Reporting weighted vs. unweighted data in descriptive analysis

[Pat Burns]

Dear Stata Community:

I am using a large nationally representative dataset where data is nested (children within schools within districts, etc.) and where populations (Asian and Native American children) were oversampled to attain a large enough sample size for these groups. As a result, for statistical analysis, I need to weight the sample using the .svyset command. My question is, when reporting the sample means and SD's for descriptive purposes only, do I report the weighted or unweighted numbers?
Many thanks,
Pat

[Clyde Schechter]

I suppose it depends on the context in which you are making a report. If you are publishing a paper or a written report, you should present both the weighted and unweighted descriptive statistics (and make sure it is clear which is which). If you are writing a brief abstract, I would present only the weighted data, with a clear statement that is the weighted, not the raw data. If you are making a presentation to a live audience, present both if time permits. If not, present only the weighted, again with a clear explanation.

[Pat Burns]

Clyde:
Thank you so much. I will present both.
Another question regarding weighted data... given the limitation of the svy: prefix with certain analytics, is there an equivalent of an anova analysis, with sheffe's post hoc test? Right now I am running the following:
.svy: mean ZX1MTHETK4, over(MOMDEG)
.estat sd
where ZXIMTHETK4 is a continuous variable of math achievement; MOMDEG is a categorical variable of mom's degree (5 categories: No high school degree, High School, Some College, College, Masters)
I want to find out if the mean differences are significant with "Some College" serving as the comparison group. Do I run a regression analysis?
I appreciate your time.
Pat

[Pat Burns]

My apologies, that question should have been directed to the entire Stata community with a different heading.