The main goals of any hiring process are to ensure that it is effective, efficient, and fair. There are many ways to determine whether you are meeting each of these goals. For example, you could conduct a validation study to determine its effectiveness. How well is each hiring tool predicting future employee behavior? Additionally, you could look at time-to-hire metrics to assess its efficiency. Are recruiters and human resource managers saving time with the hiring process? Finally, you could calculate the level of adverse impact to ensure that the system is fair for all candidates.
Adverse impact is defined as a significantly different rate of selection which negatively impacts members of a protected group. While courts tend to focus on whether adverse impact is present or not, in reality, adverse impact is assessed on a continuum and it is often difficult to have a selection system with absolutely no adverse impact.
One of the main reasons for this is due to the diversity-validity dilemma. This is the tradeoff between how well skills predict future behavior with how much they tend to show adverse impact. Meaning, some of the biggest predictors of job performance also demonstrate increased levels of adverse impact. In particular, cognitive ability has been shown to be one of the best predictors of job performance across all types of behaviors. However, cognitive ability typically demonstrates higher levels of adverse impact, especially for Blacks and Hispanics. Therefore, when developing employee assessments and creating scoring for assessments, we need to make sure that we maximize prediction, but at the same time minimize adverse impact associated with cognitive ability.
How can you really minimize adverse impact?
Researchers have suggested many ways of minimizing levels of adverse impact, including changing skills you are measuring in assessments and using different testing methods. At a recent conference, I presented with my colleagues on a research project examining whether the latter option would be effective in reducing adverse impact. In the research project, we compared levels of adverse impact between two assessment batteries for police officer job candidates.
In the first assessment battery, candidates completed a cognitive ability assessment and a traditional personality assessment. In the second assessment battery, candidates completed the cognitive ability and personality tests, but also engaged in types of exercises typically present in assessment centers. In particular, candidates participated in different oral exercises. During the exercises, candidates were presented with different work situations and had to respond appropriately to two trained assessors. Two assessors rated the candidates’ responses on skills such as oral expression, verbal comprehension, reasoning, and originality.
We hypothesized that including these oral exercises would decrease the reading requirements and cognitive load placed on candidates, which would ultimately decrease group differences. Additionally, since we were incorporating best practices of assessment centers (e.g., multiple assessors, behavioral-anchored ratings), we were not compromising on validity as assessment centers have been shown to be effective solutions for predicting job behavior.
To analyze the data, we compared overall assessment scores between the first and second assessment batteries. We found that the group differences between White and Black candidates were significantly lessened when adding oral exercises to the assessment battery.
In the end, incorporating oral exercises into your hiring process may be one method to minimize the possibility of adverse impact. In many hiring situations, it is often not feasible from a cost or resource perspective to implement a full-blown assessment center. By using the principles of assessments centers and targeting on one specific type of exercise commonly used in assessment centers, oral exercises, it’s possible to maintain high levels of prediction, reduce the possibility of adverse impact, and save time and money.