COs must ensure that contractors submit employment activity data as specified in the Itemized Listing of the Scheduling Letter to conduct statistical analyses to identify any potential indicators of discrimination.
Generally, statistical analysis assists the CO in identifying the number of individuals selected from different groups (e.g., men versus women) compared to the number expected to be selected for each group, based on each group’s proportion in the relevant applicant or employee pool. In a hiring analysis, the relevant pool is usually based on applicants for each job group or job title. If, for example, the hires are significantly less than the expected hires for women based on an IRA, standard deviation analysis64 or any other analysis permitted by the UGESP, the CO has an indicator that the contractor’s selection practice may be having an adverse impact on women. At a minimum, further analyses would be warranted. This analysis could include requesting additional data at desk audit or on-site from the contractor about its selection practices and policies, refining data for further statistical analysis based on information obtained from the contractor, and consulting with statistical experts in the DPO.
64. The standard deviation represents the distribution of the responses or data around the mean, and is often used to measure the amount of confidence in statistical conclusions. COs may run standard deviation analyses to determine whether and what types of additional information they may need to establish potential discrimination.