Organization specific predictors of job satisfaction: findings from a Canadian multi-site quality of work life cross-sectional survey

Background Organizational features can affect how staff view their quality of work life. Determining staff perceptions about quality of work life is an important consideration for employers interested in improving employee job satisfaction. The purpose of this study was to identify organization specific predictors of job satisfaction within a health care system that consisted of six independent health care organizations. Methods 5,486 full, part and causal time (non-physician) staff on active payroll within six organizations (2 community hospitals, 1 community hospital/long-term care facility, 1 long-term care facility, 1 tertiary care/community health centre, and 1 visiting nursing agency) located in five communities in Central West Ontario, Canada were asked to complete a 65-item quality of work life survey. The self-administered questionnaires collected staff perceptions of: co-worker and supervisor support; teamwork and communication; job demands and decision authority; organization characteristics; patient/resident care; compensation and benefits; staff training and development; and impressions of the organization. Socio-demographic data were also collected. Results Depending on the organization, between 15 and 30 (of the 40 potential predictor) variables were found to be statistically associated with job satisfaction (univariate analyses). Logistic regression analyses identified the best predictors of job satisfaction and these are presented for each of the six organizations and for all organizations combined. Conclusions The findings indicate that job satisfaction is a multidimensional construct and although there appear to be some commonalities across organizations, some predictors of job satisfaction appear to be organization and context specific.


Background
There appears to be no one commonly accepted definition for quality of work life. In healthcare organizations, quality of work life (QWL) has been described as referring to the strengths and weaknesses in the total work environment [1]. Characteristics that describe the overall organi-zation are viewed as part of the behaviour and reward system of the staff working in that setting. Organizational features such as policies and procedures, leadership style, operations, and general contextual factors of the setting, all have a profound effect on how staff view the quality of their work life. QWL is an umbrella term which includes many concepts. Therefore, concentrating on only one job characteristic, whether it is wages or management style, is an inadequate approach to assessing QWL. Because the perceptions held by employees play an important role in their decisions to enter, stay with or leave an organization, it is important that staff perceptions be included when assessing QWL. And although job satisfaction is not QWL, perception of QWL is often assessed using job satisfaction surveys.
Previous studies have shown that low job satisfaction is a major cause of turnover among health care providers [2][3][4]. In addition, job satisfaction may affect the quality of service and organizational commitment [5][6][7][8][9] and may be a contributing factor associated with shortages of health care providers [10]. Such findings have recently increased interest in studying job satisfaction among health care providers [11]. The results of a 1993 meta-analysis of 48 studies looking at work satisfaction in over 15,000 nurses revealed that job satisfaction was associated strongly with reduced work stress, organizational commitment, communication with supervisors, autonomy, employee recognition, fairness, locus of control, years of experience, education, and professionalism. This study also found a strong relationship between job satisfaction and QWL for nurses [12].
After reviewing the literature on QWL and job satisfaction, and considering the wide variety of health care settings, situational contexts, and organizational structures (including management styles, reporting structures, staffing complements, and levels of training and experience) in which employees work, we hypothesized that the predictors of job satisfaction would vary depending on the organization. The purpose of this study was to identify organization specific predictors of job satisfaction within a health care system that consisted of six independent and distinct organizations located in five communities in Central West Ontario, Canada.

Setting
The settings for this study included six independent and distinct health care organizations providing varying levels and types of care. All six organizations were affiliated with the St. Joseph's Health System (SJHS) located in five Central West Ontario communities. Collectively, the SJHS is one of the largest corporations in Canada devoted to health care. At the time of the study (2000), the SJHS employed 5,486 full, part and casual time (non physician) staff. Additional information about of each of the six organizations and their respective communities is provided in Table 1.

