R files labels the figures or tables they produce:
model_ols_pooled.R : Figures 4 and 15
model_ols_semipooled.R : Figures 5 and 11 and 15
model_ols.R : Figures 6 and 11
model_ols.R : Figures 7 and 17
model_interaction_effects.R : Figure 8
measure_validation.R : Figure 10 and Tables in Appendix C
negative_binomial_model.R : Figures 12 and 13
model_ols_loquacity.R : Figure 14
ideology_graph.R : Figure 16