"Patriarchy Hurts Men Too." Does Your Model Agree? A Discussion on Fairness Assumptions

TL;DR

This paper critically examines the assumptions underlying fairness measures in machine learning, revealing that common practices often assume overly simplistic bias models, which may limit or misguide fairness interventions.

Contribution

It formally analyzes implicit fairness assumptions, especially the monotonic biasing process, and discusses their implications for developing fair ML models.

Findings

Biasing process often assumed monotonic in fair scores

Many fairness models may be unnecessary if bias is predictable

Implicit assumptions can oversimplify complex bias behaviors

Abstract

The pipeline of a fair ML practitioner is generally divided into three phases: 1) Selecting a fairness measure. 2) Choosing a model that minimizes this measure. 3) Maximizing the model's performance on the data. In the context of group fairness, this approach often obscures implicit assumptions about how bias is introduced into the data. For instance, in binary classification, it is often assumed that the best model, with equal fairness, is the one with better performance. However, this belief already imposes specific properties on the process that introduced bias. More precisely, we are already assuming that the biasing process is a monotonic function of the fair scores, dependent solely on the sensitive attribute. We formally prove this claim regarding several implicit fairness assumptions. This leads, in our view, to two possible conclusions: either the behavior of the biasing process is more complex than mere monotonicity, which means we need to identify and reject our implicit assumptions in order to develop models capable of tackling more complex situations; or the bias introduced in the data behaves predictably, implying that many of the developed models are superfluous.