Re: st: mean tests an fweight

[Steven Samuels <sjhsamuels@earthlink.net>]

You should -svyset- you data to properly account for the sampling  
design. Information about the probability weights (pweights, not  
fweights) to use, the strata, if any, and "primary sampling units"  
will be in the survey documentation. Be warned that the options for - 
svyset- may not be simple.  If you have successfully -svyset- your  
data, suppose "gender" is your group variable

The command: xi: svy: reg income  i.gender

will do the survey equivalent of a t-test.

If the survey concepts in my first sentence are unfamiliar, find a  
good book a survey sampling (first choice: Sharon Lohr, 1999,   
Sampling: Design and Analysis.  Duxbury Press; second:   Lemeshow &  
Levy: Sampling of Populations, 3rd Edition, Wiley, which has Stata  
examples. I also recommend the Stata survey pages at UCLA:   http:// 
www.ats.ucla.edu/stat/stata/topics/Survey.htm

-Steve



On Oct 5, 2008, at 7:49 PM, Leonor Saravia wrote:


> Hello Statalisters,
>
> I�m comparing the mean of a sample by group, this is, comparing the
> mean of the income earned for males and females. I�m using the Stata -
> ttest - command, but i�m a bit confused because this command doesn�t
> allows using weights (fweights) and I think that this affects the
> results, doesn�t it?
> Do you know another way to compare the means of the sample using  
> weights?
>
> I�d really appreciate your comments.
>
> Thank you,
>
> Leonor
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/