Judging the Judges: A Systematic Study of Position Bias in LLM-as-a-Judge

TL;DR

This study systematically investigates position bias in LLM-based judges across various tasks, revealing factors influencing bias and its impact on evaluation reliability, with implications for improving automated judging systems.

Contribution

It introduces three metrics to measure position bias and provides a comprehensive analysis of bias sources and variability across judges and tasks in LLM-based evaluation.

Findings

Position bias is statistically significant and varies across judges and tasks.

Bias is weakly affected by prompt length but strongly by solution quality gaps.

Analysis suggests potential dataset modifications to mitigate bias effects.

Abstract

LLM-as-a-Judge has emerged as a promising alternative to human evaluators across various tasks, yet inherent biases - particularly position bias, the tendency to favor solutions based on their position within the prompt - compromise its reliability. This exploratory study evaluates position bias in LLM judges across pairwise and list-wise comparison settings, introducing three metrics: repetition stability, position consistency, and preference fairness. Our experiments, involving 15 LLM judges across MTBench and DevBench with 22 tasks and approximately 40 solution-generating models, result in over 150,000 evaluation instances. We identify Judge-Level, Candidate-Level, and Task-Level factors contributing to bias. The findings confirm that position bias is not due to random chance and varies significantly across judges and tasks. While position bias is weakly influenced by the length of prompt components, it is strongly affected by the quality gap between solutions. Our agreement and disagreement analysis among judges further provides insights into the distribution of judging difficulty across the dataset, and highlights the potential for dataset modifications.