Adverse Impact Analysis
While your organization may be required to conduct adverse impact analyses on an annual basis, these reports are for internal use and are not submitted with the Affirmative Action Plan documents.
These reports are commonly run by job group for each transaction type:
- Applicants vs. Hires
- Available for Promotion vs. Promoted
- Available for Termination vs. Retained (Involuntary)
- Available for Termination vs. Retained (All)
These reports work with data already imported and assigned to the current plan (Query method), if transaction files have been imported. Alternately, you can manually enter data to be analyzed (Manual method). This applies to both the Rates Analysis options and the Availability Analysis options.
General Report Options
Fill in the desired report title, header, footer, and select desired display options (suppress snapshot/plan date, print page numbers, allow text wrap, show report created on date).
Report-Specific Options
Rates Analysis (Manual) - Manually enter data to compare those who completed a process (i.e., hires) to those who began the process (i.e., applicants).
Rates Analysis (Query) - Use data that has already been imported and/or entered into myAAP to compare those who completed a process (i.e., hires) to those who began the process (i.e., applicants).
Availability Analysis (Manual) - Manually enter data to compare those who completed a process (i.e., hires) to the pool of those available for the job (may be final availability or another form of availability).
Availability Analysis (Query) - Use data that has already been imported and/or entered into myAAP to compare those who completed a process (i.e., hires) to the pool of those available for the job (may be final availability or another form of availability).
Statistical Methods - This will default to 80%, but can be changed if desired. If the Show Summary option is selected, the results from all methods will be displayed.
- 80% Test - a rule of "thumb test" designed to allow a compliance agency to easily find a focal employment area during investigations; the 80% test is not a legal definition of adverse impact. A violation occurs when the observed proportion of a group’s outcome (i.e., female hires) is less than 80% of the expected proportion of the group’s availability (i.e., females who applied).
- Fisher Exact (available for rates analysis only) - provides a more precise calculation of the probability value and is the preferred statistical method. The test includes a mid-p adjustment (see Agresti & Gottard, 2007; Simonoff, 2003; and Armitage, et.al., 2001). Standard deviation of 1.96 or greater or a probability value of .05 or less is considered statistically significant and will be flagged as a violation.
- Chi-Square (available for rates analysis only) - provides an estimated calculation of the probability value. This test if helpful in understanding the findings by a compliance agency that uses this older and less precise method. Standard deviation of 1.96 or greater is considered statistically significant and will be flagged as a violation.
- Exact Binomial (available for availability analysis only) - provides a more precise calculation of the probability value. Standard deviation of 1.96 or greater or a probability value of .05 or less is considered statistically significant and will be flagged as a violation.
- Generalized Binomial (available for availability analysis only) - provides an estimated calculation of the probability value. Standard deviation of 1.96 or greater is considered statistically significant and will be flagged as a violation.
Highest Rate - Select Automatic from the Highest Rate Type dropdown to have myAAP automatically select the group with the highest rate, or select Manual to select desired groups to represent the highest rate groups for gender and race.
Show Summary - This option will generate a summary report that includes all statistical calculation method results. This makes is simple to see analysis results using different calculations in one report, to help you decide which one to use.