The Employee Separation Process: Criterion-Related Issues Associated with Tenure and Turnover

Abstract

The point is made that employee separation researchers have failed to distinguish among separation criteria, leading to inaccurate parameter estimation and evaluation. It is recommended that separation be viewed as either a dichotomous (e.g., avoidable vs. unavoidable or functional vs. dysfunctional) or a dichotomized variable (e.g., membership in a longor short-tenure group). Likewise, it is argued that type of separation criterion specified determines the appropriateness of a specific estimator. In the case of a truly dichotomous criterion, the point-biserial correlation is of interest, whereas in the case of a dichotomized criterion, the biserial correlation is the proper estimator. Related statistical issues are addressed.