S.E. of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic)
830.0411 Akaike info criterion 3444841. Schwarz criterion -70.61868 Hannan-Quinn criter. 1800.478 Durbin-Watson stat 0.000000
16.58193 16.66959 16.39277 1.999883
LS(W=W2) Y C X1 X2 X3
????Dependent Variable: Y Method: Least Squares Date: 12/21/12 Time: 19:44 Sample: 13 21
Included observations: 9 Weighting series: W2
Variable C X1 X2 X3
R-squared
Coefficient -32944.61 0.131223 0.276088 279.3597
Std. Error 18525.88 0.034694 0.134632 195.2762
t-Statistic -1.778302 3.782248 2.050693 1.430588
Prob. 0.1355 0.0129 0.0956 0.2120
42578.71 10509.73 16.55059 16.63825 16.36143 1.996676
47786.93 23582.57 4353624.
Weighted Statistics
0.999035 Mean dependent var 0.998456 S.D. dependent var 817.1371 Akaike info criterion 3338565. Schwarz criterion -70.47767 Hannan-Quinn criter. 1725.117 Durbin-Watson stat 0.000000
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic)
R-squared Adjusted R-squared S.E. of regression Durbin-Watson stat
Unweighted Statistics
0.999021 Mean dependent var 0.998434 S.D. dependent var 933.1263 Sum squared resid 1.748834
LS(W=W3) Y C X1 X2 X3
????Dependent Variable: Y Method: Least Squares
Date: 12/21/12 Time: 19:45 Sample: 13 21
Included observations: 9 Weighting series: W3
Variable C X1 X2 X3
R-squared
Coefficient -44033.06 0.117363 0.329795 395.1662
Std. Error 18642.29 0.039025 0.147510 199.3679
t-Statistic -2.361999 3.007397 2.235741 1.982095
Prob. 0.0646 0.0298 0.0756 0.1043
47675.04 23275.24 16.71963 16.80729 16.53047 2.045831
47786.93 23582.57 4004549.
Weighted Statistics
0.999107 Mean dependent var 0.998571 S.D. dependent var 889.2037 Akaike info criterion 3953416. Schwarz criterion -71.23834 Hannan-Quinn criter. 1864.539 Durbin-Watson stat 0.000000
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic)
R-squared Adjusted R-squared S.E. of regression Durbin-Watson stat
Unweighted Statistics
0.999100 Mean dependent var 0.998560 S.D. dependent var 894.9356 Sum squared resid 2.057822
LS(W=W4) Y C X1 X2 X3
????Dependent Variable: Y Method: Least Squares Date: 12/21/12 Time: 19:46 Sample: 13 21
Included observations: 9 Weighting series: W4
Variable C X1 X2 X3
Coefficient
-41293.70 0.120675 0.310310 370.7943
Std. Error
19862.84 0.037543 0.142104 213.0039
t-Statistic
-2.078942 3.214344 2.183688 1.740786
Prob.
0.0922 0.0236 0.0807 0.1422
R-squared
Weighted Statistics
46721.17 22102.73 16.76827 16.85592 16.57911 2.127479 47786.93 23582.57 4437614.
0.998900 Mean dependent var 0.998240 S.D. dependent var 911.0924 Akaike info criterion 4150447. Schwarz criterion -71.45721 Hannan-Quinn criter. 1513.526 Durbin-Watson stat 0.000000
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic)
R-squared Adjusted R-squared S.E. of regression Durbin-Watson stat
Unweighted Statistics
0.999003 Mean dependent var 0.998404 S.D. dependent var 942.0843 Sum squared resid 1.856697
LS(W=W5) Y C X1 X2 X3
????Dependent Variable: Y Method: Least Squares Date: 12/21/12 Time: 19:46 Sample: 13 21
Included observations: 9 Weighting series: W5
Variable C X1 X2 X3
R-squared
Coefficient -39204.90 0.126106 0.283912 350.0404
Std. Error 19240.41 0.033338 0.127855 205.5793
t-Statistic -2.037633 3.782712 2.220580 1.702702
Prob. 0.0972 0.0129 0.0771 0.1494
44488.70 24145.09 16.61973 16.70739 16.43057 2.360020
Weighted Statistics
0.998886 Mean dependent var 0.998218 S.D. dependent var 845.8792 Akaike info criterion 3577558. Schwarz criterion -70.78880 Hannan-Quinn criter. 1495.034 Durbin-Watson stat 0.000000
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic)
R-squared
Unweighted Statistics
47786.93 23582.57 5339686.
0.998800 Mean dependent var 0.998080 S.D. dependent var 1033.411 Sum squared resid 1.569813
Adjusted R-squared S.E. of regression Durbin-Watson stat
经估计检验发现用权数W3的效果最好。下面仅给出用权数W3的结果。
③ 对所估计的模型再进行White检验，观察异方差的调整情况
对所估计的模型再进行White检验，其结果对应图所示。所对应的White检验显示，P值较大，所以接受不存在异方差的原假设，即认为已经消除了回归模型的异方差性。
2. 自相关检验 （1）图示法
由图可知，残差的图形成循环型的，因此存在正相关。
（2）杜宾—瓦森检验法
D.W.检验结果表明，在5%的显著性水平下，n=15，k=4查表得dl?0.82，du?1.75，由于4-dl=3.18>D.W=2.893273>4?du?2.25,所以无法判断是否存在自相关。
（3）拉格朗日乘数检验法
在方程窗口上点击View\\Residual Test\\ serial correlation LM test,点击后就会出现对话框），在空处填写1或2或3等，数字代表着几阶自相关。如此循环去检验。无论是一阶还是2阶或是4阶，该模型的P值都很小，说明该模型都存在自相关。下图是2阶时的数据模型。
Breusch-Godfrey Serial Correlation LM Test: F-statistic
3.807478 Probability 7.071165 Probability
Std. Error
4035.894 0.016631 0.070103 37.47232 0.229153 0.245028
t-Statistic
-0.403873 -0.321152 0.330394 0.458555 2.006376 -2.349045
0.045996 0.029142
Prob.
0.6920 0.7525 0.7457 0.6531 0.0632 0.0329
-3.71E-12 996.5221 16.75850 17.05693 1.522991 0.241391
Obs*R-squared
Test Equation:
Dependent Variable: RESID Method: Least Squares Date: 12/15/12 Time: 05:34
Presample missing value lagged residuals set to zero.
Variable C X1 X2 X3 RESID(-1) RESID(-2) R-squared
Coefficient
-1629.987 -0.005341 0.023162 17.18314 0.459768 -0.575582
0.336722 Mean dependent var 0.115630 S.D. dependent var 937.1391 Akaike info criterion 13173445 Schwarz criterion -169.9642 F-statistic 1.729152 Prob(F-statistic)
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
(4) 自相关的修正