# BenchMarker: An Education-Inspired Toolkit for Highlighting Flaws in Multiple-Choice Benchmarks

Source: [BenchMarker: An Education-Inspired Toolkit for Highlighting Flaws in Multiple-Choice Benchmarks](<https://www.sota2.com/research/paper/69896a28c97a0b3e3b5dc5ac>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-22T05:23:53.566Z

Authors: Nishant Balepur, Bhavya Rajasekaran, Jane Oh, Michael Xie, Atrey Desai, Vipul Gupta, Steven James Moore, Eunsol Choi, Rachel Rudinger, Jordan Lee Boyd-Graber

Published: 2026-02-05T21:57:50.000Z

## Abstract

Multiple-choice question answering (MCQA) is standard in NLP, but benchmarks lack rigorous quality control. We present BenchMarker, an education-inspired toolkit using LLM judges to flag three common MCQ flaws: 1) contamination: items appearing exactly online; 2) shortcuts: cues in the choices that enable guessing; and 3) writing errors: structural/grammatical issues based on a 19-rule education rubric. We validate BenchMarker with human annotations, then run the tool to audit 12 benchmarks, revealing: 1) flaws persist in MCQA benchmarks, especially automatically-made and crowdsourced data - we detect 47% of TruthfulQA appears online and 100% of HellaSwag violates multiple writing rules; 2) contaminated MCQs tend to inflate accuracy, while writing errors tend to lower it and change rankings beyond random; and 3) prior benchmark repairs address their targeted issues (i.e., lowering accuracy with LLM-written distractors), but inadvertently add new flaws (i.e. implausible distractors, many correct answers). Overall, flaws in MCQs degrade NLP evaluation, but education research offers a path forward. We release BenchMarker to bridge the fields and improve MCQA benchmark design.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Contamination detection on BenchMarker](<https://www.sota2.com/research/sota/contamination-detection-on-benchmarker.md>): Contamination Detection; BenchMarker; Accuracy: 71.2.

- [Writing flaw detection on BenchMarker Writing - In Domain, NLP](<https://www.sota2.com/research/sota/writing-flaw-detection-on-benchmarker-writing-in-domain-nlp.md>): Writing flaw detection; BenchMarker Writing - In Domain, NLP; Accuracy: not recorded.

- [Writing flaw detection on BenchMarker Writing - Out of Domain, Human](<https://www.sota2.com/research/sota/writing-flaw-detection-on-benchmarker-writing-out-of-domain-human.md>): Writing flaw detection; BenchMarker Writing - Out of Domain, Human; Accuracy: not recorded.

- [Shortcut detection on BenchMarker Shortcuts](<https://www.sota2.com/research/sota/shortcut-detection-on-benchmarker-shortcuts.md>): Shortcut detection; BenchMarker Shortcuts; Accuracy: not recorded.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2602.06221>)

[PDF](<https://arxiv.org/pdf/2602.06221.pdf>)