1.       Introduction

Higher productivity is associated with job satisfaction of employees. Thus, healthcare managers and executives are interested in identifying organizational mechanisms for increasing employees’ job satisfaction. The purpose of this empirical study is to examine the relative importance of personal and work climate factors affecting the variability in job satisfaction of professional nurses in a complex and large medical center located in Taiwan whereby a single payer or national health insurance is operative.

Based on the organizational literature on the relationship between work climate and job satisfaction, this study has raised specific four research questions as follows:

1.       Is nurse job satisfaction measurable by self-reported indicators?

2.       To what extend the safety culture, work pressure and collaboration in a large complex medical center can measure the construct of work-climate perceptions?

3.       Does job satisfaction vary by the rank of nursing staff?

4.       What is the relative contribution of each of work climate variables and personal factors in explaining the variation in nurse job satisfaction?

2.       Related Research

In conducting a systematic review of factors contributing to nurse job satisfaction, Bronwyn, Bonner and Pryor [1] found that both inter- and intra-hospital characteristics were important to understand the variability in job satisfaction. Avery, Smillie, & Fife-Schaw, argue that “job satisfaction is typically conceptualized as a multidimensional construct encompassing satisfaction with one’s manager, peers, pay, promotional opportunities, job security and company policy” [2]. This does not come as a surprise if one accepts that people have different views on goal attainments and preferences. Some researchers argue that job or work satisfaction is related to organizational culture [3], safety climate [4], work environment [5], ethical climate [6], work stress or strain [7], collaboration in nurse professional practices [8], and safety culture [9] within hospitals. 

Blegen [10] performed meta-analysis of 48 studies that addressed job satisfaction among nurses. There were a total of 15,048 subjects. She found the strongest perceptual correlates to job satisfaction were stress, communication with supervisor, and communication with peers. By contrast, age and professionalization had low correlations. Krueger et al., [11] examined staff perception of organizational factors that impact job satisfaction. They administered a survey to 1,819 employees at six independent health care organizations in Canada and tested for 40 possible predictors of job satisfaction. They identified best predictors, such as communication, keeping employees employed, teamwork, good supervisor, social support, infrequent excessive work, enough time to get work done, job classification and pay level. They concluded that job satisfaction is a “multidimensional construct and is a product of the global evaluation of one’s work place and context” [11].

This study builds on Avery, Smillie, & Fife-Schaw [2] conceptualization of job satisfaction but it also examines whether there are differences in job satisfaction with respect of the staff rank. Chien, Wan and Chen [12] conducted a path analysis of predictors of teamwork and collaboration among professional nurses in Taiwan.  They found that work related indicators are important predictors of the variability in teamwork and collaboration among nurses.

3.       Materials and Methods

3.1.  Study Design

A cross-sectional survey was conducted within a single medical center with more than 4,000 professional (registered) nurses. 

3.2.  Sample

The data for this study were collected in 2008 by a structured survey questionnaire. There was a total of 3,462 professional nurse respondents.

3.3.  Measurement of the Study Variables

The survey instrument included 40 questions pertaining to teamwork, safety culture, work pressure, management integrity, collaborations, and job satisfaction.  All of these constructs were scored from 1 (strongly disagree) to 5 (strongly agree) in a Likert scale. In addition, demographic information such as gender, staff position (rank), age, educational level, fulltime status, and years of employment was also collected and coded for each respondent.   The measurement indicators and operational definitions are presented in Appendix 1.

3.4.  Analytical Approach

In order to provide answer to the four research questions, this study utilize confirmatory factor analysis (CFA) to validate the construct validity of selected indicators for work climate and for job satisfaction, and structural equation modeling (SEM) to determine the structural relationships of personal and organizational factors to nurse job satisfaction.  

Prior to conducting multivariate analysis, the dataset was cleaned. Descriptive and correlations analyses of the data were performed. The use confirmatory factor analysis is suitable for validation of the measurement models of the endogenous construct of job satisfaction, and the exogenous construct of work-related perceptions as CFA explains the variation and covariation in a set of observed variables in terms of a set of theoretical, unobserved factors [13].  After the two measurement models were validated and further tested by multiple group analysis for supervisors and staff nurses, structural equation modeling was performed. SEM is useful for this analysis as it allows to theoretically specify complex relationships and then to test whether these relationships are present in the sample data [14]. Another advantage of the use of SEM is that it allows for testing the structural relationships among multiple constructs relevant to work climate at a large and complex medical center. The statistical software (i.e., IBM’s SPSS-AMOS) was used.

