Replication materials for Thomas Davidson, 2024, "Audience Engagement and the Dynamics of Online Activism: Far-Right Mobilization on Facebook", Mobilization 28(4): 445-470.
A pre-print is available here: https://osf.io/preprints/socarxiv/rwfjc
All necessary data to replicate the results are stored in the data directory. This also includes various intermediate datasets created by running the code below.
The weekly_dataset.csv file contains the final version of the data used for time series modeling. The bf_posts_raw.csv file contains the original text of BF's Facebook posts.
All code is contained in the code directory. Specific instructions are included within each file. The files are either in R or RMarkdown format.
The code used to run the topic model is contained in estimate_topic_model.Rmd. The fitted model is stored in the topic_model directory. The process-topic-model-results.Rmd script is used to obtain weekly topic proportions.
The main time series regression analyses can be reproduced by running main-results.Rmd. This script produces Tables 2-4 and Figure 5, as well as additional supplementary material. Note that the tables require some additional formatting to match published versions.
The negative binomial regression models estimating the effect of topic proportions and engagement bait on engagement can be reproduced by running post-level-regressions.R. This produces Figure 6 and the full table reported in the appendix.
The additional-visualizations.Rmd script can be used to recreate Figures 2-4 and additional supplementary figures.
All tables are stored in tables and figures are stored in figures.