mlsanityData to 14 Oct 2026

Large Language Models Sensitivity to The Order of Options in Multiple-Choice Questions

arXiv 2308.11483 ↗First posted 22 Aug 2023cs.CL cs.AI cs.LGcited by 263

Not on the map. Why?

Cited by 27 in LLM reasoning, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

Large Language Models (LLMs) have demonstrated remarkable capabilities in various NLP tasks. However, previous works have shown these models are sensitive towards prompt wording, and few-shot demonstrations and their order, posing challenges to fair assessment of these models. As these models become more powerful, it becomes imperative to understand and address these limitations. In this paper, we focus on LLMs robustness on the task of multiple-choice questions -- commonly adopted task to study reasoning and fact-retrieving capability of LLMs. Investigating the sensitivity of LLMs towards the order of options in multiple-choice questions, we demonstrate a considerable performance gap of approximately 13% to 75% in LLMs on different benchmarks, when answer options are reordered, even when using demonstrations in a few-shot setting. Through a detailed analysis, we conjecture that this sensitivity arises when LLMs are uncertain about the prediction between the top-2/3 choices, and specific options placements may favor certain prediction between those top choices depending on the question caused by positional bias. We also identify patterns in top-2 choices that amplify or mitigate the model's bias toward option placement. We found that for amplifying bias, the optimal strategy involves positioning the top two choices as the first and last options. Conversely, to mitigate bias, we recommend placing these choices among the adjacent options. To validate our conjecture, we conduct various experiments and adopt two approaches to calibrate LLMs' predictions, leading to up to 8 percentage points improvement across different models and benchmarks.

Key facts. Cited by 263 papers in the map (data through 14 Oct 2026). Builds on 25 of its 27 references that are in the map. First posted on arXiv 22 Aug 2023.

Builds on

25 of 27 references.

References · 25

  1. 2023GPT-4 Technical Reportcited by 18,396
  2. 2022Training language models to follow instructions with human feedbackcited by 15,971
  3. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 14,132
  4. 2019Sentence-BERT: Sentence Embeddings using Siamese BERT-Networkscited by 9,170
  5. 2020Measuring Massive Multitask Language Understandingcited by 7,963
  6. 2022PaLM: Scaling Language Modeling with Pathwayscited by 5,326
  7. 2023Universal and Transferable Adversarial Attacks on Aligned Language Modelscited by 3,104
  8. 2022Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language modelscited by 1,998
  9. 2022Language Models (Mostly) Know What They Knowcited by 1,856
  10. 2018CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledgecited by 1,822
  11. 2023Jailbroken: How Does LLM Safety Training Fail?cited by 1,651
  12. 2021Calibrate Before Use: Improving Few-Shot Performance of Language Modelscited by 1,392
13 more
  1. 2023PaLM 2 Technical Reportcited by 1,162
  2. 2023Large Language Models are not Fair Evaluatorscited by 1,085
  3. 2022Ignore Previous Prompt: Attack Techniques For Language Modelscited by 832
  4. 2022Teaching Models to Express Their Uncertainty in Wordscited by 751
  5. 2023Large Language Models are Zero-Shot Rankers for Recommender Systemscited by 309
  6. 2022Large Language Models are Better Reasoners with Self-Verificationcited by 307
  7. 2022Leveraging Large Language Models for Multiple Choice Question Answeringcited by 203
  8. 2022Can large language models reason about medical questions?cited by 194
  9. 2023Fundamental Limitations of Alignment in Large Language Modelscited by 155
  10. 2023Adversarial Demonstration Attacks on Large Language Modelscited by 63
  11. 2019From 'F' to 'A' on the N.Y. Regents Science Exams: An Overview of the Aristo Projectcited by 33
  12. 2023Large Language Models (GPT) Struggle to Answer Multiple-Choice Questions about Codecited by 32
  13. 2020Diverse Ensembles Improve Calibrationcited by 23

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2023: 20; 2024: 65; 2025: 69; 2026: 109.

263 papers.

