Re: st: Is it correct to use frequency weights?

[Rafael Claro <rafaelcla@uol.com.br>]

Dear Nicolás,

According to your description sounds like you have sampling weights and not frequency weights (check for pweight in help file for weights). As you also mentioned having other stages in your sampling plan the “frequency weight” possibility is even more improbable. This is, probably, the reason why you are facing unexpected results.

Check the help file for “survey” and also “svyset” for further information. You should not have a hard time to get everything working.

Best regards,
Rafael Moreira Claro



Em 04/05/2010 16:14, Nicolás Rojas < nicolas.srojas@gmail.com > escreveu:
Hi everyone,
                         I am using a crime survey that apparently has
frequency weights. The instruction manual says that the variable
"weights"  describes how many people each observation represent in the
population. That coincides with the  description of frrequency
weigths.
                      However, I am realizing that this is a two step
survey and therefore other variables (namely, strata variables) also
included in the survey play an important role. That would explain why
using frequency weights give results that are very bizzare ( t test of
90, evrything with three stars etc).
Should I use stata commands for complex survey or should I use another
replication method for computing the standard errors?
Does anyone knows literature specific about that subject?

Thank you very much,


Nicolás Rojas
Ponthifical Catholic University of Chile

*
*   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/