ESRA 2027 sessions by theme

Back to Overview of Sessions

Session Details

The rapid advancement of generative AI (genAI) presents opportunities and challenges for survey research. While genAI tools can enhance efficiency in data production, analysis, and reporting, their integration raises critical questions about quality and reproducibility. This session aims at bringing together divers examples of genAI usage in survey research and explores the development of best practices and guidelines for leveraging genAI in this field, ensuring robustness, transparency, and reproducibility. It is also interested in exploring the limits of this new methodology.

We invite presentations that examine the construction of synthetic survey data with genAI. This might involve (1) AI persona generated data, (2) script-based generation of synthetic data based on meta- and paradata, (3) data generation based on multivariate distributions, or (4) based on individual data forecasting synthetic data for future or alternative scenarios.
Furthermore, we invite presentations that deal with the use of genAI in survey data analyses. Areas of interest include (1) causal inference, (2) measurement, (3) missing data treatment, or (4) multiple imputation, as well as (5) predictive modeling.

We are interested in case studies and practical examples that fall into one of these domains. Presentations should include a statement on how genAI can be validated with respect to the production of simulated data and statistical analyses. What are the trade-offs in terms of, e.g., reproducibility, generalizability, accuracy, interpretability, and computational efficiency?

By fostering collaborative dialogue among researchers and methodologists, this session aims to develop a starting point for the development of standards for the responsible use of genAI for synthetizing and analyzing survey data.