I am not familiar with the software you are using to produce your data, Michelle, so I am going to answer your question with the RESULT one MUST arrive at, irrespective of what reports you need to use to get there. The short answer is that in EVERY analysis that purports to measure whether there is “adverse impact” in an organization’s selection criteria/selection process, one must measure ALL the selections in the numerator (including not just “hires” but even those who never even took the job!) And one must include ALL BUT ONLY the people who COMPETED for those opportunities, regardless of their status and regardless of what the “payroll consequence” was or would have been had your selection led to occupying the job. This may be news to you – it may be news to some other readers, and I promise it will be news to some OFCCP employees who nevertheless, if challenged (up to and including the National Office), will have to agree. Your job is for the discrimination analyses and any data that you share with the OFCCP to reflect both the LAW requires and WHAT ACTUALLY HAPPENED. According to the OFCCP’s “Uniform Guidelines on Employee Selection Procedures”, the objective of an “adverse impact” analysis is to determine whether – and to what degree – the employers’ selection process/selection criteria are “disproportionately” excluding persons on the basis of race, national origin or sex. If one is to measure “impact”, one must count ALL the individuals “impacted” by the employee’s decision making AND must also count ALL the persons NOT excluded/NOT “adversely impacted” by the EMPLOYER’S decision making. The numerator of the selection ratio is all who were selected (i.e., NOT denied/NOT “adversely impacted” and the denominator is all who competed (i.e., ALL those impacted whether adversely or favorably in the selection process). Let us walk through it step by step. I am going to contrive some situations and numbers below to illustrate. Pretend for a moment that you have a vacancy which the organization believes can be filled from within. In that case, it is the policy to post the job internally and permit only current employees to apply, which they do electronically. 10 people, 5 men and 5 women compete for this single vacancy. All of them meet the meet “Basic Qualifications” and none withdraw either explicitly or implicitly prior to you making a job offer to one of the women. But Sally changes her mind and declines what would have been FOR HER, a lateral transfer. So you offer the job to Martha who accepts what would be a nice two-level promotion FOR HER. But before she actually starts work in the new job she decides that, after all, she is happier on the shift she is on than she would be on the one for the new job, so she also declines. Finally, you offer the position to Stan. He accepts with pleasure what is FOR HIM is a DEMOTION in grade and cut in pay but is on the shift he wants so that he can go back to school. Tragically, Stan is gravely injured in an auto accident the weekend before he was to report to his new job and new shift and it’s uncertain when or even if he will be able to return to work. Note: the potential "payroll consequence" of each of these selections was different, but none was "adversely impacted" by the employer. Q. Did this selection process have an adverse impact on women? A. NO! 2 of 5 women were selected for the position or 40% versus 1 of 5 male applicants or 20%. The selection process for this vacancy did not exclude women at a faster rate than men; if anything, they were twice as successful as the male applicants. Three people SELECTED by your selection process, i.e., were NOT DENIED this opportunity, and yet not a single person was “hired”. And you still have a vacancy! You post the position and invite both internal and external job seekers to compete for it. As it happens, no additional current employee express interest. Of the external job seekers, 5 men and 5 women meet the regulatory definition of “Internet Applicant”. You offer the position to one of them and she declines. Next you offer the position to your second choice and she accepts (Hurray!) but – without notice or explanation – simply fails to show up for work on the agreed upon start date. The supervisor has now waited a month to fill this spot and is beside herself. With a sigh, you go to your third choice, he accepts and starts work as a new hire, which is the "payroll consequence" of the selection of a new employee who actually goes to work and is now on the payroll. At last, a “HIRE”! OFCCP regulations require you to measure the impact of your selection devices, whether they be Basic Qualifications, Preferred Qualifications, a standardized test, an interview/s and for some or all of the Internet Applicants (those people who possessed basic qualifications, did withdraw explicitly or implicitly at any time prior to a job offer/s being made) of whatever attributes of their previous employment you deem relevant. This might be the number of jobs held in a particular time frame, the reason for leaving a job/s, the duration of unemployment between jobs and the explanations given, etc. All of these are "tests" or "selection devices" that may include or exclude ("eliminate") applicants. When your selection criterion INCLUDE – that is, result in a SELECTION – there can be no “adverse” impact by that particular criterion or, in total at the “bottom line”, depending on what is being analyzed at the moment. ALL persons SELECTED belong in the numerator of the analysis and ALL AND ONLY persons who COMPETED (that is, both those INCLUDED and those EXCLUDED by your selection criterion/a) belong in the denominator. Selected divided by the Total of Selected + Rejected = Selection rate%. If no women are HIRED from among 10 applicants, the Hire Rate is 0%, if one of ten man is HIRED, the Hire Rate for men is 10%. So what? That is NOT a SELECTION RATE. And NEITHER measures the “adverse impact” OF THE EMPLOYER’S reasons for REJECTION nor DENIAL OF OPPORTUNITY for men nor women!!! Whether this difference is statistically significant depends entirely on the number of people of each sex who are in the pool and the number who are selected from the pool. I have said it so many times: NUMBERS COUNT: COUNT THE RIGHT NUMBERS! So, let us COUNT the number of men and women in the denominator and the number of men and women in the numerator for this single vacancy. There were 10 females “considered” (that is, an assessment of their qualifications occurred) and 10 males were similarly “considered”, five each internal and external. They competed with each other and were – presumably – assessed by the same process and same selection criteria. (If they were not, then we have disparate “treatment” issues rather than disparate “impact”. OR, if you were willing to accept lesser credentials from current employees, then the two groups did NOT compete with each other and should not be combined in any analysis.) What is the selection rate of women versus men? Remember, if you do not count the persons who were selected (albeit NOT “Hired”) in the Numerator but you DO count them in the denominator the consequence is that, mathematically, all of the people who competed but were not HIRED (or promoted, lateraled or EVEN DEMOTED) are being treated as if they were DENIED the opportunity that they sought. There are 10 men and 10 women people in the applicant pool – the denominator of each “selection rate. Of those 4 women were NOT EXCLUDED but were, in fact, “selected” – two employees and two external applicants, all of whom chose not to take the opportunity. So, there was no “payroll consequence”, not a hire, not a promotion or even a demotion among the bunch. But only 6 women were denied the OPPORTUNITY – NOT the 10 who were not “hired”. 8 men were denied the opportunity (4 employees and 4 external applicants) and two we
You can use this OFCCP audit checklist to ensure you're doing what is required to maintain OFCCP's regulations including VEVRAA, Section 503, and EO 11246. Or request a demo to streamline your compliance and recruiting efforts.