Testing the Relationship of Burnout to Job Satisfaction for Psychiatric Rehabilitation Practitioners beyond Controlled-for Background Variables

A Psychiatric Rehabilitation Practitioner (PRP) is a less-often studied rehabilitation counselor who works specifically with mental health patients. An on-line survey was filled out by 1,635 PRPs who provided background, burnout and job satisfaction data. Full data was available for 819 PRPs. Background data collected included: gender, race, age, highest education level, length of service, and salary. The study hypothesis that burnout would be negatively related to job satisfaction beyond the controlled-for background variables was supported. Although only a limited amount of job satisfaction variance was explained, the study results support continued study of burnout and job satisfaction for PRPs and other mental health professionals.


Introduction
A Psychiatric Rehabilitation Practitioner (PRP) is a type of rehabilitation counselor who works with mental health patients (Blankertz & Robinson, 1996). They are not typically studied as often as other mental health professionals (Morse, Salvers, Rollins, Monroe-DeVita, & Pfahler, 2012). The performance of PRPs can be assessed along different dimensions, including: assessment/planning; facilitating client recovery; and supporting health and wellness (Blau, Tatum, Goldberg, Viswanathan, Karnik, & Aaronson, 2014). In occupations where there is both a higher frequency and higher intensity of interpersonal contact, such as PRPs, there is a higher risk of burnout (Cordes & Dougherty, 1993). In a prior study of PRPs, Blankertz and Robinson (1996) found that burnout was negatively related to job satisfaction. However, Blankhertz and Robinson (1996) did not first control for background variables before testing if burnout was negatively related. The goal of this study was to test for the relationship of burnout to the job satisfaction of PRPs beyond controlled for background variables.

Relationship of Burnout to Job Satisfaction
Job satisfaction is an important work attitude to monitor (Spector, 1985), and consists of different facts, including: supervision, work itself, pay, co-workers, and benefits. Often low job satisfaction is a precursor to an employee leaving their job or even their occupation. General research is also supportive of a negative relationship between burnout and job satisfaction (Cordes & Dougherty, 1993;Faragher, Cass, & Cooper, 2005). Burnout is thought to be a stress response syndrome experienced by service-oriented employees, collectively consisting of three components -emotional exhaustion, depersonalization and diminished personal accomplishment (Cordes & Dougherty, 1993;Maslach, 1982). Burnout has been extensively researched using samples of allied and mental health workers such as counselors, social service employees, and rehabilitation staff (Morse et al., 2012). In their study of PRPs, Blau, Tatum and Goldberg (2013) found that personal involvement was positively correlated to two burnout dimensions, emotional exhaustion and depersonalization. They recommended that PRPs try to reduce their level of personal involvement with their clients and to utilize peer support groups whenever possible. Collectively prior research suggests that burnout among PRPs is an important topic to study. The question is whether burnout will be significantly related to job satisfaction beyond controlled-for backgrounds for PRPs. To the author's knowledge, this has not yet been empirically tested.

Controlling for Correlates of Job Satisfaction
Although the results are inconsistent across studies, prior research on job satisfaction suggests first controlling for background-related variables before testing the impact of other variables. For example, while Blau (1999) and Keshtkaran (2009) both found that age and salary had positive relationships to job satisfaction, Echchakoui and Naji (2013) did not find age to be related to job satisfaction, and Cavanagh (1992) did not find salary to be related. McIntyre and McIntyre (2010) found gender (males lower) and education level to be negatively correlated to job satisfaction. However, Blau (1999) did not find gender to be related, while Din, Zaman and Nawaz (2010) found males had higher job satisfaction. Echchakoui and Naji (2013) did find a positive education level -job satisfaction relationship. Miller, Mire and Kim (2009) did not find gender or race to be related to job satisfaction but length of service was negatively related. Keshtakaran (2009) did not find length of service to be related to job satisfaction. McNeely (1989) found a race/ethnic effect such that Black and Hispanic human service employees reported higher job satisfaction than Caucasian. Collectively this research suggests that background-related variables, i.e., gender, race, age, length of service, education level and salary, should at least be controlled for when testing the impact of burnout on job satisfaction (Nemmaniwar & Deshpande, 2016). This suggests the following study hypothesis: H1 -Burnout will be negatively related to job satisfaction beyond the controlled-for background-related variables of gender, race, age, education level, length of service and salary.

