To register for this webinar, you must use an email address linked to a valid Zoom account (sign up free).
Speaker: Patrick Sturgis, London School of Economics
Abstract
Large language models have the potential to transform many aspects of survey research, but their greatest contribution may be in improving measurement rather than replacing existing methods.
This seminar will present findings from the Economic and Social Research Council (ESRC)-funded Survey Futures project ‘Integrating Generative AI in Question Design Evaluation and Testing’.
The presentation will focus on two projects that applied generative AI to longstanding challenges in survey methodology: questionnaire pretesting and occupational coding.
The first uses large language model (LLMs) to simulate cognitive interviewing and expert review in order to identify flaws in draft survey questions, while the second uses LLMs to assign Standard Occupational Classification (SOC2020) codes from respondents’ descriptions of their jobs.
Patrick Sturgis will present evidence on the accuracy, strengths and limitations of these approaches, discuss how they compare with conventional methods, and consider what they reveal about the opportunities and challenges of integrating generative AI into the survey lifecycle.
About the speaker
Patrick Sturgis is Professor of Quantitative Social Science at the London School of Economics and Political Science, where he is Head of the Department of Methodology. His research focuses on survey methodology, social measurement and quantitative social science, with particular interests in survey quality, public opinion, and the application of new technologies to social research.
He is a Fellow of the British Academy, a Fellow of the Academy of Social Sciences and an Honorary Life Fellow of the Market Research Society.
Attendance at City St George's events is subject to our terms and conditions.