42
c. Autocorrelation Test
Test the possibility of autocorrelation aims to determine whether the error bullies at certain periods correlated with the error
bullies at other periods. Autocorrelation in the concept of linear regression mean error component correlated in order of time the
time-series data or the sequence space on cross-sectional data. In this study, the test follows the autocorrelation, with the Durbin-
Watson test statistics. Table 3.1
Classification value of D Value of D
Information
0 d d
L
There is the autocorrelation d
L
d d
U
non-conclusion 4 - d
L
d 4 non-autocorrelation
4 - d
L
d 4 - d
U
There is the autocorrelation d
U
d 4 d
U
non-conclusion
d. Heteroscedasticity Test
Heteroscedasticity means there is a variant that is not the same for different independent variables. It can be detected by
observing how the points on the scatterplot between the estimated values of Y with residual value the difference between the actual
dependent variable predictive values versus the value of the prediction spread or not to form a pattern. If the graph has a
standardized residual axis of the X and Y axis that has been predicted not to form a clear pattern wavy, widened, then
43 narrowed, as well as scattered both above and below the 0 on the Y
axis Heteroscedasticity it did not happen.
Some ways to detect the presence or absence of heteroscedasticity:
1 If there is a specific pattern, such that there are points that form
a regular pattern wavy, widened and then narrowed, it has been indicated heteroscedasticity.
2 If there is no clear pattern, and the points spread above and
below the 0 on the Y axis, then it does not happen heteroscedasticity.
2. Hypothesis Testing
The analysis tools are the simple and the multiple regression. Both are used in order to see the effects of income on government
expenditures. Simple regression is used to see the influence of each variable to the prediction of expenditures separately, whereas multiple
regression is used to see the influence of all these variables
simultaneously. a.
Multiple Regression
To test the hypothesis Ha, the method of analysis used is multiple regression, because it involves three independent variables
and one dependent variable. Regression equation model to test the hypothesis with the following formulation:
Yi = α + b1 DAU1i + b2 PAD2i + e