Questionnaire development
Items included in the "Quality of Work Life Survey 2000" were selected after a review of the literature and extensive consultation between research team members and the QWL Task Force (a management group consisting of representatives from each of the six SJHS organizations). The initial selection of items was influenced by a recently published Canadian study [13] and reports from two metaanalyses [1,12]. The QWL Task Force then refined these items to consider, among other things, issues of accuracy, relevance, readability, grammar, potential for offensiveness, and appearance of cultural or gender bias. After several months of development, the instrument was pretested on a small group of staff at two of the participating organizations (Site 2 and Site 4 -see Table 1). This pretesting was done to ensure that individuals could follow the instructions associated with the format, to obtain estimates of the time required to complete the survey instrument, to identify items that were poorly written or ambiguous, and to identify an appropriate implementation strategy. The questionnaire and implementation strategies were revised accordingly. The Co-worker and supervisor support section included 10 closed-ended and 1 open-ended questions. A 3-item supervisor social support scale included questions about supervisor helpfulness, concern about the welfare of employees, and ability to facilitate effective interaction among employees. Co-worker support was measured by a 7-item scale reflecting the extent to which co-workers were seen as competent, understanding, and supportive of employees. Both scales where adapted from Woodward et al. (1999) [13].
The Teamwork and Communication section included 9 closed-ended and 1 open-ended questions. For determining teamwork, a 7-item scale was adapted from Taylor and Bowers (1972) to measure the extent to which one's work unit coordinates efforts, solves problems and works together effectively [14]. A 2-item scale developed for this project measured how communication was practiced within the organization.
The Job Demands and Decision Authority section included 15 closed-ended and 1 open-ended questions. It included a 4-item scale adapted from Brosnan and Johnson (1980) to measure clarity regarding responsibilities, workloads and conflicting demands [15]. There was also a 9-item scale adapted from Karasek et al. (1998) to measure the extent to which respondents' jobs gave them autonomy or decision-making latitude [16], and 2 questions which reflected the demands of one's work [17].
The Characteristics of Your Organization section included 6 closed-ended and 1 open-ended questions. This section was adapted from Woodward et al. (1999) and included a 4-item scale that inquired about the extent to which the organization encouraged the best efforts from staff, and how employees were treated [14]. Two additional questions examined the extent to which staff were kept informed, and organizational recognition of employee contributions.
The The Staff Socio-Demographic Information section included 10 closed-ended questions (developed for this project) to collect information on gender, age, marital status, education, length of employment, supervisory status, time spent on job activities, job status and job classification.
Within each of the first 8 sections, employees were asked to circle the response that best described their feelings using 5-point Likert scales. Employees were also asked for written comments pertaining to each of the sections and were provided space to comment on other issues they felt were important.

Survey Procedure
Because of the diversity of organizations and staff within the SJHS, it was decided by the QWL Task Force, organization administrators and researchers that the implementation of the survey would be customized to best fit each of the organizations. It was felt that a varied approach would be more feasible for the organizations and that this would help maximize response rates. Although the procedures were not identical, all of the organizations provided as a minimum: advance notification (written or voice mail) of the survey to all staff (eligibility was based on whether the worker was active on the organization's pay roll at the time of the study and was not a physician); access to questionnaires for all staff (the QWL Task Force felt that each staff member in the SJHS should have the opportunity to complete a questionnaire); one or more reminder notices (e.g., letters, newsletters, voice mail, personal communication); and sealed drop off boxes for completed questionnaires. Pilot testing of the questionnaire revealed that employees felt that tracking individual employees for the purpose of follow-up (i.e., to increase response rate), violated the perception of anonymity and confidentiality. Therefore, to help ensure anonymity and confidentiality, follow-up attempts were limited to general reminder notices to all staff.