Sort:

By area:

  1. 2024LLM Evaluators Recognize and Favor Their Own Generationscited by 686
  2. 2023RLAIF vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedbackcited by 625
  3. 2023Large Language Models Are Not Robust Multiple Choice Selectorscited by 486
  4. 2024LLMs-as-Judges: A Comprehensive Survey on LLM-based Evaluation Methodscited by 465
  5. 2023Can Generalist Foundation Models Outcompete Special-Purpose Tuning? Case Study in Medicinecited by 329
  6. 2023Evaluating Large Language Models at Evaluating Instruction Followingcited by 302
  7. 2024Judging the Judges: Evaluating Alignment and Vulnerabilities in LLMs-as-Judgescited by 205
  8. 2023Do LLMs exhibit human-like response biases? A case study in survey designcited by 180
  9. 2024Aligning with Human Judgement: The Role of Pairwise Preference in Large Language Model Evaluatorscited by 149
  10. 2024CVQA: Culturally-diverse Multilingual Visual Question Answering Benchmarkcited by 120
  11. 2023Evil Geniuses: Delving into the Safety of LLM-based Agentscited by 106
  12. 2024MCS-SQL: Leveraging Multiple Prompts and Multiple-Choice Selection For Text-to-SQL Generationcited by 105
38 more
  1. 2023Batch Calibration: Rethinking Calibration for In-Context Learning and Prompt Engineeringcited by 94
  2. 2025Words or Vision: Do Vision-Language Models Have Blind Faith in Text?cited by 89
  3. 2024Open-LLM-Leaderboard: From Multi-choice to Open-style Questions for LLMs Evaluation, Benchmark, and Arenacited by 76
  4. 2024MovieChat+: Question-aware Sparse Memory for Long Video Question Answeringcited by 73
  5. 2024Artifacts or Abduction: How Do LLMs Answer Multiple-Choice Questions Without the Question?cited by 71
  6. 2024Perception Tokens Enhance Visual Reasoning in Multimodal Language Modelscited by 71
  7. 2024Can multiple-choice questions really be useful in detecting the abilities of LLMs?cited by 69
  8. 2024Position: LLM Unlearning Benchmarks are Weak Measures of Progresscited by 63
  9. 2024Introducing v0.5 of the AI Safety Benchmark from MLCommonscited by 62
  10. 2026Rethinking Rubric Generation for Improving LLM Judge and Reward Modeling for Open-ended Taskscited by 62
  11. 2023SALMON: Self-Alignment with Instructable Reward Modelscited by 55
  12. 2025LEXam: Benchmarking Legal Reasoning on 340 Law Examscited by 54
  13. 2024Virtual Personas for Language Models via an Anthology of Backstoriescited by 51
  14. 2024Evaluating LLMs at Detecting Errors in LLM Responsescited by 50
  15. 2024Preference Learning Algorithms Do Not Learn Preference Rankingscited by 48
  16. 2024LongGenBench: Benchmarking Long-Form Generation in Long Context LLMscited by 47
  17. 2023Hijacking Large Language Models via Adversarial In-Context Learningcited by 45
  18. 2024Talking Heads: Understanding Inter-layer Communication in Transformer Language Modelscited by 41
  19. 2023Self-Assessment Tests are Unreliable Measures of LLM Personalitycited by 39
  20. 2025Do LLM Evaluators Prefer Themselves for a Reason?cited by 38
  21. 2024Answer, Assemble, Ace: Understanding How LMs Answer Multiple Choice Questionscited by 35
  22. 2025CausalVQA: A Physically Grounded Causal Reasoning Benchmark for Video Modelscited by 35
  23. 2024CalibraEval: Calibrating Prediction Distribution to Mitigate Selection Bias in LLMs-as-Judgescited by 34
  24. 2025Investigating Non-Transitivity in LLM-as-a-Judgecited by 34
  25. 2024GSR-BENCH: A Benchmark for Grounded Spatial Reasoning Evaluation via Multimodal LLMscited by 33
  26. 2024GenderBias-VL: Benchmarking Gender Bias in Vision Language Models via Counterfactual Probingcited by 31
  27. 2023On Context Utilization in Summarization with Large Language Modelscited by 30
  28. 2024Serial Position Effects of Large Language Modelscited by 30
  29. 2023An In-depth Look at Gemini's Language Abilitiescited by 28
  30. 2024Do LLMs estimate uncertainty well in instruction-following?cited by 26
  31. 2025Break the Checkbox: Challenging Closed-Style Evaluations of Cultural Alignment in LLMscited by 26
  32. 2024Fairer Preferences Elicit Improved Human-Aligned Large Language Model Judgmentscited by 25
  33. 2024A Study on Large Language Models' Limitations in Multiple-Choice Question Answeringcited by 22
  34. 2023LoFT: Local Proxy Fine-tuning For Improving Transferability Of Adversarial Attacks Against Large Language Modelcited by 21
  35. 2026Large Language Model Reasoning Failurescited by 21
  36. 2026Brittlebench: Quantifying LLM robustness via prompt sensitivitycited by 21
  37. 2023A Systematic Comparison of Syllogistic Reasoning in Humans and Language Modelscited by 20
  38. 2024Unveiling Selection Biases: Exploring Order and Token Sensitivity in Large Language Modelscited by 20

50 most cited of 263 All 263 (JSON)