Data and code for "Capturing Humans’ Mental Models of AI: An Item Response Theory Approach"

Published at FAccT 2023.

The repository is organized as follows:

• The code for running our Mechanical Turk experiment can be found in AMT.
• The trivia questions themselves and the raw results from both our pilot study and main experiment can be found in raw_data.
• Raw data is processed in data_processing.py, output (i.e., clean data, several formats) can be found in processed_data
• Data visualizations are generated in raw_data_visualization.ipynb
• Scripts for running the IRT models can be found in modeling. Output of these scripts is in the modeling_results folder.
• Processing of these results is in modeling_results.ipynb.