# Robustness and Confounders in the Demographic Alignment of LLMs with Human Perceptions of Offensiveness

arXiv: https://arxiv.org/abs/2411.08977 · HTML: https://mlsanity.com/paper/2411.08977/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Builds on 31 of its 64 references that are in the map.
- First posted on arXiv 13 Nov 2024.

## Builds on (most cited first, 30 of 31)

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context](https://mlsanity.com/paper/2403.05530/) (2024)
- [ChatGPT Outperforms Crowd-Workers for Text-Annotation Tasks](https://mlsanity.com/paper/2303.15056/) (2023)
- [Whose Opinions Do Language Models Reflect?](https://mlsanity.com/paper/2303.17548/) (2023)
- [Out of One, Many: Using Language Models to Simulate Human Samples](https://mlsanity.com/paper/2209.06899/) (2022)
- [Automated Hate Speech Detection and the Problem of Offensive Language](https://mlsanity.com/paper/1703.04009/) (2017)
- [Can Large Language Models Be an Alternative to Human Evaluations?](https://mlsanity.com/paper/2305.01937/) (2023)
- [DecodingTrust: A Comprehensive Assessment of Trustworthiness in GPT Models](https://mlsanity.com/paper/2306.11698/) (2023)
- [Social Bias Frames: Reasoning about Social and Power Implications of Language](https://mlsanity.com/paper/1911.03891/) (2019)
- [Dealing with Disagreements: Looking Beyond the Majority Vote in Subjective Annotations](https://mlsanity.com/paper/2110.05719/) (2021)
- [Large Scale Crowdsourcing and Characterization of Twitter Abusive Behavior](https://mlsanity.com/paper/1802.00393/) (2018)
- [Annotators with Attitudes: How Annotator Beliefs And Identities Bias Toxic Language Detection](https://mlsanity.com/paper/2111.07997/) (2021)
- [Data and its (dis)contents: A survey of dataset development and use in machine learning research](https://mlsanity.com/paper/2012.05345/) (2020)
- [SOLAR 10.7B: Scaling Large Language Models with Simple yet Effective Depth Up-Scaling](https://mlsanity.com/paper/2312.15166/) (2023)
- [Quantifying the Persona Effect in LLM Simulations](https://mlsanity.com/paper/2402.10811/) (2024)
- [Is ChatGPT better than Human Annotators? Potential and Limitations of ChatGPT in Explaining Implicit Hate Speech](https://mlsanity.com/paper/2302.07736/) (2023)
- [NLPositionality: Characterizing Design Biases of Datasets and Models](https://mlsanity.com/paper/2306.01943/) (2023)
- [When the Majority is Wrong: Modeling Annotator Disagreement for Subjective Tasks](https://mlsanity.com/paper/2305.06626/) (2023)
- [Can ChatGPT Reproduce Human-Generated Labels? A Study of Social Computing Tasks](https://mlsanity.com/paper/2304.10145/) (2023)
- [When Do Annotator Demographics Matter? Measuring the Influence of Annotator Demographics with the POPQUORN Dataset](https://mlsanity.com/paper/2306.06826/) (2023)
- [Gender, Race, and Intersectional Bias in Resume Screening via Language Model Retrieval](https://mlsanity.com/paper/2407.20371/) (2024)
- [Comparing BERT against traditional machine learning text classification](https://mlsanity.com/paper/2005.13012/) (2020)
- [The Ecological Fallacy in Annotation: Modelling Human Label Variation goes beyond Sociodemographics](https://mlsanity.com/paper/2306.11559/) (2023)
- [The Perspectivist Paradigm Shift: Assumptions and Challenges of Capturing Human Labels](https://mlsanity.com/paper/2405.05860/) (2024)
- [Human and LLM Biases in Hate Speech Annotations: A Socio-Demographic Analysis of Annotators and Targets](https://mlsanity.com/paper/2410.07991/) (2024)
- ["They are uncultured": Unveiling Covert Harms and Social Threats in LLM Generated Conversations](https://mlsanity.com/paper/2405.05378/) (2024)
- [Annotation alignment: Comparing LLM and human annotations of conversational safety](https://mlsanity.com/paper/2406.06369/) (2024)
- [The Origin and Value of Disagreement Among Data Labelers: A Case Study of the Individual Difference in Hate Speech Annotation](https://mlsanity.com/paper/2112.04030/) (2021)
- [Annotation Sensitivity: Training Data Collection Methods Affect Model Performance](https://mlsanity.com/paper/2311.14212/) (2023)
- [Which Demographics do LLMs Default to During Annotation?](https://mlsanity.com/paper/2410.08820/) (2024)

## Built on by (most cited first, 7 of 7)

- [Beyond Demographics: Fine-tuning Large Language Models to Predict Individuals' Subjective Text Perceptions](https://mlsanity.com/paper/2502.20897/) (2025)
- [Generative Exaggeration in LLM Social Agents: Consistency, Bias, and Toxicity](https://mlsanity.com/paper/2507.00657/) (2025)
- [SubData: Bridging Heterogeneous Datasets to Enable Theory-Driven Evaluation of Political and Demographic Perspectives in LLMs](https://mlsanity.com/paper/2412.16783/) (2024)
- [Are Economists Always More Introverted? Analyzing Consistency in Persona-Assigned LLMs](https://mlsanity.com/paper/2506.02659/) (2025)
- [Demographic Prompting at Scale: When More Attributes Hurt LLM--Human Agreement](https://mlsanity.com/paper/2607.10590/) (2026)
- [PALMs: Using Multi Construct-Grounded Rationales for Modeling Population Preferences in LLMs](https://mlsanity.com/paper/2608.01458/) (2026)
- [LLM-as-a-Demographic: Whom Sociodemographic Prompting Helps, and Whom It Hurts](https://mlsanity.com/paper/2609.00222/) (2026)

## Abstract (arXiv)

Large language models (LLMs) are known to exhibit demographic biases, yet few studies systematically evaluate these biases across multiple datasets or account for confounding factors. In this work, we examine LLM alignment with human annotations in five offensive language datasets, comprising approximately 220K annotations. Our findings reveal that while demographic traits, particularly race, influence alignment, these effects are inconsistent across datasets and often entangled with other factors. Confounders -- such as document difficulty, annotator sensitivity, and within-group agreement -- account for more variation in alignment patterns than demographic traits alone. Specifically, alignment increases with higher annotator sensitivity and group agreement, while greater document difficulty corresponds to reduced alignment. Our results underscore the importance of multi-dataset analyses and confounder-aware methodologies in developing robust measures of demographic bias in LLMs.