4.5.1 The relationship between online information system quality dimensions and overall internetbanking service quality
Multiple linear regression analysis was employed to test hypothesis and enter approach is used to select all independent variables that would be put into the model. The relationships among the variables were tested via multiple regression analysis.
Regression models are as follows:
IBQ = β0 + β1*USE + β2*TIM + β3*CON + β4* SEC + β5* ACC + β6*AES +e
In this model, the dependent variable Overall internetbanking service quality (IBQ) would be shown by six independent variables Ease of use (USE), Timeliness (TIM), Contents (CON), Security (SEC), Accuracy (ACC), Aestheti (AES). These dimensions request for the analysis has been approved and then entered regression analysis.
The ANOVA table tests the acceptability of the model from a statistical perspective. Based on the result in table 4.11, ANOVA results showed that this model
50
had a F value of 45.005 at sig. = .000, it conducted that this regression model was suitable and explained by the independent variables mentioned.
Table 4.11: ANOVAb
Model
Sum of Squares Df
Mean
Square F Sig.
1 Regression 41.094 6 6.849 45.005 .000a Residual 30.437 200 .152
Total 71.531 206
a. Predictors: (Constant), AES, CON, SEC, USE, TIM, ACC
b. Dependent Variable: IBQ
In addition, the model summary table 4.12 showed that R2 = 0.574 and R2Adj=0.562 (smaller R2). It means that this model was fit and all independent variables already explained about 56.2% variance of the dependent ones.
Table 4.12: Model Summaryb
Model R R Square
Adjusted R Square
Std. Error of the Estimate
1 .758a .574 .562 .39011
a. Predictors: (Constant), AES, CON, SEC, USE, TIM, ACC b. Dependent Variable: IBQ
Moreover, in order to measure of the strength of the model fit, the standardized residual plots of the given dimensions were observed if having any value of error term for these dimensions. As shown in Histogram in Appendix 9, the shape of histogram approximately followed the shape of the normal curve, which had standard deviation equal 0.985 (σ # 1). Therefore, the regression assumptions were suitable. Also, after checking Scatter Diagrams (Appendix 9), all the residuals gathered around the standardized predicted value at "0" and the observed values were standing around a straight line. It proved that the normality assumption was not violated and the linear regression analysis could be used to test the relationship between the independent variables and the dependent variable of the model. Thus, the regression equation could be
51
re-written as follows:
IBQ = 0.132 + 0.137*USE + 0.181*TIM + 0.130 *CON + 0.196* SEC + 0.169* ACC + 0.143*AES +e
The relative importance of the significant predictors was determined by looking at the standardized coefficients. Accuracy (ACC) had the highest standardized coefficients and the lowest significance, which indicated that Accuracy was the best predictor. Analyzing whole table 4.13 result, the order of significance for predictors of Overall internetbanking service quality was Accuracy (ACC), Ease of use (USE), Security (SEC), Timeliness (TIM), Contents (CON), Aestheti (AES).
Table 4.13: Coefficientsa
Model
Unstandardized Coefficients
Standardiz ed
Coefficient s
t Sig.
Collinearity Statistics
B Std. Error Beta
Toleran
ce VIF
1 (Constant) .132 .238 .557 .578
USE .137 .035 .218 3.933 .000 .690 1.448
TIM .181 .052 .184 3.452 .001 .746 1.341
CON .130 .039 .162 3.327 .001 .894 1.118
SEC .196 .048 .212 4.068 .000 .782 1.279
ACC .169 .041 .234 4.160 .000 .672 1.489
AES .143 .051 .147 2.780 .006 .757 1.321
a. Dependent Variable:
IBQ (Overall Internet Banking Service Quality)
Morever, their part and partial correlations had nearly same value each other and Variance Inflation Factor (VIF) were smaller than 2 (table 4.13) . As shown in Histogram in Appendix 9, the shape of histogram approximately followed the shape of the normal curve, which had standard deviation equal 0.985 (σ # 1). Therefore, the regression
52
assumptions were suitable. Also, after checking Scatter Diagrams (Appendix 9), all the residuals gathered around the standardized predicted value at "0" and the observed values were standing around a straight line.
It means that other variables had already explained for these variables on the relationship of them with dependent variable Overall internetbanking service quality.
Therefore the hypothesis 1 is accepted.
H1a: Aesthetics is positively related to overall internet banking service quality H1b: Timeliness is positively related to overall internet banking service quality H1c: Contents is positively related to overall internet banking service quality H1d: Ease of use is positively related to overall internet banking service quality H1e: Security is positively related to overall internet banking service quality H1f: Accuracy is positively related to overall internet banking service quality 4.5.2 The relationship between Overall internet banking service quality and Customer Satisfaction
The MLR results are presented in table 4.14, 4.15 and table 4.16. The regression coefficient R2Adj is 0.568. ANOVA results showed that this model had a F value of 272.046 at sig. = .000. These results allowed us to reject the null hypothesis that the regression coefficients equal zero. To decide which hypothesis is statistically significant, partial regression coefficients were tested with t value.
Table 4.14: Model Summaryb
Model R R Square Adjusted R Square Std. Error of the Estimate
1 .755a .570 .568 .42126
a. Predictors: (Constant), IBQ b. Dependent Variable: SAT
53
Table 4.15: ANOVAb
Model
Sum of
Squares Df Mean Square F Sig.
1 Regression
48.278 1 48.278 272.04
6 .000a
Residual 36.380 205 .177
Total 84.658 206
a. Predictors: (Constant), IBQ c. Dependent Variable: SAT
Table 4.16:Coefficientsa
Model
Unstandardized Coefficients
Standardized Coefficients
t Sig.
Collinearity Statistics
B Std. Error Beta
Toleran
ce VIF
1 (Consta
nt) .674 .189 3.574 .000
IBQ .822 .050 .755 16.494 .000 1.000 1.000 a. Dependent Variable:
SAT
The results in table 4.16 show that latent variable are statistical significant . Thus, the hypothesis 2 is accepted:
H2: Overall internet banking service quality is positively related to Customer satisfaction.
In order to measure of the strength of the model fit, the standardized residual plots of the given dimensions were observed if having any value of error term for these dimensions. As shown in Histogram in Appendix 10, the shape of histogram
54
approximately followed the shape of the normal curve, which had standard deviation equal 0.998 (σ # 1). Therefore, the regression assumptions were suitable. Also, after checking Scatter Diagrams (Appendix 10), all the residuals gathered around the standardized predicted value at "0" and the observed values were standing around a straight line. Thus, the regression equation could be re-written as follows:
SAT = 0.674 + 0.822* IBQ + e
SAT: Customer Satisfaction
IBQ: Overall internet banking service quality