Predicting employee turnover from performance, satisfaction, and leader-member exchange

As part of a study of the role of supervision in predicting turnover, 105 nurses provided confidential data on leader-member exchange and satisfaction with supervision, while their supervisors provided performance evaluation information. Although results reaffirmed an inverse correlation between employee performance and turnover (r = -.23, p < .05), only directional support was obtained for the hypothesis that performance moderates the relationship between satisfaction with supervision and turnover. Turnover was also not found to be curvilinearly related to performance. Additionally, the results did not replicate an earlier finding that leader-member exchange is a strong predictor of employee turnover. The results are interpreted as underscoring the difficulty of predicting turnover from affect, and the need to develop refined theoretical models that specifically link these constructs.