[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Questions about svy commands

From   Steven Samuels <>
Subject   Re: st: Questions about svy commands
Date   Sun, 10 Feb 2008 09:56:24 -0500

On Feb 10, 2008, at 7:12 AM, Ana Gabriela Guerrero Serdan wrote:

They are difined by province but also by urban/rural.
So I would need to include dummies for urban areas as
well, I suppose?

If each province is split into urban/rural you have (no. strata)= (2 x no. provinces). You could use -xi- and include all strata with an i.urban*i.rural term. You might not need the interaction term. In that case, modify the degrees of freedom by subtracting only the number of interaction terms. If I = number of provinces, the number of interaction terms is I - 1.

Another issue is that I also run regressions for
different groups within my sub-sample of children. So
I have split children into 4 cohorts and in the areas
where they reside. So my regressions are a sub-sub of
the whole sample. This sounds confusing but the idea
is to use a diffs-diffs estimators. So I compare
cohorts in different areas.

You probably don't need other -subpop- statements. For example, if you are computing or comparing means for several cohorts, you could have:

svy, subpop(children): mean varx, over(cohort)


* For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index