做异方差问题时出了一个问题，想请教一下大侠，暂时没有积分，有详细解答的之后一定补分，谢谢啦~~

这是是原始数据做出的最小平方估计
Dependent Variable: Y   
Method: Least Squares   
Date: 11/02/10   Time: 14:46   
Sample: 1 21   
Included observations: 21   
   
Variable Coefficient Std. Error t-Statistic Prob.  
   
C -562.9074 291.5642 -1.930646 0.0686
X 5.372828 0.644239 8.339811 0.0000
   
R-squared 0.785438     Mean dependent var  1588.143
Adjusted R-squared 0.774145     S.D. dependent var  1310.975
S.E. of regression 623.0301     Akaike info criterion  15.79746
Sum squared resid 7375164.     Schwarz criterion  15.89694
Log likelihood -163.8733     F-statistic  69.55245
Durbin-Watson stat 0.430291     Prob(F-statistic)  0.000000
   

二是给每个解释变量，被解释变量，常数项分别加权得到的广义最小平平方估计
Dependent Variable: YY   
Method: Least Squares   
Date: 11/02/10   Time: 15:06   
Sample: 1 21   
Included observations: 21   
   
Variable Coefficient Std. Error t-Statistic Prob.  
   
XX0 368.6203 84.16826 4.379564 0.0003
XX 2.952837 0.822684 3.589272 0.0020
   
R-squared 0.938666     Mean dependent var  0.019657
Adjusted R-squared 0.935438     S.D. dependent var  0.026408
S.E. of regression 0.006710     Akaike info criterion  -7.080059
Sum squared resid 0.000855     Schwarz criterion  -6.980581
Log likelihood 76.34062     Durbin-Watson stat  1.706018
   


三是直接做ls(w-1/x^2)y c x得到的广义最小平方估计
Dependent Variable: Y   
Method: Least Squares   
Date: 11/02/10   Time: 15:08   
Sample: 1 21   
Included observations: 21   
Weighting series: 1/X^2   
   
Variable Coefficient Std. Error t-Statistic Prob.  
   
C 368.6203 84.16826 4.379564 0.0003
X 2.952837 0.822684 3.589272 0.0020
   
Weighted Statistics   
   
R-squared 0.938666     Mean dependent var  808.6869
Adjusted R-squared 0.935438     S.D. dependent var  1086.417
S.E. of regression 276.0477     Akaike info criterion  14.16942
Sum squared resid 1447845.     Schwarz criterion  14.26890
Log likelihood -146.7789     F-statistic  12.88288
Durbin-Watson stat 1.706018     Prob(F-statistic)  0.001956
   
Unweighted Statistics   
   
R-squared 0.625243     Mean dependent var  1588.143
Adjusted R-squared 0.605519     S.D. dependent var  1310.975
S.E. of regression 823.3933     Sum squared resid  12881555
Durbin-Watson stat 0.380812   
   
请教一下，为什么二和三中，只有r2和调整r方几个值相同，剩下的数都不相同，三中表示的没有加权的部分与原始最小平方估计的结果也是不同的？非常感谢~