Chapter 5 Results
5.11 Confirmatory Factor Analysis
5.11.4 Mediating Variable Analysis Results
Customer satisfaction has been identified as a mediator variable between service quality and customer loyalty (Caruana, 2002; Dado et al., 2012; Howat & Assaker, 2013; Ho, Kuo, & Lin, 2012; Olorunniwo, Hsu, & Udo, 2006; Yu & Ramanathan, 2012). Some studies also indicate that customer satisfaction fully mediates the effect of service quality on service loyalty (Caruana, 2002; Yu & Ramanathan, 2012). Therefore, the mediating variable analysis was designed in this study to test the effect of the mediating variable (Student Satisfaction) on the relationship between the exogenous variable (Service Quality) and the endogenous variable (Student Loyalty). To date, the Student
Involvement construct has not been tested as a mediator on the relationship between the exogenous variable (Service Quality) and the endogenous variables (Student Satisfaction and University Image). However, the statistically analysis used to test for mediating effect of the Student Involvement construct is the same as recommended by the literature (Baron & Kenny, 1986; Frazier, Tix, & Barron, 2004; Hair et al., 2010; MacKinnon & Fairchild, 2009; Meyers et al., 2013; Shrout & Bolger, 2002). Thus, the mediating variable analysis was also used to test the effect of the mediating variable (Student Involvement) on the relationship between the exogenous variable (Service Quality) and the endogenous variable (Student Satisfaction and University Image) in this study.
A mediating variable analysis starts by testing the direct (and statistically significant) effect of the exogenous variable (e.g. Service Quality) on the endogenous variable (e.g. Student Loyalty). Partial mediation occurs when the mediating variable (e.g. Student Satisfaction) enters the model. If the direct effect of the exogenous variable on the endogenous variable is reduced but still significant partial mediation is present. If the effect is reduced and no longer significant, then complete
mediation has occurred (Hair et al., 2010; Meyers et al., 2013; Shrout & Bolger, 2002). The following sections present the result of the mediating variable analysis.
5.11.4.1 The Mediating Effect of Student Satisfaction on the Relationship between Service
Quality and Student Loyalty
Initially, testing the statistically significant direct effect between the exogenous and endogenous variable found that Service Quality had a significant direct effect on Student Loyalty as the regression weight value was 0.769 which was statistically significant at the 0.001% level (see Table 5-63 or Figure 5-20).
Table 5-63 Standardized Causal Effect of the Direct Effect of Service Quality on Student Loyalty
Outcome
Determinant
Causal Effects
Result
Direct
Causal
Path
Critical Ratio
Student Loyalty Service Quality
.769
10.817
***Significant
Figure 5-20 The Direct Effect of Service Quality on Student Loyalty
Table 5-64 Goodness-of-Fit Results of the Direct Effect of Service Quality on Student Loyalty Model
Goodness-of-Fit Indices
Values
Chi-square (χ
2)
38.288
Normed Chi-square (χ
2/ df )
2.015
Goodness-of-Fit Index (GFI)
0.964
Root Mean Residual (RMR)
0.053
Comparative Fit Index (CFI)
0.990
Normed Fit Index (NFI)
0.980
The model-fit results for the direct effect of Service Quality on Student Loyalty model in Table 5-64 indicated a good model-fit to the sample data. All model fit indices were sufficiently satisfied with their relative recommended thresholds (Byrne, 2010; Hair et al., 2010).
Then, the mediator variable Student Satisfaction was included in the model. The direct effect between Service Quality and Student Loyalty was still statistically significant after Student
Satisfaction entered the model, even though the regression weight was reduced from 0.769 to 0.339. Therefore, Student Satisfaction is a partial mediator on the relationship between Service Quality and Student Loyalty (see Table 5-65 or Figure 5-21). In this case, Service Quality has a significant direct effect on Student Loyalty and also a significant indirect effect on Student Loyalty through the mediator variable Student Satisfaction.
