Utilization Tests

A utilization test is a formula that helps determine whether an organization’s actual representation rates are lower than would reasonably be expected (underutilization).  Companies use a number of methods to determine if underutilization exists, and any reasonable method, universally applied, is acceptable to the OFCCP:

• Any Difference: underutilization is declared if there is any difference between the utilization percentage (actual representation) and the availability percentage (expected representation).
• Calculations: The gender or race group’s availability percentage is multiplied by the total number of people in the job group to derive the number needed to achieve availability (the ”should have” number).  Then the actual gender or race group’s headcount is subtracted from the ”should have” number to derive the Additional Number Needed to Eliminate Problem Area.  This number will always be rounded up to the next highest number.  So if the calculation results in 6.3 people needed, then the displayed number will be 7 (because adding only 6 people still results in incumbency less than availability).
• Whole Person: underutilization is declared if the actual representation is at least one whole person lower than the number predicted by the expected representation.
• Calculations: The Additional Needed to Eliminate Problem Area uses the same calculation as the Any Difference rule, except the ”number needed” is rounded down.  So if the true result is 6.3 people, then the displayed number will be rounded to 6, the number of people needed so that incumbency will be within one ”whole person” of availability.
• 80% Rule: underutilization is declared only when the actual representation is less than 80% of the expected representation.
• Calculations: Each gender or race group’s availability is multiplied by 80%.  Then these products are multiplied by the job group’s total headcount to derive the number needed to achieve availability (the ”should have” number) for each gender or race group.  Then each gender or race group’s current incumbency is subtracted from its ”should have” number to derive the residual Additional Number Needed to Eliminate Problem Area (the additional number needed to achieve 80% of availability.
• Standard Deviation: underutilization is declared when the difference between actual and expected representation yields a standard deviation result of 1.96 or higher.
• Calculations: The Additional Number Needed to Eliminate Problem Area is determined by adding hypothetical persons to the at-issue gender or race group until the results are no longer statistically significant (the gender or race group’s standard deviation drops below 1.96).
• Exact Binomial: underutilization is declared when the difference between actual and expected representation yields a standard deviation result of 1.96 or higher.
• Calculations: This test uses the (two-tail) exact Binomial Probability test to assess whether the degree of underutilization is extreme enough to be considered ”beyond chance.”  Values less than .05 are ”statistically significant”, and will generate numbers in the Additional Number Needed to Eliminate Problem Area rows.
• 80% Rule with Whole Person: underutilization is declared when the expected representation exceeds actual representation by at least one whole person *and* the actual representation is less than 80% of the expected representation.
• Calculations: See calculations for 80% Rule with Whole Person rules.