# Fine-Tuning Strategy, Not Just Data: A Statistically Rigorous Audit of Gender and Stereotypical Bias in Embedding Models for HR Recommendation Systems

arXiv: https://arxiv.org/abs/2610.14579 · HTML: https://mlsanity.com/paper/2610.14579/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 19 of its 26 references that are in the map.
- First posted on arXiv 10 Oct 2026.

## Builds on (most cited first, 19 of 19)

- [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://mlsanity.com/paper/1908.10084/) (2019)
- [A Survey on Bias and Fairness in Machine Learning](https://mlsanity.com/paper/1908.09635/) (2019)
- [Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings](https://mlsanity.com/paper/1607.06520/) (2016)
- [Text Embeddings by Weakly-Supervised Contrastive Pre-training](https://mlsanity.com/paper/2212.03533/) (2022)
- [Language (Technology) is Power: A Critical Survey of "Bias" in NLP](https://mlsanity.com/paper/2005.14050/) (2020)
- [StereoSet: Measuring stereotypical bias in pretrained language models](https://mlsanity.com/paper/2004.09456/) (2020)
- [Semantics derived automatically from language corpora contain human-like biases](https://mlsanity.com/paper/1608.07187/) (2016)
- [How Contextual are Contextualized Word Representations? Comparing the Geometry of BERT, ELMo, and GPT-2 Embeddings](https://mlsanity.com/paper/1909.00512/) (2019)
- [CrowS-Pairs: A Challenge Dataset for Measuring Social Biases in Masked Language Models](https://mlsanity.com/paper/2010.00133/) (2020)
- [Language-agnostic BERT Sentence Embedding](https://mlsanity.com/paper/2007.01852/) (2020)
- [Gender Bias in Coreference Resolution](https://mlsanity.com/paper/1804.09301/) (2018)
- [On Measuring Social Biases in Sentence Encoders](https://mlsanity.com/paper/1903.10561/) (2019)
- [Social Bias Frames: Reasoning about Social and Power Implications of Language](https://mlsanity.com/paper/1911.03891/) (2019)
- [Bias in Bios: A Case Study of Semantic Representation Bias in a High-Stakes Setting](https://mlsanity.com/paper/1901.09451/) (2019)
- [Mitigating Bias in Algorithmic Hiring: Evaluating Claims and Practices](https://mlsanity.com/paper/1906.09208/) (2019)
- [What Do Compressed Deep Neural Networks Forget?](https://mlsanity.com/paper/1911.05248/) (2019)
- [Characterising Bias in Compressed Models](https://mlsanity.com/paper/2010.03058/) (2020)
- [Black is to Criminal as Caucasian is to Police: Detecting and Removing Multiclass Bias in Word Embeddings](https://mlsanity.com/paper/1904.04047/) (2019)
- [Fairness in Information Access Systems](https://mlsanity.com/paper/2105.05779/) (2021)

## Abstract (arXiv)

Embedding models underpin modern job-candidate matching: resumes, job descriptions, and queries are encoded as dense vectors, and recommendations come from nearest-neighbor retrieval over that space. Any social bias in the embeddings therefore feeds directly into HR recommendations. We audit six multilingual sentence-embedding models (LaBSE- and E5-family, both public and custom fine-tuned) using Bias-in-Bios (occupational gender bias) and StereoSet (stereotypes across profession, race, gender, and religion). We treat statistical validity as a first-class concern: we identify a non-independence error in a naive significance test, where pooling non-independent pairwise similarities inflated a trivial 0.0046 mean difference to p ≈ 0, and replace it with a profession-level paired Wilcoxon signed-rank test, individual-resampling bootstrap confidence intervals, and permutation tests for StereoSet. Under the corrected methodology, knowledge-distillation (KD) fine-tuning of E5-Base significantly increases measured bias on Bias-in-Bios (Delta = -0.177, p < 10-8; more negative means more bias), although this measure is sensitive to global embedding geometry. The KD result is corroborated by pro-stereotype patterns in the StereoSet profession category (uncorrected p < 0.05). Binary-classification fine-tuning of the same base model shows no statistically distinguishable change (p = 0.218). Fine-tuning strategy is thus an actionable design lever for practitioners building HR recommenders. We also report which benchmark categories are underpowered, and note that two domain-adapted models could not be included in significance testing due to a software-access constraint, which limits our conclusions.