Fragile preferences: A deep dive into order effects in large language models
PNAS Nexus · 03.08.2026 · 5 (8)
DOI: 10.1093/pnasnexus/pgag246 ↗Kurz gefasst
Fachartikel in der Zeitschrift PNAS Nexus, online erschienen am 11.08.2026. Verfasst von Haonan Yin und 2 weiteren Autor:innen. Beteiligte Einrichtung: Department of Information Systems and Business Analytics, Ivy College of Business, Iowa State University, Ames, IA 50011, USA. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 82 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Large language models (LLMs) are increasingly deployed in decision-support systems for high-stakes domains such as hiring and university admissions, where choices often involve selecting among competing alternatives. While prior work has noted position biases in LLM-driven comparisons, these biases have not been systematically analyzed or linked to underlying preference structures. We present the first comprehensive study of position biases across multiple LLMs and two distinct domains: resume comparisons, representing a realistic high-stakes context, and color selection, which isolates position effects by removing confounding factors. We find strong and consistent order effects, including a quality-dependent shift: when all options are high quality, models favor the first option, but when quality is lower, they favor later options. We also identify a previously undocumented bias: a name bias, where certain names are favored despite controlling for demographic signals. To separate superficial tie-breaking from genuine distortions of judgment, we extend the rational choice framework to classify pairwise preferences as robust, fragile, or indifferent. Using this framework, we show that order effects can lead models to select strictly inferior options. These results indicate that LLMs exhibit distinct failure modes not documented in human decision-making. We also propose targeted mitigation strategies, including a novel use of the temperature parameter, to recover underlying preferences when order effects distort model behavior.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Information Systems and Business Analytics, Ivy College of Business, Iowa State University, Ames, IA 50011, USA.
- Department of Information Systems, Muma College of Business, University of South Florida, Tampa, FL 33620, USA.
- Department of Information Systems, Paul Merage School of Business, University of California, Irvine, Irvine, CA 92697, USA.
ORCID-Profile: Haonan Yin, Shai Vardi, Vidyanand Choudhary
Zitieren
APA
Yin, H., Vardi, S., & Choudhary, V. (2026). Fragile preferences: A deep dive into order effects in large language models. PNAS Nexus, 5(8). https://doi.org/10.1093/pnasnexus/pgag246
Vancouver
Yin H, Vardi S, Choudhary V. Fragile preferences: A deep dive into order effects in large language models. PNAS Nexus. 2026;5(8). doi:10.1093/pnasnexus/pgag246
BibTeX
@article{Yin2026,
author = {Yin, Haonan and Vardi, Shai and Choudhary, Vidyanand},
title = {Fragile preferences: A deep dive into order effects in large language models},
journal = {PNAS Nexus},
year = {2026},
volume = {5},
number = {8},
doi = {10.1093/pnasnexus/pgag246}
}
EUVIMED