mlsanityData to 7 Oct 2026

Selection Bias Induced Spurious Correlations in Large Language Models

arXiv 2207.08982 ↗First posted 18 Jul 2022cs.CL cs.AIcited by 9

Not on the map. Why?

Abstract (arXiv)

In this work we show how large language models (LLMs) can learn statistical dependencies between otherwise unconditionally independent variables due to dataset selection bias. To demonstrate the effect, we developed a masked gender task that can be applied to BERT-family models to reveal spurious correlations between predicted gender pronouns and a variety of seemingly gender-neutral variables like date and location, on pre-trained (unmodified) BERT and RoBERTa large models. Finally, we provide an online demo, inviting readers to experiment further.

Key facts. Cited by 9 papers in the map (data through 7 Oct 2026). Builds on 6 of its 15 references that are in the map. First posted on arXiv 18 Jul 2022.

Builds on

6 of 15 references.

References · 6

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  2. 2019RoBERTa: A Robustly Optimized BERT Pretraining Approachcited by 13,266
  3. 2020Underspecification Presents Challenges for Credibility in Modern Machine Learningcited by 488
  4. 2018Gender Bias in Coreference Resolutioncited by 438
  5. 2020Measuring and Reducing Gendered Correlations in Pre-trained Modelscited by 222
  6. 2016Neural Text Generation from Structured Data with Application to the Biography Domaincited by 20

Built on by

Cited by year

Citing papers per year: 2022: 2; 2023: 1; 2024: 2; 2025: 4.

9 papers.

  1. 2023Causal Parrots: Large Language Models May Talk Causality But Are Not Causalcited by 146
  2. 2025Beyond Reward Hacking: Causal Rewards for Large Language Model Alignmentcited by 42
  3. 2024MM-SpuBench: Towards Better Understanding of Spurious Biases in Multimodal LLMscited by 33
  4. 2022KALM: Knowledge-Aware Integration of Local, Document, and Global Contexts for Long Document Understandingcited by 9
  5. 2025Debiasing Reward Models by Representation Learning with Guaranteescited by 6
  6. 2022Selection Collider Bias in Large Language Modelscited by 0
  7. 2024Evaluation of Large Language Models: STEM education and Gender Stereotypescited by 0
  8. 2025Recover Experimental Data with Selection Bias using Counterfactual Logiccited by 0
  9. 2025CoBA: Counterbias Text Augmentation for Mitigating Various Spurious Correlations via Semantic Triplescited by 0