Table 5-65 Standardized Mediating Effect of Student Satisfaction on the Relationship between Service Quality and Student Loyalty
Outcome
Determinant
Causal Effects
Result
Direct
Causal Path
Critical Ratio
Student Satisfaction
Service Quality
.751
12.041
***Significant
Student Loyalty
Service Quality
.339
4.804
***Significant
Student Loyalty
Student Satisfaction
.572
7.108
***Significant
Figure 5-21 The Mediating Effect of Student Satisfaction on the Relationship between Service Quality and Student Loyalty
Table 5-66 Goodness-of-Fit Results of the Mediating Effect of Student Satisfaction on the Relationship between Service Quality and Student Loyalty Model
Goodness-of-Fit Indices
Values
Chi-square (χ
2)
80.671
Normed Chi-square (χ
2/ df )
1.582
Goodness-of-Fit Index (GFI)
0.949
Root Mean Residual (RMR)
0.048
Comparative Fit Index (CFI)
0.990
Normed Fit Index (NFI)
0.972
Root Mean Square Error of Approximation (RMSEA)
0.048
The model-fit results for the mediating effect of Student Satisfaction on the relationship between Service Quality and Student Loyalty model in Table 5-66 indicated a good model-fit to the sample data. All model fit indices were sufficiently satisfied with their relative recommended thresholds (Byrne, 2010; Hair et al., 2010).
5.11.4.2 The Mediating Effect of Student Involvement on the Relationship between Service
Quality and University Image
Initially, testing the statistically significant direct effect between the exogenous and endogenous variable found that Service Quality had a significant direct effect on University Image as the
regression weight value was 0.786, which was statistically significant at the 0.001% level (see Table 5- 67 or Figure 5-22).
Table 5-67 Standardized Causal Effect of Direct Effect of Service Quality on University Image
Outcome
Determinant
Causal Effects
Result
Direct
Causal
Path
Critical Ratio
University
Image
Service Quality
.786
13.604***
Significant
Table 5-68 Goodness-of-Fit Results of the Direct Effect of Service Quality on University Image Model
Goodness-of-Fit Indices
Values
Chi-square (χ
2)
55.443
Normed Chi-square (χ
2/ df )
2.132
Goodness-of-Fit Index (GFI)
0.955
Root Mean Residual (RMR)
0.029
Comparative Fit Index (CFI)
0.987
Normed Fit Index (NFI)
0.976
Root Mean Square Error of Approximation (RMSEA)
0.067
The model-fit results for the direct effect of Service Quality on University Image model in Table 5-68 indicated a good model-fit to the sample data. All model fit indices were sufficiently satisfied with their relative recommended thresholds (Byrne, 2010; Hair et al., 2010).
Then, the mediator variable Student Involvement was included in the model. The direct effect between Service Quality and University Image was still statistically significant after Student Involvement entered the model, even though the regression weight was reduced from 0.786 to 0.407. Therefore, Student Involvement is a partial mediator on the relationship between Service Quality and University Image (see Table 5-69 or Figure 5-23). In this case, Service Quality has a significant direct effect on University Image and also a significant indirect effect on University Image through the mediator variable Student Involvement.
Table 5-69 Standardized Mediating Effect of Student Involvement on the Relationship between Service Quality and University Image
Outcome
Determinant
Causal Effects
Result
Direct
Causal Path
Critical Ratio
Student
Involvement
Service Quality
.688
9.284*** Significant
University Image
Service Quality
.407
7.106*** Significant
Figure 5-23 The Mediating Effect of Student Involvement on the Relationship between Service Quality and University Image
Table 5-70 Goodness-of-Fit Results of the Mediating Effect of Student Involvement on the Relationship between Service Quality and University Image Model
Goodness-of-Fit Indices
Values
Chi-square (χ
2)
109.760
Normed Chi-square (χ
2/ df )
1.770
Goodness-of-Fit Index (GFI)
0.937
Root Mean Residual (RMR)
0.053
Comparative Fit Index (CFI)
0.985
Normed Fit Index (NFI)
0.966
Root Mean Square Error of Approximation (RMSEA)
0.056
The model-fit results for the mediating effect of Student Involvement on the Relationship between Service Quality and University Image model in Table 5-70 indicated a good model-fit to the sample data. All model fit indices were sufficiently satisfied with their relative recommended thresholds (Byrne, 2010; Hair et al., 2010).
