conjoint-experiment

What problem does it solve?

This Skill helps you design conjoint survey experiments and turn respondent profile-choice data into interpretable estimates of attribute effects (AMCEs), including marginal means, interactions, and subgroup heterogeneity.

Core Features & Use Cases

• Conjoint profile design: Generate randomized full/partial profile sets for tasks and respondents so each attribute level is comparably varied across tasks.
• AMCE estimation with clustered uncertainty: Estimate Average Marginal Component Effects using a linear probability model (OLS) with standard errors clustered by respondent to respect within-respondent correlation.
• Marginal means, interactions, and heterogeneity: Compute marginal means, estimate interaction effects, and re-estimate AMCEs by subgroup (e.g., party or ideology) to quantify preference differences.

Quick Start

Use the conjoint-experiment skill to estimate AMCEs from your CSV at /data/conjoint_survey_results.csv for outcome column chosen, clustering by respondent_id, and using the attributes listed in your dataset.