Sample and Procedure
In the Fall, 2012 PRPs filled out an online survey, including a section asking about backgrounds, burnout and job satisfaction. The Certification Commission for Psychiatric Rehabilitation, within the United States Psychiatric Rehabilitation Association, sent out the online survey to its members. The survey was anonymous so no identifying information was required. The Institutional Review Board (IRB) approved this online survey. 1,635 PRPs or 44% (1,635/3,676) of those receiving the survey responded.
Burnout. The three frequency dimensions from the Maslach Burnout Inventory (MBI, Maslach, 1982) were measured, emotional exhaustion (9-items), depersonalization (5-items), and personal accomplishment (6-items). Due to a survey omission, two personal accomplishment items were accidentally not included. Respondents were asked to "consider your work in the past three months" as they chose their item responses. The following six-point response scale was given which directly corresponded to the other six response options on the MBI: 1 = never, 2 = once a month or less, 3 = a few times a month, 4 = once a week, 5 = a few times a week, and 6 = every day. All twenty items were summed into a general burnout score. Cronbach alpha for this measure was .90.
Job Satisfaction. Nine study-specific items were used, measuring common facets of job satisfaction, including: supervision, work itself, co-workers, pay, communication from top management, and benefits (Spector, 1985). All nine items were summed into a general job satisfaction score. Responses were made from 1 = strongly disagree to 6=strongly agree. Cronbach alpha for this measure was .94. Table 1 shows the means, standard deviations, and correlations for continuous study variables. SPSS-PC (2011) was used for all data analyses. Using listwise deletion reduced the sample size to 819. Not finding significant background differences between the incomplete versus complete data samples supports that missing data were random (Stevens, 1996). Overall, the sample had: an average age range of 31 to 40 years; generally at least an jedp.ccsenet.org Journal of Educational and Developmental Psychology Vol. 10, No. 2; 2020 associates' degree; an average salary of $40K -49,999; an average length of service over 10 years; a lower level of burnout; and a moderate level of job satisfaction. Multiple item measures were divided by their item number of items to make the response scale more meaningful. Both the burnout (.90) and job satisfaction (.94) scales had strong reliabilities (coefficient alpha). Highlighting some correlations, while age (r = -.13) and length of service (r = -.11) had modest negative correlations to burnout, only burnout (r = -.24) had a more strongly significant correlation to job satisfaction a Age (years): 1 = less than 25 to 6 = at least 61; b Education level: 1 = high school degree or less to 6 = post graduate degree; c Salary: 1 = less than 20K to 9 = over 90K; d Length of service: 1 = less than one year to 5 = over 20 years; e Burnout: 1 = never to 6 = every day; f Job Satisfaction: 1 = strongly disagree to 6 = strongly agree. g Reliability estimate (Cronbach alpha), NA = not applicable. Table 2 presents the final stepwise regression model for testing the study hypothesis, burnout will be negatively related to job satisfaction beyond controlled-for background variables. Prior to model testing, the assumptions of no multicollinearity, linearity, homoscedasticity, and normally distributed errors were satisfactorily met (Stevens, 1996). These findings give more confidence in the regression results. The results in Table 2 provide support for H1, since burnout has a significant negative relationship to job satisfaction (b = -.52) beyond the controlled-for backgrounds, accounting for 5% (p < .01) of the job satisfaction variance. Overall, these variables collectively explained only 6% of job satisfaction variance, and none of the controlled-for background variables were significantly related to job satisfaction. . 0 1

Discussion
Although the data set used was over seven years old, the author was not able to find more recent research testing this specific hypothesis with PRPs. PRPs are not as often studied as other types of mental health professionals. The study hypothesis was supported, however only 6% of PRP job satisfaction was explained. Having a stronger research design, which included such perceived work environment variables as autonomy, feedback, and opportunities for growth (Faragher et al., 2005), could have allowed for more job satisfaction variance to be explained. The job satisfaction measure used, while reliable, was more oriented towards measuring extrinsic factors, such as one's supervisor, pay, co-workers, communications from top management, and benefits. Both age and length of service had a negative relationship to burnout. While perhaps somewhat surprising, other research has found similar results (van der Ploeg, van Leeuwen, & Kwee, 1990). Younger and less experienced PRPs who perhaps "over-invest" in their clients, e.g., higher frequency and intensity of contact (Cordes & Dougherty, 1993) may experience burnout sooner.

Limitations
The lower level of burnout reported in this study was pleasantly surprising compared to prior mental health sample research (Morse et al., 2012). While the burnout scale was reliable, using only six of eight personal accomplishment items from the MBI (Maslach, 1982), and not including the year-long response option, prevents a direct comparison of PRP burnout to the accumulated normative data of MBI mental health samples (Bakker, Demerouti, & Schaufeli, 2002). Such a direct comparison would have increased the validity of this study's results.
In addition, the cross-sectional data collection used does not allow for inferring causality, i.e., that burnout causes lower job satisfaction. Such a research limitation is common in burnout research (Morse et al., 2012;Hoge, 2010).
In an on-line survey, sample self-selection is a concern (Eaton & Struthers, 2002), although the response rate and sample representativeness are in-line with prior PRP research (Blankertz & Robinson, 1996). Since age was measured using categories, the exact average, standard deviation, and range are not known. The background breakdown of this sample is consistent with prior PRP research (Blankertz & Robinson, 1996). However, it must be acknowledged that this sample was heavily female and White, which is not representative of the national population. In future research on PRPs, the United States Psychiatric Rehabilitation Association should consider an additional data collection method, such as focus groups, to make sure the opinions of under-represented PRP background groups, e.g., non-White, male, are considered. In addition, a stronger research design using a broader set of independent variables, as well as the complete MBI scale, and a more balanced (intrinsic and extrinsic items) measure of job satisfaction are recommended for future research.

Conclusion
Despite its constraints, this study contributes to the limited PRP literature. This research design used a stronger set of background controls in showing burnout to be negatively related to job satisfaction. Prior research, to the author's knowledge, has not done this with PRPs. This study's finding is consistent with general research with mental health workers showing that burnout significantly negatively affects job satisfaction (Faragher et al., 2005).
Building higher perceived quality peer relationships between PRPs and also to their supervisors can help to reduce PRP self-reported burnout (Morse et al., 2012). Lower job satisfaction can be a precursor to an employee leaving their job or even their occupation (Spector, 1985). Such research supports that burnout is an occupational hazard that needs to be continually monitored for all mental health workers.