The Statistical Report component of the Adverse Impact Report is a technical report prepared for those people with advanced statistical and Human Resources training. It provides a statistical test of whether pass rates differ between protected classes and an evaluation of the Uniform Guideline's "four-fifths rule." It is recommended you consult with a trained professional in the interpretation of this report.
The Statistical Report is a series of 2x2 tables that compares the pass rates of different protected classes at any point where decisions are made about candidates. Here is a sample table:
Each row in the table represents a protected class. The Passed column has the number of candidates for each protected class who passed that stage, and the Failed column has the number who did not pass that stage. % Passed is the percentage of candidates who passed. The text underneath the table computes the Adverse Impact Ratio (AI ratio) by dividing the lowest pass rate by the highest pass rate. A Chi-Squared test is then performed to test for differences in the pass rates.
There are five demographic variables included in this report, corresponding to the demographic questions asked of candidates: ethnicity, gender, veteran status, disability status, and age. Each variable has a single referent group to which all others are compared. These referent groups are White (not Hispanic or Latino) for ethnicity, Male for gender, Not a veteran for veteran status, Not disabled for disability status, and Under 40 for age. Candidates who chose not to disclose their demographics are excluded from analysis.
There are six stages at which adverse impact is evaluated, corresponding to the six points at which decisions regarding the candidates are made. These are:
- Qualification: The basic requirements for the job, including knowledge quizzes
- Work Styles: The assessments stage
- Homework: The open-ended questions asked of candidates
- Interview: The recorded interview stage
- Revealed: Which candidates were Unlocked to access contact information
- Hired: Which candidates were hired by your company
All statistics are based on the demographic information as self-reported by the candidates. If there are insufficient numbers of candidates passing a stage, or if there are too few members of a protected class, the AI Ratio and Chi-Squared test are not reported. Some demographic data may be obfuscated to ensure candidates do not have their individually identifiable information revealed to employers.