4.       Results

4.1.  Descriptive Statistics

Prior to the validation of the measurement models and the covariance structural model, we screened for missing values.  The descriptive analysis showed that a number of variables had missing values. In particular, the variables age, years of employment, educational attainment, fulltime status and rank had between 88 and 219 missing data points. Bearing in mind that the goal of this analysis is to validate the proposed theoretical model, all missing data points were deleted list wise.  This means that if a data point was missing in one variable, the entire case was deleted. After the deletion of respondents with missing values, the final data set contained a total of 2,969 nurse observations. Notwithstanding that a large number of data points were lost, the sample is still sufficiently large enough to perform multivariate analysis. (Table 1)

The descriptive statistics showed median of all questions with a value of 4 and the mean values between 3.53 and 3.93. The skewness of the distribution is -12.022 (-0.541/0.045) and the kurtosis is -3.511(-0.316/0.090). However, the distribution is relatively normal sufficing for multivariate analysis.

Prior to the validation of the measurement model, we examined the multi collinearity among the measurement variables or indicators.  This is aligned with the fact that the measurement model should be built on the theoretical specification and highly correlated indicators should not be used in the measurement model (Wan, 2002).   Bearing in mind that this study aims to examine if there is a difference in job satisfaction between staff nurse (coded 1) and nursing supervisors (coded 2), two correlation matrices were presented. The first correlation matrix presented in Table 2 is for group 1 (staff) and the second correlation matrix presented in Table 3 is the correlation matrix for group 2 (supervisors).

Tables 2 and 3 shows that there is a statistically significant correlation at 0.01 level among all variables, irrespective of staff rank. No serious multi collinearity issue was observed among the variables. The highest correlation is 0.788, between the indicators I17 (this a good place to work) and I18 (I am proud to work in this place).  However, as no correlation between the variables is above 0.80, there is no need to exclude any variables from the measurement model of job satisfaction.

4.2.  Confirmatory Factor Analysis (the Measurement Model of Job Satisfaction)  

Initially, the measurement model of job satisfaction with the total sample was performed. The five selected indicators used in the specification of the measurement model of job satisfaction were theoretically grounded on the literature. The endogenous construct of job satisfaction includes five indicators I15 - I19. The questions referred to whether the person likes his/her job (I15), if the work is like a family (I16), whether the hospital is a good place to work (I17), if he/she is proud to work at that place (I18) and whether the morale is high there (I19). The 5-indicators measurement model of the endogenous construct of job satisfaction was validated by confirmatory factor analysis using SPSS-AMOS.

The measurement model of the endogenous construct of job satisfaction is pictorially represented in Figure 1. All of the five factor loadings of the indicator variables were statistically significant at 0.05 level. The goodness of fit statistics in Table 4 shows that the model is reasonably fitted with the data when correlated measurement errors are not considered.

When the revised/nested measurement model with two correlated measurement errors for the indicators was examined, an improvement of the model fit to the data was achieved.  This revised and nested model is presented in Figure 2. (Table 5)

The goodness of fit statistics for the revised measurement model of job satisfaction indicates that the model has an excellent fit to the data. Therefore, there is no need for further improvements of the model.  In addition, this answers the first research question that job satisfaction can be validly measured by these five observed variables (I15-I19). Importantly, the indicator I17 (a good place to work with a factor loading of 0.90) is the most dominant variable for measuring job satisfaction.

4.3.  Multiple Group Analysis of Job Satisfaction by Staff Position

In order to provide answer to the second research question on the relationship between the five indicators I15-I19 and job satisfaction, irrespective of the rank, confirmatory factor analysis was conducted using multiple group analysis. Comparison was made between Group 1 (nurse staff) and Group 2 (nursing supervisors). The measurement models for both groups with the same specifications had comparable factor loadings; therefore, only Group 1 is presented.  

First, the factor loadings examined for all of the observed indicators showed statistically significant at 0.05 level. The goodness of fit statistics indicates that the measurement model for job satisfaction has a very good fit to the data, irrespective of staff rank or position in the nursing department. (Table 6)

Notwithstanding the fact that the model has good fit to the data, there is a relatively small difference in standardized regression weights (factor loadings) between the two groups.  The factor loadings for both groups are presented below.

Table 7 illustrates that although the standardized estimates are not exactly same for the two groups, the indicator I17 (good place to work) has a factor loading of .906 for group 1 (staff) and 0.871 for group 2 (supervisor) as it is the most reliable indicator in the measurement model of job satisfaction for both nursing groups.  As this a  good model fit, it answers the second research question that the five observed variables are good indicators for measuring job satisfaction, irrespective of the rank (or position) of professional nurse employees.

