Recent narrative reviews (eg. et al 2012; et al 2017) advise that it is timely to assess the progress made in research on voluntary employee turnover so as to guide future work.
To provide this assessment, we employed a 3-step approach. First, we conducted a comprehensive meta-analysis of turnover predictors, updating existing effect sizes and examining multiple new antecedents. Second, guided by theory, we developed and tested a set of substantive moderators, considering factors that might exacerbate or mitigate zero-order meta-analytic effects. Third, we examined the holistic pattern of results in order to highlight the most pressing needs for future turnover research.
The results of Step 1 reveal multiple newer predictors and updated effect sizes of more traditional predictors, which have received substantially greater study. The results of Step 2 provide insight into the context-dependent nature of many antecedent-turnover relationships. In Step 3, our discussion takes a bird’s-eye view of the turnover “forest” and considers the theoretical and practical implications of the results.
We offer several research recommendations that break from the traditional turnover paradigm, as a means of guiding future study.
…Our holistic effort begins with an updated meta-analytic empirical assessment of turnover research to assess main effect relationships (Step 1). Since et al 2000, a bevy of new constructs have entered into the academic vernacular, whereas other constructs have been studied in considerably more depth, perhaps warranting a revision of earlier estimates. As a point of illustration, whereas the et al 2000 analysis examined 45 predictors and 843 effect sizes, we include 57 predictors across 1,800 effects sizes (a 27% increase in constructs and a 114% increase in effects). We provide insight into new and influential predictors such as engagement, justice, and job characteristics, as well as examining potential changes in effects sizes compared to previous work.
…1.21. Individual Attribute Predictors: Among individual attributes, tenure (ρ = −0.27, outliers excluded), age (ρ = −0.21), children (ρ = −0.20), emotional stability (ρ = −0.19), and Conscientiousness (ρ = −0.16) demonstrate the strongest effects. Perhaps more important; however, age validities statistically-significantly differed compared to the et al 2000 analysis (hereafter, GHG: −0.11, here: −0.21), as did the effects sizes for abilities/skills (GHG: 0.02, here: −0.06). Implications of this larger age effect in particular (ie. also more negative for post-2000 compared to pre-2000 studies), merit comment. If older workers are less likely to quit, younger workers are, equally, more likely to quit. Some scholars (eg. 2016) find support for the idea that younger workers hold higher—perhaps even unrealistic—expectations than do older workers regarding what they want from their employers. Looking forward, researchers might monitor this trend, and if/how the broader definitions of careers and work relationships change, and what that means for theory and practice.
