ESRA 2027 sessions by theme

Back to Overview of Sessions

Session Details

Generative AI, and particularly large language models (LLMs), offers new opportunities to support the full questionnaire design, evaluation and testing (QDET) lifecycle – from conceptualisation and item generation to questionnaire adaptation, diagnostic review, synthetic pretesting, AI-assisted cognitive interviewing, and question revision and optimisation. Yet researchers responsible for high-quality academic and official surveys remain appropriately cautious about the validity, reliability and practical utility of AI-assisted QDET tools, and about how they should be integrated into established survey development workflows. Realising the potential of generative AI therefore requires rigorous methodological research to determine where these tools add value, where they introduce new sources of error or bias, and how their performance varies across models, tasks, populations and survey contexts.

This session invites empirical and methodological contributions investigating the use of generative AI across the QDET lifecycle. We particularly welcome studies that evaluate and validate alternative models, compare AI-assisted approaches with established QDET methods, investigate human-AI workflows, or develop and test custom tools for questionnaire design and evaluation. Contributions from academic, government and other survey-research settings are encouraged, with a particular interest in research that advances the responsible use of generative AI in high-quality, general-population and probability-based surveys.