Bridging Disciplinary Boundaries (January 11 - 14, 2007)

Purpose: Previous research suggests that social workers' burnout is significantly related to the higher turnover rate (Drake, & Yamada, 1996), and subsequently turnover phenomena negatively affect the quality and stability of social work practice (Mor Barak, Nissly, & Levin, 2001). However, burnout studies have considered social workers as a single occupational group, and little attention has been paid to the subgroup characteristics of social workers. Although Maslach Burnout Inventory (MBI; Maslach & Jackson, 1981) is undoubtedly the most widely used measure of burnout, its validity for use with different types of social workers has not been established. No empirical studies investigate the factorial invariance of MBI across various subgroups of social workers. The current study aims (a) to test the factorial validity of MBI for social workers; (b) to examine factorial invariance of MBI across social workers in public and private organization settings; and (c) to investigate factorial invariance of MBI across Licensed Clinical Social Workers (LCSW) and Associated Social Workers (ASW). The current study will better inform intervention and preventive efforts to address burnout among social workers, and ultimately contribute to the improvement of social work practice.

Method: For the study's cross-sectional survey design, 1,500 social workers were randomly selected from the total population of California state-registered social workers. Of the 529 returned questionnaire, 468 cases were eligible for the study, yielding a survey completion rate of 31 percent. The MBI for Human Service Survey is self-report measure of burnout and consists of Emotional Exhaustion (EE), Depersonalization (DP) and Personal Accomplishment (PA) subscales. Structural Equation Modeling techniques were used to test the hypothesized model of MBI using AMOS5.0 software.

Results: The results of confirmatory factor analyses indicated that the four factor model of MBI best fit the present data (CFI=.87, RMSEA=.083), which comprises EE, DP, and two sub-dimensions of PA (perceived achievement & professional efficacy). Additionally, as Byrne (1997) suggested, the deletion of items 12 and 16 yielded a better fitting model (CFI=.90, RMSEA=.074). A series of multi-group confirmatory factor analyses showed that factor structure of MBI was invariant across social workers in different organizational settings. However, the factor structure of MBI was variant across LCSWs and ASWs (Chi-square change(19)=33.8, p<.05). In particular, the results of partial invariant model testing indicated that the factor structure of professional efficacy is variant across the groups (Chi-square change(3)=18.1, p<.05). The results of factor mean difference tests indicated that social workers in public setting and ASWs exhibited higher levels of burnout comparing to their counterparts.

Discussion: The findings of the study indicates the underlying two-dimensions of PA. This finding is consistent with the research by Densten (2001). The presence of variant factor structure in the professional efficacy across LCSW and ASW groups suggests qualitative differences in the interpretation of MBI PA scale across the groups. The utilization and interpretation of MBI should be carefully made depending on the characteristics of social workers.