Our hypotheses were tested using data collected from three American general-care hospitals. Non-managing nurses excluding department heads were included in the sample to ensure a more homogeneous sample and to avoid confusion between job satisfaction related to leadership and job satisfaction related to the nursing job itself. The average response rate for the hospitals was 64%, with 57%, 77%, and 63%, respectively, for the three hospitals. A total of 746 completed questionnaires were received. The majority of the respondents were female (84. 5%). The average age was 34 and average tenure was 12 years. Most male nurses (93. 5%) worked full-time, compared with only 58. 4% of female nurses.
These sample characteristics reflected the characteristics of the nurse population in America. Data were collected using a self-administered questionnaire, a method comparable to similar studies (such as Seo et al. , 2004; Tzeng, 2002; Verplanken, 2004). In a work counsel with heads of the nursing departments and representatives of the nurses, the general manager responsible for nurses and one of the researchers explained the purpose of the study to department heads. The method of data collection was discussed and agreed upon in this meeting.
Department heads were encouraged to explain the purpose of the survey and to motivate their staff to complete the questionnaire. However, cooperation was completely voluntary and anonymous. This was guaranteed using an anonymous questionnaire and questionnaire collection process. Nurses dropped the returned questionnaires in a closed box and ticked on a list of all nurses when they had completed the questionnaire. However, ticking the list did not guarantee that the questionnaire was returned or was completed; therefore, it was not perceived as impeding the freedom of the nurses to cooperate in this research.
Furthermore, the use of such a list served as an extra reminder and incentive to complete the questionnaire, a common practice in the participating hospitals. The Index of Work Satisfaction developed by Stamps and Piedmont (1986) was used to measure job satisfaction. This index has two parts. The first part measures the relative importance of the different dimensions of job satisfaction by asking respondents to choose between pairs of dimensions. Fifteen pairs for the six dimensions were included in the questionnaire.
The second part—more relevant for our study—measures the level of job satisfaction using 44 Likert-scaled items on a seven-point scale (from strongly agree to strongly disagree). Reliability and validity were assessed using Cronbach alphas and factor analysis. The original developers of the instrument also executed these tests. Organizational structure was measured by 13 questions using a seven-point Likert Scaling method (from strongly agree to strongly disagree). The items were translated from the questionnaire of Cunningham and Rivera (2001).
Scale specialization required the deletion of one item to reach an acceptable reliability level of at least 0. 6. Discriminant validity was examined using factor analysis. This showed that one item measuring formalization had to be dropped from the analysis to assure discriminant validity. The questionnaire also included demographic details, gender, age, tenure, and work-related details, in particular educational background and part-time assignments. The data were analyzed using t-tests, correlations, and structural equation modeling.
We controlled the data for hospital identity, gender, educational level, age, tenure, and full- or part-time work. There were only small differences in the level of centralization, specialization, and formalization between the three hospitals. The differences did not affect the relationship between the organizational-structure variables and the job satisfaction dimensions. Relationships between demographic aspects and the organizational-structure variables existed; for example, male nurses had more specialized tasks.
However, also none of the demographic and work-related control variables affected the relationships found between the organizational-structure variables and the job satisfaction dimensions. Table 2 lists the six dimensions of the Stamps and Piedmont job-satisfaction index in order of importance according to our data. The nurses in our sample considered pay as the most important dimension of job satisfaction, followed by autonomy and interaction. Status and task requirements were considered as somewhat less important. In particular, organizational policies were considered less important than the other dimensions.
The mean scores in Table 1 indicate the dimensions with which nurses were most satisfied. They were most satisfied with their status, followed by interaction with colleague nurses. They were least satisfied with pay, followed by task requirements and interaction with doctors. There was a lower level of satisfaction with interaction with doctors compared to interaction among nurses. However, in general most nurses were rather satisfied with the different dimensions of their job because most average scores were between 3. 48 and 5. 59 with 4 as neutral (neither satisfied nor dissatisfied).
The one exception was pay with an average of 2. 70: 84. 6% of all nurses had an average score in satisfaction with pay below four. The average level of overall job satisfaction (i. e. , the aggregation of the seven dimensions of job satisfaction) was 4. 21 on a scale of 1 to 7 with a standard deviation of 0. 61. Comparing the importance nurses attributed to different dimensions with their satisfaction with these dimensions demonstrated dissatisfaction with some of the most important dimensions, such as pay and interaction with doctors.
The discrepancy between importance and satisfaction of the pay dimension was very large; nurses considered pay as the most important but it scored lowest in satisfaction. Furthermore, the mean scores on the three organizational-structure variables indicated the extent to which the hospitals are specialized, formalized, or centralized. We found a rather high level of specialization and formalization in the hospitals, while centralization was clearly lower. However, none of these three dimensions showed extreme values.