Why averaging LLM benchmark scores is fundamentally broken

A new study finds that averaging benchmark scores produces misleading rankings when evaluation data is sparse and item difficulty varies widely, with Spearman rank correlation dropping from 1.000 to 0.809 under such conditions. The researchers show that Item Response Theory (IRT) maintains near-perfect accuracy (ρ ≥ 0.996) across all tested scenarios, including in autonomous vehicle safety and cybersecurity domains.

Computer Science Machine Learning Submitted on 11 May 2026 Title:The Scaling Law of Evaluation Failure: Why Simple Averaging Collapses Under Data Sparsity and Item Difficulty Gaps, and How Item Response Theory Recovers Ground Truth Across Domains View PDF /pdf/2605.11205 HTML experimental https://arxiv.org/html/2605.11205v1 Abstract:Benchmark evaluation across AI and safety-critical domains overwhelmingly relies on simple averaging. We demonstrate that this practice produces substantially misleading rankings when two conditions co-occur: 1 the evaluation matrix is sparse and 2 items vary substantially in difficulty. Through controlled simulation experiments across four domains -- NLP GLUE , clinical drug trials, autonomous vehicle safety, and cybersecurity -- we show that Spearman rank correlation $\rho$ between simple-average rankings and ground-truth rankings degrades from $\rho = 1.000$ at 100% coverage to $\rho = 0.809$ at 67% coverage with high difficulty heterogeneity mean over 20 seeds . A standard two-parameter logistic 2PL Item Response Theory IRT model maintains $\rho \geq 0.996$ across all conditions. A 150-condition grid sweep over sparsity $S \in 0, 0.70 $ and difficulty gap $D \in 0.5, 5.0 $ confirms that ranking error forms a failure surface with a strong $S \times D$ interaction $\gamma 3 = +0.20$, $t = 13.05$ , while IRT maintains $\rho \geq 0.993$ throughout. We discuss implications for Physical AI benchmarking, where evaluation matrices are often incomplete and difficulty gaps are extreme. References & Citations Loading... Bibliographic and Citation Tools Bibliographic Explorer What is the Explorer? https://info.arxiv.org/labs/showcase.html arxiv-bibliographic-explorer Connected Papers What is Connected Papers? https://www.connectedpapers.com/about Litmaps What is Litmaps? https://www.litmaps.co/ scite Smart Citations What are Smart Citations? https://www.scite.ai/ Code, Data and Media Associated with this Article alphaXiv What is alphaXiv? https://alphaxiv.org/ CatalyzeX Code Finder for Papers What is CatalyzeX? https://www.catalyzex.com DagsHub What is DagsHub? https://dagshub.com/ Gotit.pub What is GotitPub? http://gotit.pub/faq Hugging Face What is Huggingface? https://huggingface.co/huggingface ScienceCast What is ScienceCast? https://sciencecast.org/welcome Demos Recommenders and Search Tools Influence Flower What are Influence Flowers? https://influencemap.cmlab.dev/ CORE Recommender What is CORE? https://core.ac.uk/services/recommender IArxiv Recommender What is IArxiv? https://iarxiv.org/about arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs https://info.arxiv.org/labs/index.html .