Re: st: weighted regression --- *PLEASE IGNORE THE POST*

["Nikolaos A. Patsopoulos" <npatsop@cc.uoi.gr>]

Nikolaos A. Patsopoulos wrote:
> Dear statalist members,
>
> in a weighted regression with what do I have to divide the weighted SE 
> to get the unweighted one?
>
> Some sample data and results:
> WEIGHTED
> var1    var2    weight
> 0    3.4    1
> 1    4.7    2
> 2    9.2    2
>
> regress  var2 var1 [aweight= weight]
>
> (sum of wgt is   5.0000e+00)
>
>      Source |       SS       df       MS              Number of obs 
> =       3
> -------------+------------------------------           F(  1,     1) 
> =    9.37
>       Model |  16.4437701     1  16.4437701           Prob > F      =  
> 0.2010
>    Residual |  1.75542889     1  1.75542889           R-squared     =  
> 0.9035
> -------------+------------------------------           Adj R-squared 
> =  0.8071
>       Total |   18.199199     2  9.09959951           Root MSE      =  
> 1.3249
>
> ------------------------------------------------------------------------------ 
>        var2 |      Coef.   Std. Err.      t    P>|t|     [95% Conf. 
> Interval]
> -------------+---------------------------------------------------------------- 
>        var1 |   3.128571   1.022203     3.06   0.201    -9.859744    
> 16.11689
>       _cons |   2.485714   1.445613     1.72   0.335    -15.88254    
> 20.85397
> ------------------------------------------------------------------------------ 
>
> UNWEIGHTED
> var1    var2
> 0    3.4
> 1    4.7
> 2    9.2
> 1    4.7
> 2    9.2
>
> . regress  var2 var1
>
>      Source |       SS       df       MS              Number of obs 
> =       5
> -------------+------------------------------           F(  1,     3) 
> =   28.10
>       Model |  27.4062836     1  27.4062836           Prob > F      =  
> 0.0131
>    Residual |  2.92571481     3   .97523827           R-squared     =  
> 0.9035
> -------------+------------------------------           Adj R-squared 
> =  0.8714
>       Total |  30.3319984     4  7.58299959           Root MSE      =  
> .98754
>
> ------------------------------------------------------------------------------ 
>        var2 |      Coef.   Std. Err.      t    P>|t|     [95% Conf. 
> Interval]
> -------------+---------------------------------------------------------------- 
>        var1 |   3.128571   .5901689     5.30   0.013      1.25039    
> 5.006752
>       _cons |   2.485714   .8346249     2.98   0.059    -.1704347    
> 5.141863
> ------------------------------------------------------------------------------ 
>
> I believe I have to divide the weight SE (1.022203) with the 
> ResidualMS (1.75542889), which gives 0.582309.
>
>
> Is this correct?
>
> Thanks in advance,
>
> Nikos
>
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
Sorry for this, wrong weight used ( aweight instead of the correct 
fweight). STATA as it seems does correct automatically weighted SEs (not 
like SPSS).

Again sorry for the apparent spam.

Nikos

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/