A closer look at decisions to quit

Aspects of March and Simon’s (1958) subjective expected utility model and a prediction of Lee and Mitchell’s (1994) unfolding model of voluntary employee turnover were tested. A policy capturing simulation that varied high, moderate, and low levels of five job characteristics was used to model voluntary turnover decision processes for 532 nurses. Survey measures of these job characteristics obtained over the next 2 years were multiplied by weights derived from nurse simulations to yield turnover likelihood estimates. These estimates exhibited 80%, 127%, and 190% more predictive power (depending on turnover operationalization) than post-employment survey estimates of turnover intention, job satisfaction, and job availability. Groups of nurses with homogeneous voluntary turnover decision models were also identified, though no groups with homogeneous job perceptions were observed. Evidence suggested nurses responded to “shocks” as predicted. March and Simon’s model of voluntary turnover was supported and implications drawn for managing voluntary nursing turnover.

► A simulation modeled decisions to quit for 532 newly hired nurses in a hospital.
► A biodata instrument was also created to predict job tenure.
► Decision models and biodata scores accurately predicted who quit over 6 years.
► Decision models predicted turnover better than prior theory-based measures.
► Groups of nurses with similar turnover decision models were identified.