Multi-Linear regression analysis

Một phần của tài liệu Factors influencing customer satisfaction a study of online shopping in vietnam (Trang 55 - 60)

4.3. Testing the research model and the hypotheses

4.3.3. Multi-Linear regression analysis

Multiple linear regression analysis is the appropriate technique to analyze the linear relationship between a dependent variable and multiple independent variables by estimating coefficients for the equation for a straight line (Hair et al., 2006).

Multiple linear regression analysis will be used to test the six hypotheses of this research.

Through correlation analysis, the chosen model is multiple linear regression, reflected by the following formulation:

β0 + β1 β2 β3 IQ + β4 CS + β5 DM + β6 MA + ε

In this multiple regression model, there are total of 7 variables, including 6 independent variables, namely Website Design (WD), Security and Privacy (SAP), Information Quality (IQ), Customer Service (CS), Delivery Method (DM), Merchandise Attribute (MA) and 1 dependent variable of Online Shopping Customer Satisfaction (OCS). Six independent variables are hypothesized as the factors influencing the dependent variable. The multi-linear regression analysis would be used to test the model. The author used Enter method to input all variables at once and find out the acceptable variables. The MLR analysis results are presented in Table 4.7, Table 4.8 and Table 4.9.

Table 4.7. MLR results using Enter technique R R Square Adjusted R

Square

Std. Error of the Estimate

Std. Error of the Estimate

.759a .576 .567 .67498 .67498

a. Predictors: (Constant), MA, CS, WD, DM, SAP, IQ b. Dependent Variable: OCS

Table 4.8. ANOVA

Model Sum of

Squares df

Mean

Square F Sig.

1 Regression 165.515 6 27.586 60.548 .000a

Residual 121.646 267 .456

Total 287.161 273

a. Predictors: (Constant), MA, CS, WD, DM, SAP, IQ b. Dependent Variable: OCS

Table 4.9. MLR variables coefficients Model Unstandardize

d Coefficients

Standardized Coefficients

t Sig. Collinearity Statistics

B Std.

Error

Beta Tolerance VIF

(Constant) .424 .281 1.510 .132

WD .112 .056 .108 2.012 .045 .550 1.820

SAP .039 .055 .040 .706 .481 .489 2.044

IQ .159 .063 .151 2.532 .012 .446 2.241

CS .176 .046 .201 3.858 .000 .587 1.704

DM .286 .053 .301 5.386 .000 .506 1.975

MA .163 .047 .174 3.501 .001 .642 1.557

The output in table 4.7 shows the value of adjusted R2 (.567) is smaller than R2 (.576). It helps to ensure that the model is safer, because the accordance of research model would not be exaggerated. The indicator of adjusted R2 = .567 means that the compatibility of the model is 6.7%, or in other words, it’s about 56.7% variance of the dependent variable of Online Shopping Customer Satisfaction (OCS) could be explained by six corresponding factors, which are 6 independent variables in the model.

The coefficients showed in Table 4.9 are used to test the linear relationship between the dependent variable of OCS and independent variables of WD, SAP, IQ, CS, DM, MA and check whether or not the linear relationship between OCS variable and the whole group of independent variables (WD, SAP, IQ, CS, DM, MA). The value of statistical indicator F in Table 4.8 is 60.548 which is calculated based on R Square value at very small of significant level (p = 0.000), indicating that it would be safe to reject the null hypothesis H0: β1=β2 =β3 =β4 =β =β6=0, so the multi-linear regression model is suitable with the data and could be used.

The variance inflation factors (VIF) were scrutinized and all were found to be within the range of 1.557-2.241. Myers (1990) indicates that only if the value of VIF is above ten is there cause for concern about multicollinearity. Therefore, multicollinearity and autocorrelation were well within acceptable limits and not unduly influencing the regression estimates.

With the significance level of 5% chosen in normal studies, if p-value <.05, we could conclude that independent variables have impacts on dependent variables.

Results presented in the Table 4.9 showed that the p-values of 5 variables, including WD, IQ, CS, DM, MA, are smaller than .05, so we could conclude that these five variables have the statistical significance in the regression model and they have impacts on the online shopping customer satisfaction. Furthermore, because the values of Beta are all positive, the five independent variables have positive impacts on the dependent variables.

Due to the SAP variable has p-value = .481 > .05, so we could conclude that the Security and Privacy (SAP) does not have statistical significance in the regression model, or in other word, online shopping customer satisfaction would not be influenced by security and privacy.

In summary, the regression results indicated that the satisfaction of online shopping customers is influenced by 5 factors, namely Website Design (WD),

Information Quality (IQ), Customer Service (CS), Delivery Method (DM) and Merchandise Attribute (MA), and not be influenced by Security and Privacy (SAP).

The results of testing six hypotheses is showed in Table 4.10, in which five hypotheses, namely H1, H3, H4, H5, H6, were supported (p-value < .05), while the H2 was not supported (p-value > .05).

Table 4.10. Results of testing the hypotheses

Hypothese Unstandardized

Beta

Sig. Conclusion

H1: Website design is positively related to online shopping customer satisfaction.

.112 .045 Supported

H2: Security and Privacy are positively related to online shopping customer satisfaction.

.039 .481 Unsupported

H3: Information quality is positively related to online shopping customer satisfaction.

.159 .012 Supported

H4: Customer service is positively related to online shopping customer satisfaction.

.176 .000 Supported

H5: Delivery method is positively related to online shopping customer satisfaction.

.286 .000 Supported

H6: Merchandise attribute is positively related to online shopping customer satisfaction.

.163 .001 Supported

One possibly practical explanation for the reason why H2 is not supported, or online shopping customer satisfaction don’t be influenced by security and privacy when they purchase online is because Vietnam lacks a completed online transaction credit system, and majority of Vietnamese online shopping customers prefer to pay

by cash when they receive the goods from online shopper rather than by credit card via Internet. Therefore, Vietnamese online customer no need to input financial information directly in the website, then they don’t perceive security and privacy seriously. This explanation is supported by the results of Kim and Stoel’s (2004), that privacy and security have significant impact on online shopping intention but not on online shopping satisfaction. It’s also supported in context of Vietnam market according to report of Yahoo! Kantar Media Net Index Vietnam (2011), in which 93% Vietnamese online customer use cash as payment method when they purchase online.

The regression equation could be re-written as follows (excluded SAP variable and based on the unstandardized beta coefficients).

.424 + .112 * .159 * IQ + .176 * CS + .286 *DM + .163 * MA + ε

The standardized beta coefficients indicate the impact level of each independent variable on dependent variable. The linear regression equation could be re-written based on the standardized beta as follows:

OCS = .108 * WD .151 * IQ + .201 * CS + .301 *DM + .171 * MA According to the equation, the level of importance of factors influencing online shopping customer satisfaction is as follows: (1) Delivery Method, (2) Customer Service, (3) Merchandise Attribute, (4) Information Quality, (5) Website Design. And the research model is updated and presented as in Figure 4.1.

Một phần của tài liệu Factors influencing customer satisfaction a study of online shopping in vietnam (Trang 55 - 60)

Tải bản đầy đủ (PDF)

(97 trang)