Designing Selection Systems to Reduce Turnover
Bush, Joshua David
MetadataShow full item record
This study sought to explore and predict voluntary turnover as a critical dependent variable in selection systems. I describe how using competency modeling to build measures of person-organization (P-O) fit may serve as a theoretical launching point for designing selection assessments to predict voluntary turnover. I tested this theoretical framework by analyzing an archival selection database of 3,332 current and former employees who went through a competency model-based selection system, which includes five different assessment methods all designed to measure P-O fit. I hypothesized that scores on these assessments would negatively predict turnover risk because research has demonstrated a strong negative relationship between P-O fit and turnover. Survival analysis demonstrated that the selection system predicted probability of voluntary turnover (i.e., turnover risk). Regarding specific assessments, interest inventory scores did not significantly relate to turnover risk, and structured interview scores demonstrated a significant positive relationship with turnover risk. The dispositional inventory, biographical inventory, and SJT scores all demonstrated significant negative relationships with turnover risk as hypothesized. When combining these assessments, dispositional scores accounted for all the biographical inventory scores’ relationship with turnover risk, and most of the SJT scores’ relationship. Therefore, dispositional inventory scores were the strongest predictor. Turnover risk was reduced by 38% for each standard unit increase in dispositional inventory scores.