Correlation analysis and linear regression analysis

Statistical results showed that there is not the linear correlation between scales of the customer satisfaction within research model (Appendix 6). Thus, there is not

Service Capacity

Tangible Facilities

Image

Reliability

Empathy

Cost

Satisfaction

33

multicollinearity in regression analysis.

Besides, analyzing results show the level of linear correlation between each of above scales with the scale of customer satisfaction. In which, the highest correlation relation is between the scale of “tangible facilities” and “customer satisfaction” scale with r = 0.446.

4.8.1.2 Testing model and research hypotheses The multiple linear regression model has the form:

SHL=β0+ β1*PTHH + β2*KNPV + β3*HA +β4*CP * + β5*TCAY + β6* DCAM

To perform the multiple linear regression analysis, the variables are put into the model by Enter method. Testing standard is the standard built into test method of F statistical value, determining the corresponding probability of F statistical value, testing the suitability level between sample and overall through the coefficient of determination R². Diagnostic tool helping to detect the existence of multicollinearity in data which is assessed collinear level degenerating the estimated parameter is Variance inflation factor – VIF. The principle here is when VIF exceeds 10; it is the sign of multicollinearity (Trong & Ngoc 2005).

The result of multiple regression analysis shows that the model has R2 = 0.713 and adjusted R2 is 0.702. This means that the suitability level of the model is 70.2% or in other words, 70.2% of the variation of the satisfaction variable (HLC) is explained in general by six observed variables. Thus, the model has the good the explanation level.

ANOVA analysis shows that F parameter has sig. = .000, demonstrating that the model is suitable with the collected data and the variables which are put into the analysis also have the statistical significanceof 5%. Thus, the independent variables within the model have the relation to the dependent variable HLC.

The analyzing result of regression coefficients showed that the sig. value of all independent variables is less than 0.05. Therefore, we can say that all independent variables have effect on customer satisfaction. All these factors have the significance in the model and the positive relationship with the customer satisfaction because the regression coefficients bear positive sign. Details are as follows:

Standardized regression coefficient of tangible facilities variable (PTHH) is β1 = 0.446 which is the highest one in standardized regression coefficients of variables. This also demonstrates that this variable is evaluated importantly compared with the others at this

34

recreation area. Next, standardized regression coefficient of service capacity variable (KNPV) is β2 = 0.382. It ranks second after tangible facilities variable and it is evaluated quite important at this recreation area. The third one is the standardized regression coefficient of image variable (HA) with β3 = 0.339. Next is the standardized regression coefficient of cost variable (CP) with β4 = 0.309. Lastly, they are reliability variable (TCAY) and empathy variable (DCAM) that have the regression coefficients of β5 = 0.297 and β6 = 0.265. All β values that are different from zero have the statistical significance.

Additionally, there are not any factor having β= 0 or β≠0 has no statistical significance.

Simultaneously, they are combined with condition of t>2 and sig< 0.05. We concluded that the factors are not eliminated. In this moment, we make the new regression equation with coefficient β’ as follows:

SHL = β’1*PTHH + β’2*KNPV + β’3*HA +β’4CP * + β’5*TCAY + β’6* DCAM

35

Final regression results for the second time are summarized as follows (See Appendix 5):

Table 4.6: Model Summary and ANOVA

Model R R Square Adjusted R Std. Error of the

Durbin-Watson

Square Estimate

1 .845(a) .713 .702 .54574333 1.769

a Predictors: (Constant), CPHI, DCAM, TCAY, HA, PTHH, KNPV b Dependent Variable: HLC

ANOVA(b)

Model

Sum of

df Mean Square F Sig.

Squares

1 Regression 113,431 6 18,905 63,475 ,000(a)

Residual 45,569 153 ,298

Total 159,000 159

a Predictors: (Constant), CPHI, DCAM, TCAY, HA, PTHH, KNPV b Dependent Variable: HLC

According to the experience of the researcher Nguyen Trong Hoai (University of Economics Ho Chi Minh City), for cross section data, adjusted R Square ranging from 0.20 -0.40 is acceptable; from 0.40 – 0.60 is good; from 0.6 0- 0.80 is very good; and above 0.80 is infrequent.

The adjusted coefficient of determination adjusted R-Square is 0,702¹. It means that the linear regression model that has been built is suitable with data up to 70.2%.

This also shows the relation between the dependent variable and the independent variable is quite close. All of six above variables are to explain 70.2% of the difference of the observed level of customer satisfaction with the service quality of recreation area.

F-type test used in ANOVA is still a hypothesis test regarding the suitability of overall linear regression model. According to the result of Table 4.6, we see that the F-type test has the value of 63.475 with Sig. = .000(a), demonstrating that the multiple linear regression model is suitable with data and can be used.