Analysis
All closed-ended (or quantitative) responses were entered directly from the questionnaires into SPSS (version 10.0.5 for Windows, SPSS, Inc., Chicago, 1999). Prior to data analysis, most of the survey questions were re-coded. Questions which asked participants to select one response within a five point scale (never to always; very dissatisfied to very satisfied; very poor to very good; no, definitely not to yes, definitely) were collapsed into two categories. For example, for the response scale (1=very dissatisfied, 2=dissatisfied, 3=not sure, 4=satisfied, 5=very satisfied) those who indicated they were either satisfied or very satisfied were re-coded as "satisfied" while all others were re-coded "not satisfied" by default. In several instances, it was appropriate to combine two or more of the questions into a composite scale score. See "Questionnaire Development" section and Additional File: Statistically Significant Organization Specific (Univariate) Predictors of Job Satisfaction for additional details on how the composite scale scores were calculated. In total, there were eight scale scores (supervisor social support; co-worker support; teamwork; communication; role clarity; decision latitude; organization/staff relations; patient/resident care). Scale scores were generated by summing the participant responses (i.e. one to five) for all questions that made up the scale. In the rare situation where a participant failed to answer one or more of the questions that made up a scale score, missing values were replaced with mean values for that organization. Scale scores were categorized into meaningful dichotomous categories prior to analysis (e.g., satisfied or not satisfied).
For the purpose of this study, QWL was operationally defined using the global question "Overall, how satisfied are you with your job?". Employees rated job satisfaction from very dissatisfied to very satisfied using a five point scale (very dissatisfied, dissatisfied, not sure, satisfied, very satisfied). For the analysis, however, those indicating they were either satisfied or very satisfied were considered to be "satisfied" with their jobs. All others were considered "not satisfied" with their jobs.
Prior to analysis, study researchers reached a consensus on which survey questions to include as potential predictors of job satisfaction. In total, there were eight scale scores and 32 questions that were rationalized a priori as potential predictors of job satisfaction. Data from each of the organizations, as well as all of the organizations combined (representing the SJHS), were analyzed separately to identify predictors of job satisfaction. T-test, chi-square analyses and, when appropriate, Fisher exact tests were used to determine which of the variables were statistically associ-ated with job satisfaction i.e., were potential predictors of job satisfaction. Descriptive information (numbers and percentages) for each of the variables was calculated by whether or not staff were satisfied with their jobs. In addition, p-values, odds ratios, and 95% confidence intervals for the odds ratios were calculated for each potential predictor of job satisfaction.
Separate logistic regression analyses were used to identify the best predictors of job satisfaction for each organization and for all organizations combined (SJHS). Only variables which had a statistically significant association with job satisfaction were included in these analyses. Adjusted odds ratios and corresponding 95% confidence intervals are reported for each organization and the SJHS. The logistic regression analyses produces odds ratios which have been simultaneously adjusted for all other variables in their respective final models. The goodness of fit of the logistic regression models were assessed using the rhosquared statistic [18]. A rho-square value between 0.20 and 0.40 suggests a very good fit of the model. A probability level of <0.05 was used to determine statistical significance. SPSS and Epi-Info (version 6.04a, Centers for Disease Control and Prevention, Atlanta, 1995) were used for statistical computations. Table 1 provides additional information about each of the six health care organizations, including the type of organization, number of staff, number of beds or visits/year, and the size of the community where the organization was located.

Respondent participation rate
Response rates are often used as an indicator of the representativeness of a sample of respondents. Of the combined 5,486 staff, 1,819 (33.2%) returned a completed questionnaire. Organization specific response rates varied from 25.3% to 55.3% (Table 2). In an attempt to assess the representativeness of respondents, a comparison was made of available socio-demographic information between respondents and all staff within each of the organizations. Overall, female employees were more likely to respond than male employees (it should be noted, however, that the vast majority of staff (82% to 98%), were females within each of these organizations), as were fulltime employees compared to part-time, casual or temporary employees. There were also some differences in respondents, across organizations, based on job classification. All organizations, however, had respondents within each job classification. A statistical estimating procedure was also used to assess how accurately respondents represent staff at each of the organizations [19]. This calculation suggests that the organization specific findings were accurate plus or minus 3.6% to 8.8%, 19 times out of 20 (Table 2).

