Is My Prediction Arbitrary? Measuring Self-Consistency in Fair Classification

Variance in predictions across different trained models is a significant, under-explored source of error in fair classification. Empirically, the variance on some instances is so large that decisions can be effectively arbitrary. To study this problem, we perform a large-scale empirical study and make four overarching contributions: We 1) Define a metric called self-consistency, derived from variance, which we use as a proxy for measuring and reducing arbitrariness; 2) Develop an ensembling algorithm that abstains from classification when a prediction would be arbitrary; 3) Conduct the largest to-date empirical study of the role of variance (vis-a-vis self-consistency and arbitrariness) in fair classification; and, 4) Release a toolkit that makes the US Home Mortgage Disclosure Act (HMDA) datasets easily usable for future research. Altogether, our empirical results reveal shocking insights about reproducibility. Most fairness classification benchmarks are close-to-fair when taking into account the amount of arbitrariness present in predictions. Subgroup error rates are similar before we even try to apply common fairness interventions