BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20251120T110349EST-82014RgrOk@132.216.98.100
DTSTAMP:20251120T160349Z
DESCRIPTION:Heng Zhang\n\nW.P. Carey Business School\n	Arizona State Univers
 ity\n\nLarge Language Models for Market Research: A Data-augmentation Appr
 oach\n\nDate: Friday\, November 21\, 2025\n	Time: 11:00 AM - 12:00 PM\n	Loca
 tion: Bronfman Building\, Room 045\n\n\nAbstract\n\nLarge Language Models 
 (LLMs) have transformed artificial intelligence by excelling in complex na
 tural language processing tasks. Their ability to generate human-like text
  has opened new possibilities for market research\, particularly in conjoi
 nt analysis\, where understanding consumer preferences is essential but of
 ten resource-intensive. Traditional survey-based methods face limitations 
 in scalability and cost\, making LLM-generated data a promising alternativ
 e. However\, while LLMs have the potential to simulate real consumer behav
 ior\, recent studies highlight a significant gap between LLM-generated and
  human data\, with biases introduced when substituting between the two. In
  this paper\, we address this gap by proposing a novel statistical data au
 gmentation approach that efficiently integrates LLM-generated data with re
 al data in conjoint analysis. This results in statistically robust estimat
 ors with consistent and asymptotically normal properties\, in contrast to 
 na\'ive approaches that simply substitute human data with LLM-generated da
 ta\, which can exacerbate bias. We further present a finite-sample perform
 ance bound on the estimation error. We validate our framework through an e
 mpirical study on COVID-19 vaccine preferences\, demonstrating its superio
 r ability to reduce estimation error and save data and costs by 24.9% to 7
 9.8%. In contrast\, naive approaches fail to save data due to the inherent
  biases in LLM-generated data compared to human data. Another empirical st
 udy on sports car choices validates the robustness of our results. Our fin
 dings suggest that while LLM-generated data is not a direct substitute for
  human responses\, it can serve as a valuable complement when used within 
 a robust statistical framework.\n
DTSTART:20251121T160000Z
DTEND:20251121T170000Z
LOCATION:Room 045\, Donald E. Armstrong Building\, CA\, QC\, Montreal\, H3A
  3L1\, 3420 rue McTavish
SUMMARY:Management Science Research Centre (MSRC) Seminar: Heng Zhang 
URL:/dobson/channels/event/management-science-research
 -centre-msrc-seminar-heng-zhang-368977
END:VEVENT
END:VCALENDAR