对此回归模型通过残差图及 Spearman 相关系数检验可知无异方差性

This regression model by residual plots and Spearman correlation coefficient test shows no different variance exists. Autocorrelation of the inspection by way of illustration we first test method, and found that most of the points fall II, IV quadrant, then we do exact test with DW test, to give DW = 2.402, falls (2.4) between these show random disturbance Ɛi there is a negative correlation. Regression model (1), the variance of the available scale factor (the VIF) to determine the presence of multiple co-linear. Calculated from the statistical software SPSS:

This regression model can be seen to exist by the residual graph and Spearman correlation coefficient test. The test of self-correlation first through the graphical test, found that most of the points fall in the II, IV quadrant, and then we use the DW test to do the exact test, get DW s 2.402, fall between (2.4), which indicates that there is a negative correlation of the random perturbation item i. For regression model (1), the presence of multiple collinearity can be judged by the variance expansion factor (VIF). Calculated by the statistical software SPSS:

In this regression model, we can see that there is no heteroscedasticity through residual graph and Spearman correlation coefficient test. For the test of autocorrelation, we first find that most of the points fall in quadrants I I and IV through the graph test, and then we use DW test to do the accurate test, and we get DW = 2.402, falling between (2.4). All these indicate that there is a negative correlation in the random disturbance term Ɛ I. For regression model (1), the existence of multicollinearity can be judged by the variance expansion factor (VIF). According to the statistical software SPSS:<br>