Targeted Learning for Data Fairness
This work addresses fairness in data for stakeholders concerned with algorithmic bias, though it is incremental by expanding existing fairness inference methods to data processes.
The paper tackles the problem of evaluating fairness in the data generating process itself, rather than just in predictive algorithms, by using targeted learning for statistical inference on fairness metrics like demographic parity and equal opportunity, and validates the approach through simulations and real data applications.
Data and algorithms have the potential to produce and perpetuate discrimination and disparate treatment. As such, significant effort has been invested in developing approaches to defining, detecting, and eliminating unfair outcomes in algorithms. In this paper, we focus on performing statistical inference for fairness. Prior work in fairness inference has largely focused on inferring the fairness properties of a given predictive algorithm. Here, we expand fairness inference by evaluating fairness in the data generating process itself, referred to here as data fairness. We perform inference on data fairness using targeted learning, a flexible framework for nonparametric inference. We derive estimators demographic parity, equal opportunity, and conditional mutual information. Additionally, we find that our estimators for probabilistic metrics exploit double robustness. To validate our approach, we perform several simulations and apply our estimators to real data.