3.4. Analysis of the impact of factors affecting overall satisfaction
Impact analysis is the author considers the individual impact of each measured variable and their simultaneous impact on overall satisfaction. That is, we conduct univariate linear regression analysis and multiple linear regression.
3.4.1. Univariate linear regression analysis
After analyzing the collected data through the Cronbach alpha reliability analysis step, the research model remains the same as Figure 1.5 proposed by the author in Chapter 1, including 3 factors of service quality, corporate image and service price. In which, the service quality factor includes 5 factors: Reliability, response, tangibility, assurance, empathy.
In which the hypotheses proposed to be tested are:
- H1: The higher the satisfaction with service quality, the more satisfied the customer is.
- H2: The higher the satisfaction with the image, the more satisfied the customer is.
- H3: The higher the satisfaction with service price, the more satisfied the customer is.
The author conducts a univariate linear regression analysis between an independent variable and a dependent variable to test the hypothesis and find the relationship between the variables as a basis for building a multiple linear regression model.
Simple linear regression is where we look at the relationship between two quantitative variables. When analyzing simple linear regression, we assume that only one variable affects customer satisfaction, other factors do not affect customer satisfaction.
Hypothesis testing can be based on many bases. However, in the scope of the article, the author relies on the main basis: Based on the suitability of the simple linear regression equation and the significance of the simple linear regression equation to evaluate more specifically the impact of an independent variable on the dependent variable.
Table 3.14: Results of simple linear regression Model Summary(b)
Model
R | R Square | Adjusted R Square | Std. Error of the Estimate | Change Statistics | Durbin-Watson |
Maybe you are interested!
-
Analysis of Factors Affecting the Amount of Savings Deposits of Individual Customers at Abbank Cai Rang Branch -
Factors affecting the choice of bank for deposit transactions of individual customers - 1 -
Results of Linear Regression Analysis on Factors Affecting Land Complaints in Vinh City -
Exploratory Factor Analysis (EFA) Results for Factors Affecting Customer Satisfaction -
Summary of Factors Affecting the Debt Repayment Ability of Individual Customers at Vsb Dong Nai from Experimental Model
R Square Change | F Change | df1 | df2 | Sig. F Change | ||||||
CLDV | .902(a) | 0.814 | 0.813 | 0.198 | 0.814 | 775,977 | 1 | 177 | 0.000 | 1,808 |
HADN | .516(a) | 0.266 | 0.262 | 0.394 | 0.266 | 64,153 | 1 | 177 | 0.000 | 1,724 |
GDV | .522(a) | 0.273 | 0.269 | 0.392 | 0.273 | 66,422 | 1 | 177 | 0.000 | 1,843 |
Coefficients(a)
Model
Unstandardized Coefficients | Standardized Coefficients | t | Sig. | |||
B | Std. Error | Beta | ||||
CLDV | (Constant) | 0.826 | 0.113 | 7,292 | 0 | |
CLDV | 0.843 | 0.030 | 0.902 | 27,856 | 0 | |
HADN | (Constant) | 2,781 | 0.150 | 18,568 | 0 | |
HADN | 0.293 | 0.037 | 0.516 | 8.010 | 0 | |
GCDV | (Constant) | 3,052 | 0.115 | 26,582 | 0 | |
GCDV | 0.238 | 0.029 | 0.522 | 8,150 | 0 | |
From the results of single-variable regression in appendices 4, 5, 6 and summary table 3.14, we successively build a single-variable linear regression model.
- Univariate linear regression model
The generalized simple linear regression model is built from sample data in the form:
Ŷ i = B 0 + B 1 * X i.
In there:
X i : Is the i-th observation value of the independent variable
Ŷ i : Is the predicted value (or theoretical value) of the dependent variable i, the caret represents the predicted value.
B 0 and B 1 : Are the regression coefficients mentioned above, the method used to determine B 0 and B 1 is the ordinary least squares method (OLS).
From the general regression model, we will build a single-variable linear regression equation describing the relationship between CLDV - "Service quality" as the independent variable and SHL - "Satisfaction" (1), HADN - "Business image" as the independent variable and SHL (2), GDV - "Service price" as the independent variable and SHL (3). The equations of the straight lines have the form:
1) SHL cl = B 0cl + B 1cl * CLDV
2) SHL ha = B 0ha + B 1ha * HADN
3) SHL gd = B 0gc + B 1gc * GDV
Table 3.14 - Coefficients provides us with information about the regression coefficients estimated by the OLS method, the slope and the constant are shown in column B of the results table. We can write the corresponding univariate linear regression equation as follows:
1) SHL cl = 0.826 + 0.843 * CLDV
2) SHL ha = 2.781 + 0.293 * HADN
3) SHL gd = 3.052 + 0.238 * GDV
Once the regression equation is determined, we need to proceed with the steps of evaluating the suitability of the univariate linear regression model.
