README – Pre-publication data and code deposit Status This is a pre-publication data and code deposit for the associated preprint on callback rates for conscripts and Ukraine veterans. It contains the dataset and Stata do-files required to reproduce the main numerical results and figures reported in the current preprint. Following journal publication, this repository will be updated (if necessary) to correspond to the final accepted version. 1. Contents The repository currently includes the following files: 1. Raw_Data.csv • Unit of observation: Application (120 rows, 60 paired applications). • Key variables: o pair_id: Numeric identifier for résumé pairs. o veteran_status: Indicator for applicant type (0 = conscript, 1 = Ukraine veteran). o callback: Binary indicator for whether the résumé received a callback. o industry: Job type (string). o city: Vacancy location (string). • Additional variables are identifiers or convenience fields not used in the main text. • This dataset is used throughout the paper and by all do-files listed below. 2. Replication.do • Master script that reproduces the core numerical results reported in the paper. • Main tasks: o Imports Raw_Data.csv. o Computes callback rates for conscripts and veterans. o Implements the Beta–binomial analysis of the callback-rate difference (Δ = p_C − p_U, where p_C is the callback probability for conscripts and p_U for veterans). o Estimates a Bayesian logistic regression using bayes: logit. o Estimates a linear probability model (LPM) with job-type and region controls. o Performs the classical two-sample proportion test. • Output includes point estimates, intervals, and p-values reported in the Results section and Table 1 of the preprint. 3. Figure 1.do • Generates the “Prior vs posterior distribution on delta” figure. • Uses a Beta(1,1) prior for each group and Beta–binomial posteriors to plot: o the prior distribution (dashed), and o the posterior distribution (solid) for Δ = p_C − p_U. 4. Figure 2.do • Generates the “Beta posterior vs Bayesian-logit MCMC” figure. • Plots the Beta–binomial posterior for Δ and overlays the posterior implied by the Bayesian logistic regression. • Illustrates the close agreement between the two Bayesian approaches. 5. Figure 3.do • Generates the “Delta vs cumulative data” (sequential evidence) figure. • Aggregates applications by pair and cumulative batches and, at each step, computes: o the Beta(1,1) posterior for Δ (posterior mean and 95% credible interval), and o the corresponding frequentist point estimate and 95% confidence interval. • Plots the sequential posterior mean and credible band alongside frequentist estimates, illustrating how evidence for a veteran penalty accumulates as more pairs are added.
2. Software and requirements • Statistical software: Stata 16 or later (also tested in Stata 17). • Additional packages: None. All scripts rely exclusively on standard Stata commands. • Random seeds: Fixed seeds are set in simulation-based scripts to ensure exact reproducibility of posterior draws and figures. 3. How to reproduce the results 1. Place Raw_Data.csv, Replication.do, and all figure do-files in the same working directory. 2. Open each .do file in Stata and, if necessary, adjust any cd commands to point to this directory. 3. Run Replication.do to reproduce the numerical results reported in the Results section and summary table. 4. Run Figure 1.do, Figure 2.do, and Figure 3.do to generate the three main figures used in the text. o Output figures are saved to the Visuals/ folder referenced in the paper. 4. Version and planned updates • Current version: Pre-publication / preprint replication package. • Planned updates: o If the analysis is revised during peer review, the dataset and do-files will be updated to match the accepted version. o Any changes will be documented in an updated README and in the repository metadata (version notes or changelog). 5. Contact Sergey Alexeev¹ ² ¹ Senior Research Associate, Nura Gili: Centre for Indigenous Programs; Co-Design Health Research and Innovation Team, UNSW Sydney, Ground Electrical Engineering Building, Kensington NSW 2033, Australia. ² Adjunct Senior Lecturer, NHMRC Clinical Trials Centre, Faculty of Medicine and Health, The University of Sydney, Camperdown NSW 2050, Australia. Corresponding author: Dr Sergey Alexeev Email:
[email protected] Additional emails:
[email protected];
[email protected] Phone: +61 457 871 540
