Self-Preference Bias in Rubric-Based Evaluation of Large Language Models

TL;DR

This paper investigates self-preference bias in rubric-based evaluation of large language models, revealing its persistence and impact on model scoring, even with objective criteria and ensemble methods.

Contribution

It is the first study to analyze self-preference bias in rubric-based LLM evaluation, highlighting its effects and mitigation strategies.

Findings

SPB can cause up to 50% bias in objective rubric settings.

Ensembling judges reduces but does not eliminate SPB.

SPB can skew model scores by up to 10 points in subjective evaluations.

Abstract

LLM-as-a-judge has become the de facto approach for evaluating LLM outputs. However, judges are known to exhibit self-preference bias (SPB): they tend to favor outputs produced by themselves or by models from their own family. This skews evaluations and, thus, hinders model development, especially in settings of recursive self-improvement. We present the first study of SPB in rubric-based evaluation, an increasingly popular benchmarking paradigm where judges issue binary verdicts on individual evaluation criteria, instead of assigning holistic scores or rankings. Using IFEval, a benchmark with programmatically verifiable rubrics, we show that SPB persists even when evaluation criteria are entirely objective: among rubrics where generators fail, judges can be up to 50\% more likely to incorrectly mark them as satisfied when the output is their own. We also find that, similarly to other evaluation paradigms, ensembling multiple judges helps mitigate SPB, but without fully eliminating it. On HealthBench, a medical chat benchmark with subjective rubrics, we observe that SPB skews model scores by up to 10 points, a potentially decisive margin when ranking frontier models. We analyze the factors that drive SPB in this setting, finding that negative rubrics, extreme rubric lengths, and subjective topics like emergency referrals are particularly susceptible.