Evaluating the Regression Model of Overall Job Satisfaction

Get Complete Project Material File(s) Now! »

Employee Empowerment

The eleven items measuring the facets of employee empowerment – information sharing, rewards, training and trust – were subjected to principle component analysis (PCA). Prior to performing the PCA, the data for factor analysis were assessed. Inspection of the correlation matrix (Appendix 4) revealed the presence of a number of coefficients with a value of 0.3 and greater. The Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy was found to be 0.739, which exceeds the recommended value of 0.6. Bartlett’s Test of Sphericity reached statistical significance (p= 0.000).
The initial analysis went on to reveal the presence of three components whose eigen-values exceeded 1.0. Ideally, there would have been four components with the first rotation each measuring the four facets of empowerment. The three components (Appendix 6) explained 32.32%, 15.79% and 10.25% of the variance respectively. This means that these three factors should ideally be retained for rotation. However, Pallant (2007) states that Kaiser’s criterion of using eigen-values tends to retain too many factors in some situations. Therefore, the scree plot obtained with the first analysis was inspected (Appendix 5). This graph plots each of the eigen-values obtained in the analysis and involved indentifying a point on the plot at which the shape of the graph changes direction. All of the points above this point should then be retained. The scree plot from the analysis revealed only two points. To be absolutely certain to retain either two or three factors, an additional analysis known as parallel analysis was carried out. This involved using the Monte Carlo PCA Parallel Analysis computer software program. The number of variables, subjects and replications were put into the software which then calculated a randomly generated data set of which is referred
to as the criterion value. The output was then compared to the output from the PCA. Table PCA was also carried out on the fifteen items of contextual performance. An initial assessment of the data revealed that many coefficients with a value greater than 0.30 were present in the correlation matrix. The KMO was calculated at 0.794 and deemed acceptable gainst the recommended cut off value of 0.60. Bartlett’s Test of Sphericity reached statistical significance (p = 0.000). The initial analysis revealed a presence of four components with eigen-values of above 1.0, revealing variances of 34.49%, 16.46%, 10.06% and 6.71% respectively (Appendix 9). That meant retaining four components. However, the scree plot (Appendix 8) indicated only two points before a break in the graph could be seen. As previously mentioned, to be absolutely certain, parallel analysis was carried out to ascertain that only two or more components should be retained. The criterion values obtained from the parallel analysis is shown in Table 5.4 alongside the actual eigen-values and the decision to accept or reject the component.

READ ZnO as a semiconductor

Correlations and Multicollinearity

The correlations (Appendix 10) between the four independent variables in our model (training and rewards, information and trust, interpersonal facilitation and job dedication behaviours) and job satisfaction showed a relationship between the dependent variable of overall job satisfaction. Training and rewards correlate quite highly with overall job satisfaction at 0.481. Information and trust correlated with overall job satisfaction at 0.119. Interpersonal facilitation and job dedication correlated with overall job satisfaction at 0.233 and 0.284. In addition, we also checked that the correlation between each of the independent variables was not too high. Upon close inspection of the correlation matrix (Appendix 4), we can conclude that there is not a high inter-correlation between the independent variables.
In addition to checking for inter-correlations between the independent variables, the tolerance and VIF values in the coefficient output were also checked. Pallant (2007) explains that tolerance values indicate how much of the variability in the independent variables are not explained by other independent variables within the same model. Tolerance values of less than 0.10 indicate that multiple correlation with other variables is high and thus problems of multicollinearity may be evident. In the case of the resulting output from our analysis the tolerance values for all of the four independent variables are quite high and well above the 0.10 recommended cut-off value. Thus, we can safely assume that there was no violation of the multicollinearity assumption. The VIF values are simply the inverse of the tolerance values and any VIF values above 10 would be a concern indicating violations of multicollinearity. However, the VIF values in our regression model all range between 1.003 and 1.130 thus indicating very low multicollinearity.

1 Introduction
1.1 Background
1.2 The Scandic Hotels of Jönköping
1.3 The Problem Discussion
1.4 Purpose
1.5 Delimitations
2 Theoretical Framework
2.1 Job Satisfaction
2.2 Employee Empowerment
2.3 Contextual Performanc
2.3.1 Other Behavioural Patterns
2.3.1.1 Organizational Citizenship Behaviour .
3 Method
3.1 Research Approach
3.2 The Case Study Approach .
3.3 Data Collection
3.4 Data Analysis
3.5 Method Reliability and Validity
4 Empirical Data
4.1 Descriptive Statistics
4.2 Overall Job Satisfaction
4.3 Employee Empowerment
4.3.1 Information Sharing
4.3.2 Reward
4.3.3 Training
4.3.4 Trust & Influenc
4.4 Contextual Performance
4.4.1 Interpersonal Facilitation
4.4.2 Job Dedication
4.5 Summary
5 Data Analysis
5.1 Factor Analysis
5.1.1 Employee Empowerment
5.1.2 Contextual Performance
5.2 Standard Multiple Regressio
5.2.1 Correlations and Multicollinearity
5.2.2 Evaluating the Regression Model of Overall Job Satisfaction
5.2.3 Evaluating Contribution of the Dependent Variables in Overall
Job Satisfaction
6 Discussio
6.1 Employee Empowerment .
6.2 Contextual Performance .
7 Conclusion, Limitations and Further Research
7.1 Conclusion
7.2 Limitations & Further Research
7.2.1 Limitations
7.2.2 Further Research
References