ABSTRACT
This chapter describes a set of best practices for structuring traditional 2×2 tables for adverse impact analyses. In this context, 2×2 simply refers to a dichotomous protected group variable (e.g., Male versus Female, White versus Black, 40 and Older versus Under 40) crossed with a dichotomous employment outcome variable (e.g., Hired or Not, Promoted or Not, Terminated or Not). Although this structure is conceptually intuitive, 2×2 table analyses are often more complicated than analysts realize. In many situations, data, analysis structure, the appropriateness of statistical methods and interpretation decisions are often nuanced, without a “right answer” per se, and lacking meaningful guidance from the academic literature.
