2M03 Analysis of Objective Criteria for Adverse Impact

a. Pool of Individuals Used for Statistical Calculation. In an adverse impact calculation, the CO compares the number of individuals from each group (by race, ethnicity, and sex) who were assessed using the criterion at issue with the number from each group who met the criterion. This yields a “pass rate” or “selection rate” for each group, and the CO then tests the difference in pass rates between the groups for adverse impact.

b. Performing the Adverse Impact Calculation. First, the CO must determine whether each objective criterion is a pass/fail requirement, (i.e., it eliminates an individual from further consideration if it is not met), or whether it is considered along with other criteria in making a final decision.

  • For each pass/fail criterion, the CO must determine whether the difference in pass rates for members of each group (by race, ethnicity, and sex) shows adverse impact.
  • When the criteria are not considered singly but are considered together in reaching a decision, the CO must determine how the contractor weighted each criterion to make a selection. Using the weights used by the contractor, the CO must determine whether this is causing the adverse impact. If it is not possible to isolate the contribution of the individual criterion to the bottom line effect, the CO will calculate the cumulative effect of all the criteria. This calculation may be appropriate when the contractor did not keep records of the effects of the individual criteria.

c. Measurement of Statistical Significance. Measurement of the statistical significance of a difference in selection rates, whether by standard deviations or another method, indicates the probability that a particular disparity in those rates could or could not have occurred by chance. The greater the disparity between the percentage of a specific group (e.g., women) actually selected and the percentage of the group expected, the more likely the variation did not happen by chance but was a result of the selection factor used to screen. When there is a significant probability that the disparity in selection rates for members of a nonfavored group could not have happened by chance, this is strong evidence of adverse impact.