An Agent-based Model for Predicting Turnover in the Science, Technology, Engineering, and Mathematics (STEM) Workforce Open Access
Downloadable ContentDownload PDF
The United States (U.S.) government forecasts a shortage of Science, Technology, Engineering, and Mathematics (STEM) workers, putting STEM workforce sustainability at risk. The focus of this dissertation is to understand the micro-behaviors and antecedent variables that contribute to engineer turnover, the largest segment of the STEM workforce. Few Systems Engineering (SE) or Engineering Management (EM) studies have focused on collective organizational environment factors for analyzing worker turnover. An interdisciplinary systems approach was used to examine STEM and engineer turnover through the use of agent-based modeling (ABM) to generate testable predictions on organizational level turnover.In part I, the ABM is validated by turnover data for the National Aeronautics and Space Administration (NASA) and two control group organizations. Data for model verification and validation is based on the U.S. Office of Personnel Management (OPM) 2005-2014 Federal Human Resources Data and the 2014 Federal Employee Viewpoint Survey (FEVS) Results. The model confirms that STEM density is negatively related to worker turnover in general. Model observations further show the emergence of a negative relationship between organizational STEM density and worker team size for knowledge workers in general, but a positive relationship to STEM workers, particularly engineers. The ABM output generates hypotheses for testing with multiple regression analysis (MRA) as it relates to engineer turnover and supervisor structure as a representation of work group size.Part II examines the relationship between organizational supervisor ratio and engineer supervisor percentage with engineer turnover and total organizational turnover, within which supervisor ratio is defined as the percentage of supervisors to total workers and engineer supervisor percentage is defined as the percentage of engineer-trained supervisors to total engineers. Hierarchical MRA was used to test hypotheses for 85 organizational-level samples across 17 United States (U.S.) large independent federal government agencies from 2010-2014.The findings suggest that engineer turnover is negatively related to supervisor ratio and positively related to engineer supervisor percentages. The study found mediating effects of engineer supervisors in reducing the negative relationship between supervisor ratio and engineer turnover. Engineer turnover was also found to moderate the relationship between supervisor ratio and non-engineer knowledge worker turnover rate; lower engineer turnover is a significant predictor of lower non-engineer turnover in low supervisor ratio organizational environments, suggesting a cross-over effect that reverses the non-engineer worker turnover relationship with high supervisor ratio. This study provides a theory-building prediction framework for the role of organizational supervisor ratio in mitigating turnover risk for both engineers and all other knowledge workers in interdisciplinary work environments. By applying value-congruence theory, engineering managers can use these findings to understand the functional role of supervisor ratio and its impact on organizational turnover. This study concludes with practical systems and engineering management recommendations.