Catherine Watt <[email protected]> asks:
> I am using svylogit followed by adjust, using pweights. When I try to set
> certain factors to the grand mean when using adjust, stata seems to be
> forgetting about the pweights.
>
> <cut>
>
> I tried the following commands:
>
> logit y factor1 factor2 [pweight=w]
> adjust factor1, by(factor2)
>
> The adjust results tell me that the average if factor 1 is 0.5, which it
> clearly isn't if one uses pweights (I calculate that it is 0.92)
>
> Am I missing something? Perhaps for my "real" data I need to go through by
> hand to calculate the average value of each factor, and specify it myself.
-adjust- does not examine the weight from the estimation command
in setting factor1 (in your example) to its mean. If you want to
have -adjust- produce predictions with factor1 set to its
weighted mean, you will have to use syntax such as
. adjust factor1=.92, by(factor2)
You can use -svymean factor1- (after you have -svyset- your data)
to obtain the pweighted mean of factor1 that you then supply to
-adjust- (.92 in your example). If you need to do this often,
then you might want to write a wrapper ado-file that automates
these steps. If it were some other kind of weight (fweights,
iweights, or aweights), you could use -summarize- instead of
-svymean- to obtain the weighted mean.
Ken Higbee [email protected]
StataCorp 1-800-STATAPC
*
* 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/
