eviews 回归分析 正态分布

eviews 回归分析 正态分布
第一次做回归分析的时候结果是这样.
Dependent Variable:L
Method:Least Squares
Date:11/26/10 Time:12:31
Sample:1975 2009
Included observations:35
Variable Coefficient Std.Error t-Statistic Prob.
C -139.1324 43.75109 -3.180090 0.0033
K 0.000363 0.000112 3.245454 0.0028
T 2.44E-06 4.68E-07 5.219729 0.0000
N -4.46E-05 1.42E-05 -3.147523 0.0036
R-squared 0.886214 Mean dependent var 26.11714
Adjusted R-squared 0.875202 S.D.dependent var 15.59284
S.E.of regression 5.508436 Akaike info criterion 6.357649
Sum squared resid 940.6288 Schwarz criterion 6.535403
Log likelihood -107.2589 F-statistic 80.48043
Durbin-Watson stat 0.804796 Prob(F-statistic) 0.000000
我对上边的结果做了对数.结果是这样.
Dependent Variable:L
Method:Least Squares
Date:11/26/10 Time:16:36
Sample (adjusted):1976 2009
Included observations:34 after adjustments
Variable Coefficient Std.Error t-Statistic Prob.
K -0.009192 0.038484 -0.238842 0.8129
T 0.056958 0.030722 1.853986 0.0736
N -0.058760 0.017315 -3.393607 0.0020
L(-1) 0.985595 0.009576 102.9252 0.0000
R-squared 0.999754 Mean dependent var 3.120385
Adjusted R-squared 0.999730 S.D.dependent var 0.578109
S.E.of regression 0.009503 Akaike info criterion -6.364222
Sum squared resid 0.002709 Schwarz criterion -6.184650
Log likelihood 112.1918 Durbin-Watson stat 1.776801
我觉得模型稳定了一些,但是别的检验都没事就有正态分布检验的时候违背假设.
这下怎么办.这个模型能用上吗?还我应该改别的吗?

一般用SPSS、EVIEWS来检验.
最简单的方法就是通过画正态分布图来判断,或者Q-Q图,也可以通过用非参数检验中的单样本K-S进行检验