party2d/README.md

party2d

Code and processed model inputs for generating two-dimensional party-position estimates from text and expert data. The model combines manifesto and media text indicators with expert survey placements in a Bayesian dynamic item-response framework.

Repository contents

• src/r/ — scripts that prepare processed model inputs from source datasets.
• src/julia/ — Stan data preparation, model fitting, post-estimation, enrichment, and validation.
• models/ — Stan model specification.
• data/ — processed party-level inputs used by the model.
• metadata/ — data dictionary and source-support documentation.
• docs/ — raw data source documentation, coding decisions, and operational notes.

Processed inputs needed by the model are included in data/ so the estimation step can be reproduced from the model-ready data.

Running the pipeline

Run the full workflow with:

bash run_estimation.sh full

This executes the numbered workflow scripts:

bash scripts/01_prepare_data.sh
bash scripts/02_fit_model.sh
bash scripts/03_extract_estimates.sh
bash scripts/04_enrich_estimates.sh

The numbered scripts can also be run manually in that order. scripts/05_validate_estimates.sh runs validation checks after estimates have been generated.

The Bayesian model is computationally expensive. The production run used 4 cores on an AMD Ryzen 9 7945HX and took 60,372 seconds, approximately 16 hours 46 minutes.

If model output is already available, rebuild estimates without refitting Stan:

bash run_estimation.sh reuse

reuse reruns source-data processing, post-estimation, and enrichment while skipping the Stan fitting step.

To check the local setup without fitting the model, run:

bash run_estimation.sh dry-run

Data inputs

The model-ready inputs are included under data/.

Original raw source files are not redistributed. See docs/RAW_DATA_SOURCES.md for the list of original data sources, access information, and expected local filenames for regenerating the processed inputs.

Output variables

The two position variables are scaled from 0 to 1:

• economic_lr: economic left to economic right.
• galtan: cosmopolitan/socially liberal to traditionalist/nationalist.

Column definitions are in metadata/data_dictionary.csv.