- Assessment of the suitability of the univariate linear regression model
The commonly used measures to evaluate the goodness of fit of a simple linear regression model are the coefficient of determination R 2 and the adjusted R 2. The adjusted R 2 is often used to more closely evaluate the goodness of fit of the model. The higher the adjusted R 2 and the closer it is to 1, the better the goodness of fit of the model. In Table 3.14 - Model Summary we see:
Firstly, the univariate linear regression model between CLDV and SHL has an adjusted R2 value = 0.813, meaning that 81.3% of the variation in customer satisfaction with lending services is explained by the service quality variable or the model fit is 81.3%. Thus, the level of explanation of the CLDV variable for the SHL variable is very high.
At the same time, we also see that the Sig. F = 0.000 value is very small, so the simple linear regression model is suitable to explain the relationship between the variables CLDV and SHL.
Second, the univariate linear regression model between HADN and SHL has an R 2 value
adjusted = 0.262, meaning that 26.2% of the variation in customer satisfaction is explained by the corporate image variable. This shows that the suitability of the single linear regression model between the independent variable HADN and the dependent variable SHL is 26.2%. We see that Sig.F is very small, so the single linear regression model is suitable to explain the relationship between the variables HADN and SHL.
Third, the univariate linear regression model between GDV and SHL has an effective R 2 value.
= 0.269, meaning that 26.9% of the variation in customer satisfaction is explained by the GDV variable. This shows that the suitability of the single linear regression model between the independent variable GDV and the dependent variable SHL is 26.9%. We also see that the Sig. F value is small, so the single linear regression model is suitable to explain the relationship between GDV and SHL.
Thus, the level of explanation of the service quality variable for the customer satisfaction variable is the highest.
- Explain the equation
All three univariate linear regression equations were fit and significant.
The first equation of simple linear regression between CLDV - SHL:
SHL cl = 0.826 + 0.843 * CLDV
We see that when satisfaction with service quality increases, customer satisfaction increases because the coefficient B 0cl = 0.826 > 0 and B 1cl = 0.843 > 0, the relationship between service quality and satisfaction is a positive linear relationship. Therefore, hypothesis H1: The higher the satisfaction with service quality, the more satisfied the customer is accepted.
The second equation of simple linear regression between HADN - SHL:
SHL ha = 2.781 + 0.293 * HADN
We see that when HADN satisfaction increases, customer SHL increases due to coefficient B 0ha
= 2.781 > 0 and B 1ha = 0.293 > 0 so the relationship between HADN and SHL is linear.
positive. Therefore, hypothesis H2: The higher the satisfaction with the image, the more satisfied the customer is is accepted.
The third equation of simple linear regression between GDV - SHL:
SHL gd = 3.052 + 0.238 * GDV
We see that when satisfaction with service price increases, customer satisfaction increases because the coefficient B 0gd = 3.052 > 0 and B 1gd = 0.238 > 0, so the relationship between service price and satisfaction is a positive linear relationship. Therefore, hypothesis H3: The higher the satisfaction with service price, the more satisfied the customer is accepted.
- Conclude
Univariate linear regression analysis shows that service quality is the best explanatory variable for customer satisfaction. All three variables of service quality, corporate image and satisfaction have a positive linear relationship with the satisfaction variable, which is the basis for the author to build a multiple linear regression model. At the same time, the following hypotheses have been tested:
Table 3.15: Hypothesis testing results
Hypothesis
Test results | |
- H1: The higher the satisfaction with service quality, the more satisfied the customer is. | Accept |
- H2: The higher the satisfaction with the image, the more satisfied the customer is. | Accept |
- H3: The higher the satisfaction with service price, the more satisfied the customer is. | Accept |
3.4.2. Multiple linear regression analysis
In fact, a dependent variable is often affected by two or more independent variables. The multiple linear regression model, also known as the multivariate linear regression model, is one of the statistical models commonly used in testing scientific theories (testing research models). When using the multiple linear regression model, we need to pay attention to the suitability of the model and check the assumptions.
Multiple regression analysis is used to study customer satisfaction to determine the causal relationship between the dependent variable: Customer satisfaction
- SHL and independent variables: "Service quality" - CLDV; "Business image" - HADN; "Service price" - GDV. The purpose of multiple regression analysis is to predict the value of the dependent variable when the values of the independent variables are known in advance, through which we can consider the impact of the independent variables on the dependent variable. The chosen analysis method is the Steepwise method, a popular and suitable method to explore the relationship between variables. The analysis results are shown specifically in Table 3.16.
