Barak, M.E.M., Travis, D.J., Pyun, H., & Xie, B. (2009). The Impact of supervision on worker outcomes: A meta-analysis. Social Service Review, 83(1), 3-32.
Effective supervision is a vital aspect of social service delivery. There have been many studies of the effect of various supervisory practices on worker outcomes; this is a report of a meta-analysis of 27 such studies, published between 1990 and 2007, representing a combined sample of 10,867 workers--social service (13 studies), child welfare (7 studies) and mental health (7 studies). Computerized searches were made of 3 data bases (PsycInfo, Social Work Abstracts, Social Service Abstracts) supplemented by manual searches of 8 key journals; a list of key words used in these searches is provided. Authors of all included articles were also contacted to ask about possible unpublished “file drawer” studies; some additional manuscripts were identified, but none met the criteria for inclusion.
This analysis used a theoretical model, based on the literature and social exchange theory, with:
- 3 supervisory dimensions : task assistance, social/emotional support, interpersonal interaction
- 5 detrimental worker outcomes: anxiety, depression, job stress, burnout, employee turnover.
- 5 beneficial worker outcomes: psychological well-being, job satisfaction, organizational commitment, effectiveness, job retention.
Each study was coded to reflect these dimensions. Measures used in various studies included Caplan’s Social Support Index and Maslach’s Burnout Scale. Effect sizes reported were 54 correlation coefficients (r) and 20 standardized regression coefficients (beta); where a study included several r values for different measures of the same construct, these were averaged. All effect sizes were converted to Fisher Z-scores to achieve more nearly normal distributions. Appropriate tests were performed for homogeneity of effect sizes across studies, and for possible publication bias.
Results are reported for 5 combinations of supervisory dimension and worker outcome category; there were no studies testing the effect of task assistance on detrimental worker outcomes. All three supervisory dimensions had significant positive relationships with beneficial worker outcomes, the task assistance dimension having the strongest effect. Social/emotional support and interpersonal interaction were associated with a decrease in detrimental outcomes, such as stress and burnout. All effect sizes were moderate (r = .30 - .40). These findings underscore the importance of effective supervision; the authors suggest that agencies should devote more resources to training supervisors across all supervisory dimensions.
This appears to have been a thorough and well-conducted meta-analysis, with all the usual statistical precautions taken and all procedures clearly described. The authors identify the most important limitations as the small number of studies available with data relevant to this issue, resulting in a small number of studies for each worker outcome category. A possible strategy here might have been to combine the beneficial and detrimental categories, as some of the outcomes appear to be mirror images (e.g., turnover & retention). The favorable results in this study suggest some options for further research. One logical next step might be a cost-benefit analysis; could extra agency expenditures on supervisor training lead to savings on staff turnover? Also, one of the measured beneficial outcomes was worker effectiveness; a logical hypothesis would be that this might lead to better client outcomes.