Hypothesis Testing About the Significance of Regression Coefficients

With significance level α = 5%

Testing the hypothesis on the significance of the regression coefficients, in Table 4.15, the t values ​​correspond to sig < 0.05. Therefore, reject the hypothesis H 0 and conclude that the independent variables (1) Need for information on management accounting from the unit's leaders; (2) Awareness of the unit's leaders; (3) Organizational structure of the unit's management apparatus; (4) Methods and techniques; (5) Accounting human resources; (6) Application of IT and (7) Level of market competition affect the Application of management accounting at Binh Duong Water - Environment Joint Stock Company .

4.4.2 Testing for multicollinearity

There are many ways to detect multicollinearity such as: large R2 coefficient but small t, high pair correlation of explanatory variables, auxiliary regression, using variance inflation factor - VIF (Hoang Ngoc Nham et al., 2008). Here, the author chooses to use VIF coefficient, if VIF > 10 then multicollinearity may occur (Nguyen Dinh Tho, 2011). The results show that the VIF coefficients of the variables are all within the allowable range (NCTT, NTLD, BMQL, PPKT, NLKT, IT, MDCT coefficients are 1.078; 1.184; 1.363; 1.156; 1.264, 1.185, 1.236 respectively, showing that the model is not multicollinear), and the acceptability of the variable is greater than 0.1, meaning that multicollinearity does not occur.

Table 4.16: Multicollinearity test

Factor

MDCT

(Source: Author's synthesis from data analysis results)

4.4.3. Autocorrelation test

The Durbin-Watson test is performed to test the assumption of independence of errors (no autocorrelation). If the residuals do not have first-order serial correlation with each other, the value of d is greater than 1. The value of d = 1.082 is within the acceptable range, meaning there is no first-order serial correlation or in other words, there is no correlation between the residuals (Hoang Trong and Chu Nguyen Mong Ngoc, 2008).

Table 4.17: Durbin-Watson run results

Model

(Source: Author's synthesis from data analysis results)

4.4.4. Testing for normal distribution of residuals

The residuals may not follow a normal distribution for reasons such as: using the wrong model, non-constant variance, the number of residuals is not large enough for analysis,... Therefore, the author decided to examine the distribution of the residuals by constructing a histogram of the residual frequencies.

Figure 4.2: Histogram of residuals – normalized

(Source: Author's synthesis from data analysis results)

The results in the Histogram are bell-shaped and the standard deviation value Std.Dev = 0.982 (close to 1) and Mean = 0. Thus, it can be concluded that the distribution of the residuals is approximately normal.

4.5. Checking regression model assumptions

In order to draw conclusions from a regression model, it is necessary to test the assumptions.

4.5.1 Testing the assumption of constant variance of errors (residuals)

Check the Scatter plot for the standardized residuals and standardized predicted values. The results show that the residuals are randomly scattered around the line through zero, and do not form any particular shape.

Figure 4.3: Scatter plot between predicted values ​​and residuals from regression

(Source: Author's synthesis from data analysis results)

Looking at the graph, we see that there is no relationship between the predicted values ​​and the residuals, they are randomly distributed, without any rule. Therefore, the assumption of linear relationship and equal variance is satisfied, so we conclude that the above linear regression model can be used.

4.5.2. Check the assumption of normal distribution of residuals

Residuals may not follow a normal distribution for reasons such as using the wrong model, the variance is not constant, the number of residuals is not large enough for analysis... (Hoang Trong - Mong Ngoc, 2008). The frequency chart (Histogram, QQ plot, PP plot) of the residuals (standardized) is used to test this assumption.

Figure 4.4: PP Plot of residuals – standardized

(Source: Author's synthesis from data analysis results)

The results from the PP plot show that the points are scattered around the expected. This also indicates that the assumption of normal distribution of the residuals is not violated.

4.6 Discussion of research results

4.6.1. Scale of influencing factors

According to the results of the research model testing above, the application of international accounting at Binh Duong Water - Environment Joint Stock Company is affected by 7 factors with 29 observed variables (also known as measurement scales) including: (1) Demand for international accounting information from the unit's leaders; (2) Awareness of the unit's leaders; (3) Organizational structure of the unit's management apparatus; (4) Methods and techniques; (5) Human resources.

accounting capacity; (6) IT application and (7) Market competitiveness. The scales of each factor are shown as follows:

Table 4.18: Scales of factors affecting the application of management accounting at Binh Duong Water - Environment Joint Stock Company

STT