philswatton / SocDem-Parties-EU-Replication

Replication materials for my paper "Social Democratic Party Positions on the EU: The Case of Brexit"

Replication mateirals for "Social democratic party positions on the EU: The case of Brexit"

This repository contains replication materials and R code for my article published in Party Politics, Social democratic party positions on the EU: The case of Brexit. The DOI for the article is: https://doi.org/10.1177/13540688231177557

This README contains an explanation of the contents of this repository, and notes some files which are not included.

To the best of my ability, I have checked over the code and made sure it remains correct for current usage. This paper was done largely in three different time periods during my PhD, with the analysis being almost wholly re-written at one point. The one part that has not been rewritten was the script implementing Bayesian-Aldrich-McKelvey scaling, which is included here but is more likely to be outdated for any user of these files using more recent versions of JAGS/rjags.

All of the scripts - particularly the more recent ones - contain many comments which hopefully along with this README should be sufficient to guide a reader through the steps taken. If not, or if you have any issues in using these replication materials where I can be of help, please visit my website to find my latest contact details: https://philswatton.github.io/. Please note I may take some time to respond!

Main Scripts

The main point of this repository is to offer the scripts used to prepare and analyse the data used in the paper. They are as follows:

1. 01_bes.R: Selects the relevant variables of choice, filters, and cleans the BES data used in the paper. Outputs from this script are saved in the 'data' folder.
2. 02_scaling.R: Applies Bayesian Aldrich-McKelvey scaling to the data, imputes values of '5' in likeability data, scales likeability data, and computes distances/same side variables. Outputs from this script are saved in the 'data' folder.
3. 03b_clogit.R: Runs conditional logit models and simulations for the main analysis and the additional analyses in the appendix. Outputs from this script are saved in the 'results' and 'data' folders.
4. 04_results.R: Runs functions that generate the plots and tables presented in the main analysis and the additional analyses in the appendix. Outputs from this script are saved in the 'plots' and 'tables' folders.

Additional Scripts

1. 03b_clogit_functions.R: Functions for the steps performed in 03b_clogit.R
2. 04b_output_functions.R: Functions for the steps performed in 04b_results.R
3. bam.jags: JAGS code for the Bayesian-Aldrich-McKelvey scaling. Taken from the University of Essex version of the asmcjr R package. Link to the package: https://github.com/uniofessex/asmcjr
4. labsim.R: Variables used during the simulation and useful to refer back to later on.

Folders

1. data: Contains data files generates in scripts 01 and 02. unscaled.rds contains BES data prior to scaling, while scaled.rds contains the BES data post-scaling. Finally, parties.rds contains the esimtated party positions from Aldrich-McKelvey scaling.
2. plots: All output plots. Plots beginning with sims are plots of simulations, with uns are UNS results, and with urs are URS results. Postfixes if present refer to which appendix result they belong to: left_shift and right_shift for shifts in the centre position, _sq for squared distance function results, and _no_ctrl for the results without controls. Finally,
3. results: Due to the size of the objects created, this folder does not exist in the repository, but will be crated if you run script 3. The folder contains the results from script 03. Each file is a .Rdata file with postfixes in the manner above. Each result is a list with two elements. Replacing results with the appropriate R object name, the first is results$models, which contains both models results$model$p_model for the proximity-only model and the second is results$model$c_model for the proximity plus categorisation model. The third element is results$model$test which contains the test dataset. The second element of results is results$sims, with results$sims$simProbs_p and results$sims$simProbs_c containing the simulated vote choices from the proximity-only and proximity plus categorisation models respectively.
4. tables: This contains the latex tables generated in script 04. reg.tex contains the main regression results while cv.tex contains the cross-validation table.

Not In This Repository

Users of this repository should be able to run scripts 2 and 3 without making any further changes. Script 3 must be run before script 4 is run. The original BES data files used are not however included, and thus script 1 cannot be run without making further downloads. They can be easily downloaded however. The files used are:

1. BES Internet Panel Wave 17, version 1.0-2, STATA file
2. BES 2019 General Election Results File, version 1.1, STATA file

They can be downloaded for free from the BES website: https://www.britishelectionstudy.com/

These are used in script 1. Note that as versions etc change over time, you may need to change the lines in script 1 to reflect new file names.