Fragile Preferences: A Deep Dive Into Order Effects in Large Language Models

• URL: http://arxiv.org/abs/2506.14092v2
• Date: Sun, 17 Aug 2025 03:47:06 GMT
• Title: Fragile Preferences: A Deep Dive Into Order Effects in Large Language Models
• Authors: Haonan Yin, Shai Vardi, Vidyanand Choudhary,
• Abstract summary: We present the first comprehensive study of position biases across multiple large language models (LLMs)<n>We find strong and consistent order effects, including a quality-dependent shift.<n>We also identify two previously undocumented biases in both human and machine decision-making.
• Score: 2.3936613583728064
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: 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 order 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 two previously undocumented biases in both human and machine decision-making: a centrality bias (favoring the middle position in triplewise comparisons) and 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, and that position biases are typically stronger than gender biases. 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.

Related papers

• Selective Mixup for Debiasing Question Selection in Computerized Adaptive Testing [50.805231979748434]
  Computerized Adaptive Testing (CAT) is a widely used technology for evaluating learners' proficiency in online education platforms.<n> Selection Bias arises because the question selection is strongly influenced by the estimated proficiency.<n>We propose a debiasing framework consisting of two key modules: Cross-Attribute Examinee Retrieval and Selective Mixup-based Regularization.
  arXiv  Detail & Related papers  (2025-11-19T08:55:01Z)
• Positional Bias in Binary Question Answering: How Uncertainty Shapes Model Preferences [8.389558188293641]
  We quantify and analyze positional bias across five large language models under varying degrees of answer uncertainty.<n>We observe that positional bias is nearly absent under low-uncertainty conditions, but grows exponentially when it becomes doubtful to decide which option is correct.
  arXiv  Detail & Related papers  (2025-06-30T11:30:23Z)
• The Root Shapes the Fruit: On the Persistence of Gender-Exclusive Harms in Aligned Language Models [91.86718720024825]
  We center transgender, nonbinary, and other gender-diverse identities to investigate how alignment procedures interact with pre-existing gender-diverse bias.<n>Our findings reveal that DPO-aligned models are particularly sensitive to supervised finetuning.<n>We conclude with recommendations tailored to DPO and broader alignment practices.
  arXiv  Detail & Related papers  (2024-11-06T06:50:50Z)
• Diverging Preferences: When do Annotators Disagree and do Models Know? [92.24651142187989]
  We develop a taxonomy of disagreement sources spanning 10 categories across four high-level classes. We find that the majority of disagreements are in opposition with standard reward modeling approaches. We develop methods for identifying diverging preferences to mitigate their influence on evaluation and training.
  arXiv  Detail & Related papers  (2024-10-18T17:32:22Z)
• Investigating Implicit Bias in Large Language Models: A Large-Scale Study of Over 50 LLMs [0.0]
  Large Language Models (LLMs) are being adopted across a wide range of tasks. Recent research indicates that LLMs can harbor implicit biases even when they pass explicit bias evaluations. This study highlights that newer or larger language models do not automatically exhibit reduced bias.
  arXiv  Detail & Related papers  (2024-10-13T03:43:18Z)
• Mitigating Selection Bias with Node Pruning and Auxiliary Options [11.835002896308545]
  Large language models (LLMs) often exhibit systematic preferences for certain answer choices when responding to multiple-choice questions.<n>This bias reduces the accuracy and reliability of LLM outputs, limiting their usefulness in decision-critical applications.<n>We introduce two methods: Bias Node Pruning (BNP), which prunes parameters that contribute to selection bias, and Auxiliary Option Injection (AOI), which introduces an answer choice to reduce bias in both white-box and black-box settings.
  arXiv  Detail & Related papers  (2024-09-27T15:53:54Z)
• Eliminating Position Bias of Language Models: A Mechanistic Approach [119.34143323054143]
  Position bias has proven to be a prevalent issue of modern language models (LMs)<n>Our mechanistic analysis attributes the position bias to two components employed in nearly all state-of-the-art LMs: causal attention and relative positional encodings.<n>By eliminating position bias, models achieve better performance and reliability in downstream tasks, including LM-as-a-judge, retrieval-augmented QA, molecule generation, and math reasoning.
  arXiv  Detail & Related papers  (2024-07-01T09:06:57Z)
• Going Beyond Popularity and Positivity Bias: Correcting for Multifactorial Bias in Recommender Systems [74.47680026838128]
  Two typical forms of bias in user interaction data with recommender systems (RSs) are popularity bias and positivity bias. We consider multifactorial selection bias affected by both item and rating value factors. We propose smoothing and alternating gradient descent techniques to reduce variance and improve the robustness of its optimization.
  arXiv  Detail & Related papers  (2024-04-29T12:18:21Z)
• Causality and Independence Enhancement for Biased Node Classification [56.38828085943763]
  We propose a novel Causality and Independence Enhancement (CIE) framework, applicable to various graph neural networks (GNNs) Our approach estimates causal and spurious features at the node representation level and mitigates the influence of spurious correlations. Our approach CIE not only significantly enhances the performance of GNNs but outperforms state-of-the-art debiased node classification methods.
  arXiv  Detail & Related papers  (2023-10-14T13:56:24Z)
• HANS, are you clever? Clever Hans Effect Analysis of Neural Systems [1.6267479602370545]
  Large Language Models (It-LLMs) have been exhibiting outstanding abilities to reason around cognitive states, intentions, and reactions of all people involved, letting humans guide and comprehend day-to-day social interactions effectively. Several multiple-choice questions (MCQ) benchmarks have been proposed to construct solid assessments of the models' abilities. However, earlier works are demonstrating the presence of inherent "order bias" in It-LLMs, posing challenges to the appropriate evaluation.
  arXiv  Detail & Related papers  (2023-09-21T20:52:18Z)
• Large Language Models Are Not Robust Multiple Choice Selectors [117.72712117510953]
  Multiple choice questions (MCQs) serve as a common yet important task format in the evaluation of large language models (LLMs) This work shows that modern LLMs are vulnerable to option position changes due to their inherent "selection bias" We propose a label-free, inference-time debiasing method, called PriDe, which separates the model's prior bias for option IDs from the overall prediction distribution.
  arXiv  Detail & Related papers  (2023-09-07T17:44:56Z)
• Delving into Identify-Emphasize Paradigm for Combating Unknown Bias [52.76758938921129]
  We propose an effective bias-conflicting scoring method (ECS) to boost the identification accuracy. We also propose gradient alignment (GA) to balance the contributions of the mined bias-aligned and bias-conflicting samples. Experiments are conducted on multiple datasets in various settings, demonstrating that the proposed solution can mitigate the impact of unknown biases.
  arXiv  Detail & Related papers  (2023-02-22T14:50:24Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.