4.4.  Work Climate

The measurement model of work-related perceptions is conceptualized as measured by safety culture, work pressure and collaboration.  Appendix 1 presents the definitions and operational measures. Safety culture is measured by observed variables I7- I15 variables, work pressure by I20-I23 and collaborations is measured with I38-I40. The generic measurement model of the exogenous construct of work related perceptions is illustrated in Figure 4. All of the observed variables in the measurement model have critical values larger than 1.96 which means that they are statistically significant at 0.05 level. However, the goodness of fit statistics indicates that the model is not perfectly fitted to the data.  (Table 8)

Figure 4: The measurement model of work climate with three sub-latent constructs (Safety culture, work pressure, and collaboration).

The above statistics indicate that the model can be further revised and improved. Therefore, modification indices were examined and the model was revised. The revised model is presented in Figure 5.

In this measurement model, all factor loadings were statistically significant at 0.05 level. The goodness of fit of the revised model with correlated error measures indicates that the measurement model fits the data reasonably well and no additional changes should be makes (Table 9).  This provides answer to the third research question and indicates that the perceptions of work climate are shared in common with three indicators such as safety culture, work pressure and collaboration.

4.5.  Covariance Structure Model

After validating the measurement models, we combined the exogenous (predictor) and endogenous (response) variables in a structural equation model to determine the relative effect of each of the predictors on job satisfaction.  The covariance structural model is presented below.

In order to evaluate the structural model, the factor loadings were examined. This revealed that the variables rank and educational attainment were not statistically significant at 0.05 level. In particular, the critical value of the factor loading of education is 1.129 and the critical value of staff rank is 1.283. As there were not statistically significant, we removed them from the revised predictive model. A final model is presented in Figure 6.

All of the factor loadings in the revised model are statistically significant.  In addition, the goodness of fit statistics indicates that the model is a reasonable fit to the data. (Table 10)

The goodness of fit statistic reveals that the model has a good fit to the data. In an absence of new theoretical or empirical information, the model was not further improved. With regards to the relative importance of the exogenous indicators or predictors in explaining the variation in nurse job satisfaction, most important predictor is safety culture with a standardized regression coefficient of 0.675. Importantly, this study also demonstrates that work pressure has a negative impact on job satisfaction (standardized regression coefficient of -0.072). Older and male professional nurses were more satisfied with their work at the medical center than younger and female nurses.

5.       Discussion 

5.1.  Implications 

This study provides relevant answers to four research questions. For the first research question, we demonstrated the validity of nurse job satisfaction measured by five indicators (I15-I19). The literature suggests that additional indicators such as salary level, compensation, mobility and flexibility of the work schedule could be further considered.  With regards to the second research question, the five observed variables are shown to be good indicators for measuring job satisfaction, irrespective of nursing position (rank). Thus, these indicators can be effectively used as the measurement variables for nurse job satisfaction without concern about the rank of a nursing position. With regards to the third research question, work climate can be adequately reflected by safety culture, work pressure and collaboration. However, there is an unexplained common variance not accounted for by the three work climate constructs. The organizational culture, as an additional work climate related construct, should be included in the future construction of a better measurement instrument of work climate. The literature also suggests that additional sub-constructs such as empowerment, autonomy, communication, and fulfillment could be included in the measurement model. With regards to the fourth research question, the covariance structure model of perceived work climate and job satisfaction of nurses offers evidence in support of the relative contribution of each predictor variable of work climate and personal characteristics for explaining the variability in job satisfaction.  Safety culture is one of the most influential predictors of job satisfaction of nurses. 

5.2.  Limitations 

This cross-sectional study has two limitations. First, the sample of professional nurses drawn from a single medical center in Taiwan is relatively large.  However, it might not necessarily represent nursing employees in other large hospitals.  Future research needs to select a diverse sample of nurses from multiple hospitals.  Second, job satisfaction of nurses may be influenced by nursing unit characteristics such as the quality of administrative and technical support, the type of nursing systems, specialties in nursing practice, and unit-level size.  Future research on nurse job satisfaction should employ a two-level design in collecting data at the nurse and nursing unit levels.  Thus, the relative effects of personal and work climate factors on job satisfaction of nurses can be ascertained in multi-hospital studies. 

6.       Conclusion             

Job satisfaction could affect turnover [7,15,16]. Consequently, hospitals would be sage to pay attention to the determinants of job satisfaction. While compensation may be an important contributing factor to staff turnover, other perceptual aspects of work environment may also influence job satisfaction. Thus, job satisfaction is a multi-dimensional construct influenced by a complex set of personal and organizational factors. This study confirms the fact that safety culture is an important factor influencing nurse job satisfaction.  Furthermore, nurse job satisfaction does not vary by their ranks.

