# BiasCause: Evaluate Socially Biased Causal Reasoning of Large Language Models

arXiv: https://arxiv.org/abs/2504.07997 · HTML: https://mlsanity.com/paper/2504.07997/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Causal reasoning in LLMs, one of 17 directions in LLM reasoning.
- Builds on 21 of its 58 references that are in the map.
- First posted on arXiv 8 Apr 2025.

## Placed in

- LLM reasoning / [Causal reasoning in LLMs](https://mlsanity.com/lr/causal-reasoning-in-llms/)

## Builds on (most cited first, 21 of 21)

- [Self-Consistency Improves Chain of Thought Reasoning in Language Models](https://mlsanity.com/paper/2203.11171/) (2022)
- [TruthfulQA: Measuring How Models Mimic Human Falsehoods](https://mlsanity.com/paper/2109.07958/) (2021)
- [BloombergGPT: A Large Language Model for Finance](https://mlsanity.com/paper/2303.17564/) (2023)
- [Measuring Faithfulness in Chain-of-Thought Reasoning](https://mlsanity.com/paper/2307.13702/) (2023)
- [Gender Bias in Coreference Resolution](https://mlsanity.com/paper/1804.09301/) (2018)
- ["Kelly is a Warm Person, Joseph is a Role Model": Gender Biases in LLM-Generated Reference Letters](https://mlsanity.com/paper/2310.09219/) (2023)
- [Can Large Language Models Infer Causation from Correlation?](https://mlsanity.com/paper/2306.05836/) (2023)
- [A Mechanistic Interpretation of Arithmetic Reasoning in Language Models using Causal Mediation Analysis](https://mlsanity.com/paper/2305.15054/) (2023)
- [Understanding Causality with Large Language Models: Feasibility and Opportunities](https://mlsanity.com/paper/2304.05524/) (2023)
- [CHBias: Bias Evaluation and Mitigation of Chinese Conversational Language Models](https://mlsanity.com/paper/2305.11262/) (2023)
- [Prompting Fairness: Integrating Causality to Debias Large Language Models](https://mlsanity.com/paper/2403.08743/) (2024)
- [Does Reasoning Introduce Bias? A Study of Social Bias Evaluation and Mitigation in LLM Reasoning](https://mlsanity.com/paper/2502.15361/) (2025)
- [Fairness through Difference Awareness: Measuring Desired Group Discrimination in LLMs](https://mlsanity.com/paper/2502.01926/) (2025)
- [ChatGPT Based Data Augmentation for Improved Parameter-Efficient Debiasing of LLMs](https://mlsanity.com/paper/2402.11764/) (2024)
- [Does Safety Training of LLMs Generalize to Semantically Related Natural Prompts?](https://mlsanity.com/paper/2412.03235/) (2024)
- [Fairness Testing of Large Language Models in Role-Playing](https://mlsanity.com/paper/2411.00585/) (2024)
- [Evaluating and Mitigating Social Bias for Large Language Models in Open-ended Settings](https://mlsanity.com/paper/2412.06134/) (2024)
- ["Im not Racist but...": Discovering Bias in the Internal Knowledge of Large Language Models](https://mlsanity.com/paper/2310.08780/) (2023)
- [How Strategic Agents Respond: Comparing Analytical Models with LLM-Generated Responses in Strategic Classification](https://mlsanity.com/paper/2501.16355/) (2025)
- [REFINE-LM: Mitigating Language Model Stereotypes via Reinforcement Learning](https://mlsanity.com/paper/2408.09489/) (2024)
- [Gender Bias Evaluation in Text-to-image Generation: A Survey](https://mlsanity.com/paper/2408.11358/) (2024)

## Built on by (most cited first, 2 of 2)

- [SCRuB: Social Concept Reasoning under Rubric-Based Evaluation](https://mlsanity.com/paper/2605.06444/) (2026)
- [Side-by-side Comparison Amplifies Dialect Bias in Language Models](https://mlsanity.com/paper/2605.24384/) (2026)

## Abstract (arXiv)

While large language models (LLMs) play increasingly significant roles in society, research shows they continue to generate content that reflects social bias against sensitive groups. Existing benchmarks effectively identify these biases, but a critical gap remains in understanding the underlying reasoning processes that produce them. This paper addresses this gap by evaluating the causal reasoning of LLMs when answering socially biased questions. We propose a formal schema that categorizes causal reasoning into three types (mistaken, biased, and contextually-grounded). We then synthesize 1788 questions covering eight sensitive attributes, with each set of questions designed to probe a specific type of causal reasoning. All questions are then manually validated, and each of them prompts the LLM to generate a causal graph behind its answer. We evaluate four state-of-the-art LLMs and find that all models exhibit biased causal reasoning on most questions eliciting it. Moreover, we discover that LLMs are also prone to "mistaken-biased" reasoning, where they first confuse correlation with causality to infer sensitive group membership and subsequently apply biased causal reasoning. By examining the cases where LLMs produce unbiased causal reasoning, we also identify three strategies LLMs employ to avoid bias (i.e., explicitly refusing to answer, avoiding sensitive attributes, and adding contextual restrictions), which provide insights for future debiasing efforts.