Combining predictors to achieve optimal trade-offs between selection quality and adverse impact

Area of Science:

• Industrial-Organizational Psychology
• Human Resource Management
• Quantitative Psychology

Background:

• Personnel selection aims to balance candidate quality with legal and ethical considerations like adverse impact.
• Existing methods may not adequately address the trade-offs between selection quality and adverse impact.
• Quantifying the relative importance of these competing objectives is crucial for effective selection.

Purpose of the Study:

• To develop a procedure for identifying Pareto-optimal predictor composites in personnel selection.
• To determine the relative importance of quality and adverse impact objectives for each trade-off.
• To assess the robustness of these Pareto-optimal composites under varying selection parameters.

Main Methods:

• Utilized a procedure to derive predictor composites that represent Pareto-optimal trade-offs.
• Analyzed the relative importance of quality and adverse impact objectives.
• Tested the performance of these composites with simulated variability in selection parameters (e.g., validity, predictor intercorrelations, subgroup differences).

Main Results:

• The procedure successfully identified predictor composites offering optimal trade-offs between selection quality and adverse impact.
• The relative importance of quality and impact objectives was quantifiable for each trade-off.
• Pareto-optimal composites demonstrated robust performance despite variability in selection parameters.

Conclusions:

• The proposed procedure offers a viable strategy for addressing the quality-adverse impact problem in personnel selection.
• This method is particularly useful when selection parameter estimates are available from validation studies or meta-analyses.
• The robustness of the identified composites supports their practical application in real-world selection contexts.