Table 3.16: Results of multiple linear regression analysis
Model Summary(d)
Mode
l
R | R Square | Adjusted R Square | Std. Error of the Estimate | Change Statistics | Durbin- Watson | |||||
R Square Change | F Change | df1 | df2 | Sig. F Change | ||||||
1 | .902(a) | .814 | .813 | .19831 | .814 | 775,977 | 1 | 177 | .000 | |
2 | .919(b) | .845 | .843 | .18189 | .030 | 34,400 | 1 | 176 | .000 | |
3 | .954(c) | .909 | .908 | .13928 | .065 | 125,151 | 1 | 175 | .000 | 1,932 |
ANOVA(d)
Model
Sum of Squares | df | Mean Square | F | Sig. | ||
1 | Regression | 30,516 | 1 | 30,516 | 775,977 | .000(a) |
Residual | 6,961 | 177 | .039 | |||
Total | 37,477 | 178 | ||||
2 | Regression | 31,654 | 2 | 15,827 | 478,402 | .000(b) |
Residual | 5,823 | 176 | .033 | |||
Total | 37,477 | 178 | ||||
3 | Regression | 34,082 | 3 | 11,361 | 585,629 | .000(c) |
Residual | 3,395 | 175 | .019 | |||
Total | 37,477 | 178 |
Coefficients(a)
Model
Unstandardized Coefficients | Standardized Coefficients | t | Sig. | Collinearity Statistics | ||||
B | Std. Error | Beta | Tolerance | VIF | ||||
1 | (Constant) | .826 | .113 | 7,292 | .000 | |||
CLDV | .843 | .030 | .902 | 27,856 | .000 | 1,000 | 1,000 | |
2 | (Constant) | .653 | .108 | 6,046 | .000 | |||
CLDV | .773 | .030 | .828 | 25,602 | .000 | .845 | 1,184 | |
HADN | .108 | .018 | .190 | 5,865 | .000 | .845 | 1,184 | |
3 | (Constant) | .482 | .084 | 5,732 | .000 | |||
CLDV | .703 | .024 | .753 | 29,364 | .000 | .787 | 1,270 | |
HADN | .100 | .014 | .175 | 7,072 | .000 | .842 | 1,187 | |
GCDV | .122 | .011 | .267 | 11,187 | .000 | .907 | 1.102 | |
Excluded Variables(c)
Model
Beta Print | t | Sig. | Partial Correlation | Collinearity Statistics | ||||
Tolerance | VIF | Minimum Tolerance | ||||||
1 | HADN | .190(a) | 5,865 | .000 | .404 | .845 | 1,184 | .845 |
GCDV | .276(a) | 10,230 | .000 | .611 | .910 | 1,099 | .910 | |
2 | HADN | |||||||
GCDV | .267(b) | 11,187 | .000 | .646 | .907 | 1.102 | .787 | |
- Multiple linear regression model
The multiple linear regression model is an extension of the single linear regression model by adding independent variables to better explain the dependent variable. In the content of the article, the author examines the factors affecting customer satisfaction with the bank's lending services. The factors proposed in the model are: CLDV - service quality, HADN - corporate image, GDV - service price. Therefore, the proposed multiple linear regression model has the following form:
SHL = β 0 + β 1 * CLDV + β 2 * HADN + β 3 * GDV
Where β are called regression coefficients.
Table 3.16 - Coefficients gives us information about the regression coefficients estimated by the OLS method, the slope and the constant are shown in column B of the results table. We can write the multiple linear regression equation:
SHL = 0.482 + 0.703 * CLDV + 0.1 * HADN + 0.122 * GDV
Once the regression equation is determined, we need to conduct steps to evaluate and test the model's suitability to see whether the model from the sample estimate is suitable to apply to the population or not. In testing the suitability of the multiple linear regression model, we need to perform some additional tests compared to the single linear regression model.
- Assessment of the suitability of the multiple linear regression model
To evaluate the suitability of a multiple linear regression model similar to a single linear regression model, we need to determine the coefficient R 2 or the coefficient R 2 adjusted (Adjusted R Square). Almost no straight line can fit the data set perfectly, there is always a deviation between the predicted values given by the straight line and the actual values shown through the residuals. R 2 = 1 shows that the multiple linear regression model fits 100% of the sample data set. In table 3.16 - Model summary, the value R 2 = 0.909. Because we cannot get the R 2 of the crowd because there is no crowd data, we must use the coefficient of determination R 2 adjusted (Adjusted R Square) from R 2 (R square) to estimate for the crowd. Therefore, R 2 adjusted from R 2 is used to more closely reflect the suitability of the multiple linear regression model. Comparing adjusted R 2 and R 2 , adjusted R 2 is smaller (Adjusted R Square = 0.908) while R square = 0.909, when using it to evaluate the suitability of the model it is safer because it does not inflate the suitability of the model. Thus, the adjusted R 2 of the model is 0.908, showing that the suitability of the multiple linear regression model with the observed variable is very high (90.8% of the variation in customer satisfaction with service quality is explained by the linear relationship of the independent variables). This means that the multiple linear regression model is suitable to explain the relationship between individual customer satisfaction with the lending services of GP. Bank, Vung Tau branch. We also see that the adjusted R 2 of the multiple linear regression model is higher than the three simple linear regression models analyzed above. This shows that the regression model