Can LLM-based 'silicon samples' accurately reflect human decisionmaking in conjoint tasks? The short answer is "not really" -- but there's more to the story.
I'll be presenting my paper on synthetic sampling in conjoint experiments tomorrow at 10! #APSA2025