Potential predictors of job satisfaction
Organization specific and combined SJHS (univariate) analyses (t-test, chi-square analyses and, when appropriate, Fisher exact tests) were used to determine which of the potential predictor variables were statistically associated with job satisfaction. Included in these analyses were the 40 potential predictor variables (

Best predictors of job satisfaction
Separate logistic regression analyses were then used to identify the best predictors of job satisfaction for each organization and for all organizations combined (SJHS). All variables found to be statistically associated with job satisfaction from the univariate analyses were entered into these logistic regressions analyses. The best predictors of job satisfaction are presented in Table 3. The ranking assigned to these variables relates to the order in which variables were added to the logistic regression models. For example, the rank "1" refers to the first variable that was added to the model i.e., the variable which best improved the fit of the model (or the most important variable). A more detailed description of the magnitude (as represented by the size of the odds ratios) and statistical significance (as represented by the 95% confidence intervals of the odds ratios) of the association between each of these predictors and job satisfaction is presented below for each organization and all organizations combined (SJHS). The best predictors of job satisfaction are again ranked according to their importance. All of the odds ratios presented below have been simultaneously adjusted for all other variables in their respective final logistic regression models. All logistic regression models achieved a rho-square between 0.20 and 0.40 suggesting they were very good (fitting) models for predicting job satisfaction.

Site 1 (community hospital)
The most important predictors of job

Site 5 (community hospital)
The most important predictors of job satisfaction were: 1) belief the organization carries out its Mission statement

Predictors of Job Satisfaction Site 1 Rank Site 2 Rank Site 3 Rank Site 4 Rank Site 5 Rank Site 6 Rank SJHS 3 Rank
Believes organization carries out its Mission Statement 1 Separate logistic regression analyses were done to identify the best predictors of job satisfaction for each of the organizations and for the System as a whole. 2 The rank reflects the order that variables were added to the logistic regression models. For example, the rank "1" refers to the first variable that was added to the model (i.e. the variable which best improved the fit of the model). 3 In order to account for the organizational setting, site was forced into the final logistic regression analysis for the SJHS. 4 Composite score from 2 (5-point) questions. A score of 8 to 10 indicates good communication. 5 Composite score from 9 (5-point) questions. A score of 36 to 45 indicates good decision latitude. 6 Composite score from 4 (5-point) questions. A score of 16 to 20 indicates good role clarity. 7 Composite score from 7 (5-point) questions. A score of 28 to 35 indicates good teamwork. 8 Composite score from 4 (5-point) questions. A score of 16 to 20 indicates good organization/staff relations. 9 Decision authority is a subscale of decision latitude. It was found to be a better predictor of job satisfaction for this organization than decision latitude. It is a composite score from 3 (5-point) questions. A score of 12 to 15 indicates good decision authority. 10

Discussion
The results of this survey were intended to assist decisionmakers in identifying key workplace issues, as perceived by employees, in order to develop strategies to address and improve the quality of working conditions for staff within each of the individual health care organizations and the SJHS as a whole. This research represents the first step of an ongoing process to ensure better QWL for employees. There are several positive attributes of this study. First, to our knowledge, it is the largest QWL investigation of health care workers in Canada with 1,819 completed interviews. Second, it is also unique in that we collected information from staff at six distinct and functionally diverse health care organizations. Third, because we could not find an "off-the-shelf" QWL instrument that suited our needs and collected all the information desired by key stakeholders, we developed (through a combination of modifying existing instruments and creating our own questions and scales) our own questionnaire. Finally, although the response rates were not as high as we would have hoped, the findings: appear to be consistent with what we expected a priori (the study's investigators had offices within 5 of the 6 organizations thus having inside knowledge about these organizations); appear consistent with the published literature; and were judged credible by management and staff at each of the sites. The statistical estimating procedure to assess how accurately respondents represent staff at each of the organizations also suggest that our findings were fairly representative of staff within these organizations, particularly the larger organizations.

Conclusions
The results of this research show that job satisfaction is a multidimensional construct and is a product of the global evaluation of one's work place and context. This report provides valuable information about how employees in specific health care settings view their work environment. A number of organization specific predictors of job satisfaction were identified as a result of this study. The implications of these findings are currently being deliberated as they relate to improving QWL within each of the six health care organizations that make up the SJHS. These findings, may also be of relevance and value to employees, researchers, evaluators, human resource planners and administrators of similar health care organizations.
The results of this survey can also be used as baseline measures against which the findings of future job satisfaction surveys can be compared. Such comparisons place this type of research within a continuous quality improvement framework.