Predicting employee turnover from communication networks

This article investigates three social network models of employee turnover: a structural equivalence model, a social influence model, and an erosion model. It was predicted that structurally equivalent individuals would be more likely to behave similarly (i.e., leave or stay at their position). The social influence model predicted that employees with a greater percentage of direct communication links with leavers would be more likely to leave their job. The erosion model posited that individuals located on the periphery of a social network would be more likely to leave their job or "fall off" the edges of the social network. A total of 170 employees, a census of an organization, were administered a communication network questionnaire asking them to identify the people with whom they communicate at work. Network self-report data were analyzed using NEGOPY and UCINET software. Results provided support for all three models of turnover, with the erosion model explaining more of the variance than the other two models. Future research examining communication networks and employee turnover is proposed, on the basis of these results.