neubig / BiasMonkey

BiasMonkey

This repository contains the data and analysis code for our paper:
Do LLMs exhibit human-like response biases? A case study in survey design
Lindia Tjuatja*, Valerie Chen*, Sherry Tongshuang Wu, Ameet Talwalkar, Graham Neubig

Contents

• Dataset
  • The original and modified questions used in our study can be found in here.
  • Original Pew questions were acquired from the OpinionsQA dataset (Santurkar et al. 2023)
• LLM Responses
  • Raw responses from LLMs are in results/<model>/*.pickle.
  • Formatted responses that are used in the analysis scripts are in results/<model>/csv/. The script to generate these files from the raw responses is format_results.py.
• Analysis
  • Main results
    • full_analysis.ipynb: Generates results for all models across response biases and non-bias perturbations.
    • correlation_human_behavior.ipynb: Computes human and model distributions for all relevant questions and wasserstein distance between the two distributions.
  • Additional results
    • uncertainty_analysis.ipynb: Generate uncertainty measures for all models across response biases and non-bias perturbations.
    • topic_analysis.ipynb: Visualizes model behavior broken down by topic.
    • steering_analysis.ipynb: Analyzes the effect of steering model behavior.
    • ext_gen_analysis.ipynb: Analyzes the effect of extended generation.