5.11.4.3 The Mediating Effect of Student Involvement on the Relationship between Service
Quality and Student Satisfaction
Initially, testing the statistically significant direct effect between the exogenous and endogenous variable found that Service Quality had a significant direct effect on Student Satisfaction as the regression weight value was 0.75, which was statistically significant at the 0.001% level (see Table 5- 71 or Figure 5-24).
Table 5-71 Standardized Causal Effect of Direct Effect of Service Quality on Student Satisfaction
Outcome
Determinant
Causal Effects
Result
Direct
Causal
Path
Critical Ratio
Student
Satisfaction
Service Quality
.751
11.987***
Significant
Figure 5-24 The Direct Effect of Service Quality on Student Satisfaction
Table 5-72 Goodness-of-Fit Results of the Direct Effect of Service Quality on Student Satisfaction Model
Goodness-of-Fit Indices
Values
Chi-square (χ
2)
31.486
Normed Chi-square (χ
2/ df )
1.657
Goodness-of-Fit Index (GFI)
0.970
Root Mean Residual (RMR)
0.032
Comparative Fit Index (CFI)
0.993
Normed Fit Index (NFI)
0.984
Root Mean Square Error of Approximation (RMSEA)
0.051
The model-fit results for the direct effect of Service Quality on Student Satisfaction model in Table 5- 72 indicated a good model-fit to the sample data. All model fit indices were sufficiently satisfied with their relative recommended thresholds (Byrne, 2010; Hair et al., 2010).
Then, the mediator variable Student Involvement was included in the model. The direct effect between Service Quality and Student Satisfaction was still statistically significant after Student Involvement entered the model, even though the regression weight was reduced from 0.751 to 0.504. Therefore, Student Involvement is a partial mediator on the relationship between Service Quality and Student Satisfaction (see Table 5-73 or Figure 5-25). In this case, Service Quality has a significant direct effect on Student Satisfaction and also a significant indirect effect on Student Satisfaction through the mediator variable Student Involvement.
Table 5-73 Standardized Mediating Effect of Student Involvement on the Relationship between Service Quality and Student Satisfaction
Outcome
Determinant
Causal Effects
Result
Direct
Causal Path
Critical Ratio
Student
Involvement
Service Quality
.689
9.243*** Significant
Student Satisfaction
Service Quality
.504
7.135*** Significant
Student Satisfaction Student Involvement
.358
4.912*** Significant
Figure 5-25 The Mediating Effect of Student Involvement on the Relationship between Service Quality and Student Satisfaction
Table 5-74 Goodness-of-Fit Results of the Mediating Effect of Student Involvement on the Relationship between Service Quality and Student Satisfaction Model
Goodness-of-Fit Indices
Values
Chi-square (χ
2)
78.595
Normed Chi-square (χ
2/ df )
1.541
Goodness-of-Fit Index (GFI)
0.950
Root Mean Residual (RMR)
0.056
Comparative Fit Index (CFI)
0.990
Normed Fit Index (NFI)
0.972
Root Mean Square Error of Approximation (RMSEA)
0.047
The model-fit results for the mediating effect of Student Involvement on the relationship between Service Quality and Student Satisfaction model in Table 5-74 indicated a good model-fit to the sample data. All model fit indices were sufficiently satisfied with their relative recommended thresholds (Byrne, 2010; Hair et al., 2010).