--- On Tue, 8/12/09, P C <[email protected]> wrote:
> I have a survey data set. In this variable I have a
> race/ethnicity variable, and a dependent variable (obesity
> which is a dichotomous variable: obese=1; not obese=0). I
> have two waves of data for obesity.
>
> At wave 1, I calculate percentage of obese respondents for
> each race/ethnic group. Then at Wave 2, I do the same thing.
> Now I want to know the change for the percentage of obese
> respondents for each racial group is significant or not. I
> also need to know the confidence interval of change.
These proportions are just means, so you can use -regress- of
obesity on wave with the -svy- prefix and the appropriate
-subpop- option. The "effect" of the wave dummy is the change
in the percentage within a racial group. Alternativly you
could use -logit- with the appropriate -svy- prefix and
-subpop- option in combination with -margins-. This might
reasure people (referees, advisors, etc.) who don't like the
linear probability model, it should give very similar results
though.
Hope this helps,
Maarten
--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany
http://www.maartenbuis.nl
--------------------------
*
* 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/
