Re: Testing median for baseline and endline data

[Gebreselassie Tesfayi]

Greetings Stata Users,

I am working on baseline and endline surveys data collected using complex survey method (multistage sampling). I wanted to test whether the baseline and endline median consumption expenditure are statistically significantly different. Is there a built in statistical test in Stata that support complex survey design (i.e., with svy command)? Please advise.

Thank you.

[Carlo Lazzaro]

Gebreselassie:
you may want to take a look at Stas Kolenikov's (who was, in the past, a frequent and outstanding contributor to this forum and I hope that he will return to bein the near future) reply on a similar issue (https://forum.ipums.org/t/using-svy-...ing-median/562).

[Gebreselassie Tesfayi]

Thank you, Carlo.

[Gebreselassie Tesfayi]

Dear Stata users, I downloaded epctile program from Stas Kolenikov's website and was able to generate confidence intervals. However, since the sampling distribution of the median is not N(0,1) can I use the t-test to test whether the baseline and endline median consumption expenditure are statistically the same/different? Thank you.

[Carlo Lazzaro]

Gebreselassie:
have you already considered the -bootstrap-?