Now, we check the possibility of the multicollinearity phenomena between independent variables:

VIF <2: The multicollinearity phenomena between independent variables do not affect the model significantly.

2 ≤VIF ≤ 10: The multicollinearity phenomena between independent variables significantly affectthe model.

36

VIF > 10: Sign of multicollinearity

We discover that all of VIF values = 1: Multicollinearity phenomenon among dependent variables does not affect the model significantly².

The model also meets conditions about residual; the residual has approximately normal distribution (Mean =0.00, standard deviation Std.Dev = .98) (see Figure 4.2).

Durbin-Watson statistical quantity (d) = 1.769 which is less than 2, so residual in the model has no correlation with each other (see Table 4.6)

37

Table 4.7: Standardized regression coefficient of equation

Model Unstandardized Standardized t Sig, Collinearity Statistics

Coefficients Coefficients

B Std, Beta Tolerance VIF

Error

1 (Constant) 9,769E-18 ,043 ,000 1,000

KNPV .382 .043 .382 8.823 .000 1.000 1.000

PTHH .446 .043 .446 10.300 .000 1.000 1.000

HA .339 .043 .339 7.832 .000 1.000 1.000

TCAY .297 .043 .297 6.871 .000 1.000 1.000

DCAM .265 .043 .265 6.125 .000 1.000 1.000

CPHI .309 .043 .309 7.131 .000 1.000 1.000

The results show that the β’ coefficients are all different from zero and Sig, <0,05. This proves that all above components participate in customer satisfaction. Comparing value (amplitude) of β’ shows that tangible facilities is the most important matter. It has the greatest influence on customer satisfaction (β’= 0.446). Each (standardized) unit changes at tangible facilities, satisfaction level of customer changes 0.446 unit. Its influence dominates over that of other factors: service capacity (β’= 0.382); image (β’= 0.339); cost (β’= 0.309); reliability (β’=

0.297); empathy (β’=0.265).

From the above results, we write the equation to forecast the customer satisfaction with service quality of Vinpearl land following the independent variables as below:

Standardized regression equation of the model:

SHL = 0.446*PTHH + 0.382*KNPV + 0.339*HA + 0.309*CP + 0.297*TCAY + 0.265*DCAM

Such model could explain 70.2% of the change of HLC variable caused by independent endogenous variables. The remaining 29.8% of variability is explained by exogenousvariables.

The model demonstrates that those independent variables have positive correlationwith level of customer satisfaction with reliability of 95%. Through the regression equation, we discover that when the evaluation scorefor tangible facilities increases by 1, level of customer satisfaction will have the average increase of 0.446; other independent variables are unchanged.

Similarly, when evaluation scorefor service capacity increases by 1, level of customer satisfaction will have an average increase of 0.382; other independent variables are unchanged.

When evaluation scorefor image of recreation area increases by 1, level of customer satisfaction will have an average increase of 0.339; other factors are unchanged. Evaluation scorefor cost increases by 1, satisfaction level will have an average increase of 0.309; other factors are unchanged. Evaluation scorefor reliability level of recreation area increases by 1, satisfaction level of user will have an average increase of 0.297; other factors are unchanged.

38

Finally, evaluation scorefor reliability level increases by 1 mark, satisfaction level of user will have an average increase of 0.265; other factors are unchanged.

With the above analyzing results, we discover that research model is totally suitable, and confirm that there is a close correlation between scales and customer satisfaction with Vinpearl land recreation area.

Table 4.8: Results of hypothesis testing

Hypothesis Testing result

When service capacity of staffs in recreation area is evaluated highly or lowly by customers, their level of satisfaction with that service is also increased or decreased accordingly.

Acceptable H1

H2

When tangible facilities of one service is evaluated highly or lowly by customers, their satisfaction level for that service is also increased or decreased accordingly.

Acceptable

H3 When the image of recreation area is evaluated highly or lowly by customers, their satisfaction level is also increased or decreased accordingly.

Acceptable

H4

When reliability level of customers for one service is evaluated highly or lowly, their satisfaction level is also increased or decreased accordingly.

Acceptable

H5

When the empathy of customers for one service is evaluated highly or lowly, their satisfaction level is also increased or decreased accordingly.

Acceptable

H6 The higher perception of customer about the appropriateness of the service price is, the more higher their satisfaction with service is.

Acceptable

Through the above results, we discover that the hypotheses H1, H2, H3, H4, H5 and H6 are acceptable because the increase of these criteria will increase the customer satisfaction.

In other words, when perception of customers about service quality at this recreation area is increased, customer satisfaction will increased accordingly.