Budget, Cost, or Both? An Empirical Exploration of Mechanisms in Quadratic Surveys Data Analysis

This repository contains notebooks and data used for the paper ``Budget, Cost, or Both? An Empirical Exploration of Mechanisms in Quadratic Surveys'' published in Collective Intelligence 2025. The main notebooks in the repository load models created elsewhere and generate the figures found in output/ and `plots/`.

Repository Structure

• QV_intensity_analysis.ipynb – Post modeling analysis notebook for intensity conditions.
• QV_sign_analysis.ipynb – Post modeling analysis notebook for sign conditions.
• modeling_notebook/ – Public versions of the notebooks used to fit the models.
• data/ – Experiment data files used by the notebooks.
• output/ and plots/ – Generated CSVs and PDF figures.

Reproducing the Results

The analysis notebooks expect pre‑computed model traces. These traces are large and are not stored in this repository. To generate them:

1. Clone and run the separate modeling repository that accompanies this analysis.
2. Execute the notebooks inside that repository to fit the models using the data found here under data/.
3. Note the location where the model files (.trace NetCDF files) are saved and update the corresponding paths in the analysis notebooks.

After the models are created, edit the MODEL_PATH variable near the top of each root notebook so it points to the folder containing the saved models. You may also adjust DATA_PATH, OUTPUT_PATH, and PLOT_PATH if your directory layout differs.

Once the paths are configured, run the notebooks in this repository to reproduce the plots and CSV outputs.

Dependencies

The notebooks were developed with Python and require packages such as numpyro, arviz, pandas, and seaborn. The modeling notebooks include cells that install these packages via pip. It is recommended to use a virtual environment or conda environment to manage dependencies prior to running the notebooks.

Cite

Please cite the paper should the data or models be useful for your work:

@inproceedings{cheng2025budget,
  author    = {Ti-Chung Cheng, Tiffany Wenting Li, Karrie Karahalios, and Hari Sundaram},
  title     = {Budget, Cost, or Both? An Empirical Exploration of Mechanisms in Quadratic Surveys},
  booktitle = {Proceedings of the Collective Intelligence Conference (CI 2025)},
  year      = {2025},
  month     = {August},
  pages     = {1--13},
  address   = {San Diego, CA, USA},
  publisher = {ACM},
  doi       = {10.1145/3715928.3737474},
  isbn      = {979-8-4007-1489-4/2025/08},
  url       = {https://doi.org/10.1145/3715928.3737474}
}