Replication Data for the Paper:
Autonomy and Accountability: Strategic Behaviour of German State Leaders During the COVID-19 Pandemic
by Salvatore Barbaro, Reyn van Ewijk, and Julia M. Rode
Public Choice (2025)
DOI: https://doi.org/10.1007/s11127-025-01357-3
This repository contains R, Stata, and Python scripts, along with embedded datasets, for replicating the analyses presented in the paper.
The main replication files are the two .do files. However, the R and Python scripts also reproduce most of the empirical evaluations reported in the paper.
In addition to the index values derived by human coders, we provide AI-generated index values obtained via the OpenAI API. The connection to this API was established through a Python script, while the matching to the existing dataset was implemented in the R replication file.
The following table summarises which analyses can be reproduced by which replication file.
| Assessment | MainStata.do | ReplicationR.R | ReplicationPython.py | TweetsStata.do | aiprompt2.py |
|---|---|---|---|---|---|
| Table 2 | x | x | x | ||
| AI-generated index values | x | ||||
| Table 3 (1)–(3) | x | x | |||
| Table 3 (4)–(5) | x | ||||
| Table 4 | x | ||||
| Table 5 | x | x | |||
| Table C1 | x | x | x |
The following table indicates which script produces each figure.
| Figure | Script |
|---|---|
| Fig. 1 | Fig01.R |
| Fig. 2 | Fig02.py (or Fig02.R) |
| Fig. 3 | Fig03.R |
| Fig. 4 | Fig03.R |
-
IDtable.csv
Lookup table linking AI-generated index values to the corresponding rows in the main dataset. -
Main_data_set_replication.dta
Main dataset. For convenience, both R and Python scripts import this file directly. -
Tweets_replication.dta
Matches the dates of statements to the corresponding Prime Minister Conferences. -
Statement23.csv
Contains statements from the second and third waves. This dataset is used byaiprompt2.py.
- Scripts were tested on Debian Linux.\
- The Stata files are compatible with Windows OS.
These are standalone scripts. Please check the header comments in each script for required library dependencies.
You can run the scripts: - Line-by-line in an editor such as RStudio, Positron, or Stata, or\
- As complete scripts via the terminal:
```bash Rscript <script_name.R>