This study revealed how work climate indicators such as safety culture, work pressure and collaboration were associated with job satisfaction of registered nurses in Taiwan. Safety culture, the strongest indicator of work climate, accounted for a majority of variance in job satisfaction of nurses.  Hospital and nursing executives should pay greater attention to enhance better safety culture.  Thus, hospitals can achieve optimal retention of professional nurses by implementing and practicing safety culture activities.

Figure 1: Measurement model of job satisfaction for the total sample.

Figure 2: A revised/nested measurement model of job satisfaction with correlated errors.

Figure 3: Measurement model of job satisfaction (Group 1).

Figure 4: The measurement model of work climate with three sub-latent constructs (Safety culture, work pressure, and collaboration).

Figure 5: A revised/nested measurement model of perceived work climate with Correlated Errors.

Figure 6: Generic model of Predictors of Job Satisfaction.

Appendix 1: The Description of Study Variables and Measurements.

Table 1: Descriptive statistics of the study variables^* (N=2,969).

Table 2: Correlations matrix for group 1(staff nurses) (N=2,279).

Table 3: Correlations matrix for group 2 (Supervisors) (N=690).

Table 4:  Goodness of fit statistic for the measurement model of job satisfaction.

Table 5: Goodness of fit for revised measurement model of job satisfaction.

Table 6:  Goodness of fit statistic for multiple-group measurement model of job satisfaction.

Table 7: Factor loadings for the measurement model of job satisfaction for nursing groups.

Table 8: Goodness of fit statistic of the measurement model of perceived work climate.

Table 9: Goodness of fit statistics of the measurement model of perceived work climate (revised model).

Table 10: Goodness of Fit Statistics for revised Covariance Structure Model of Job Satisfaction.

Table 11: Regression weights of predictors of job satisfaction.

1.       Bronwyn H, Bonner A, Pryor J (2010) Factors contributing to nurse job satisfaction in the acuter hospital setting: review of recent literature. J Nursing Management 18: 804-814.

2.       Avery RE, Smillie LD, Fife-Schaw CR (2015) Employee achievement orientations and personality as predictors of job satisfaction facets. Personality and Individual Differences 76: 56-61.

3.       Lin BY, Wan TTH, Hsu CP, Hung FR, Juan CW et al. (2012) Relationships of hospital-based emergency department culture to work satisfaction and intent to leave of emergency physicians and nurses. Health Services Management Research   25: 68-77.           

4.       Glendon AI, Litherland DK (2001) Safety climate factors, group differences and safety behavior in road construction. Safety Science 39: 157-188.

5.       Laschinger S, Heather K (2008) Effect of empowerment on professional practice environments, work satisfaction, and patient care quality: further test of nursing work life model. J Nursing Care Quality 23: 322-330.

6.       Tsai MT, Huang CC (2008) The relationship among ethical climate types, facets of job satisfaction, and the three components of organizational commitment: A study of nurses in Taiwan. J Business Ethics 80: 565-581.

7.       Andrews DR, Wan TTH (2009) The importance of mental health to the experience of job strain: An evidence-guided approach to improve retention. Journal of Nursing Management, 17: 340-351.

8.       Mark BA, Salyer J, Wan TTH (2003) Professional nursing practice: impact on organizational and patient outcomes. J Nursing Administration 33: 224-234.

9.       Aiken LH, Sermeus W, Van der Heeds K, Sloane DM, Busse R, et al. (2012) Patient safety, satisfaction, and quality of hospital care: Cross-sectional surveys of nurses and patients in twelve countries in Europe and the U.S.  British Med J 344: e1717.

10.    Blegen MA (1993) Job satisfaction: A meta-analysis of related variables. Nursing Research 42: 36- 41.

11.    Krueger P, Brazil K, Lohfeld L, Edward HG, Lewis D,  et al. (2002) Organizational specific predictors of job satisfaction: findings from a Canadian multi-site quality of work life cross-sectional survey. BMC Health Services Research 2: 6-8.

12.    Chien SF, Wan TTH, Chen YC (2012) Factors influencing teamwork and collaboration    within a tertiary medical center: A path analysis.  World J Methodology 2: 18-22.

13.    Wan TTH (2002) Evidence-Based Health Care Management: Multivariate Modeling Approaches.  Boston, MA:  Kluwer Academic Publishers.

14.    Weston R, Gore PA (2006) A brief guide to structural equation modeling. The Counseling Psychologist 34: 719-751.

15.    Curry JP, Wakefield DS, Price JL, Mueller CW, McCloskey JC (1985) Determinants of turnover among nursing department employees. Research in Nursing Health 8: 397- 411.

16.    Yeh Y, Lin BY, Lin WH, Wan TTH (2009)  Job stress: Its relationship to hospital pharmacists’ insomnia and work outcomes. Inter  J of Behavl Med 17: 143-153.