Causal Fairness Evaluation

Causal fairness evaluation is a method that employs structural causal models and counterfactual reasoning to distinguish direct, indirect, and path-specific effects in biased decision-making systems. It leverages techniques like back-door adjustment, inverse probability weighting, and doubly robust estimation to quantify fairness criteria under various causal assumptions. Applications across ...

However, the most recent notions of fairness are causal-based and reflect the now widely accepted idea that using causality is necessary to appropriately address the problem of fairness. This paper examines an exhaustive list of causal-based fairness notions and studies their applicability in real-world scenarios.

Abstract Over the past several years, multiple different methods to measure the causal fairness of machine learning models have been proposed. However, despite the growing number of publications and implementations, there is still a critical lack of literature that explains the interplay of causality-based fairness notions with the social sciences of philosophy, sociology, and law. We hope to ...

In contrast, causal-based measures focus on estimating the causal effects of sensitive attributes to eval-uate fairness and avoid statistical bias [Zuo et al., 2022; Li et al., 2024b]. In Potential Outcome Framework (POF), it is widely believed that, for fair models, there should be no causal relationship between sensitive attributes and outcomes.

Outline Review basic causal concepts in the context of fairness. Introduce the foundations of fairness analysis based on causal inference, including theory of decomposing variations, causal measures, and the fairness map. Discuss connections with previous literature.

Moving forward, it's essential to keep these visual contexts in mind when discussing Causal Fairness Evaluation.

Motivated by the problems posed by such measurement biases, we offer a framework based on graphical causal inference to operationalize assumptions about data quality issues, alongside methods adapted from causal sensitivity analysis for statistical quantification of their impacts on fairness evaluations.

To evaluate the discovery of the different classes of sensitive variables and the problematic structures, we consider accuracy. 6.2 Evaluation of the Impact on Causal Fairness Notions.

Causal fairness provides a more comprehensive evaluation framework by preserving causal structure, but current synthetic data generation methods do not address it in health settings.