| Re: Statistical difference in median ages [message #13781 is a reply to message #13780] |
Thu, 21 December 2017 16:53  |
Reduced-For(u)m
Messages: 292
Registered: March 2013
|
Senior Member |
|
|
How about a quantile regression at the median, including both age-at-first-marraige AND an interaction of age-at-first-marriage with "second survey round"? Then the coefficient on the interaction gives you the difference between the two rounds and you can do a simple T-test (if the interaction coefficient is statistically significant, that's a test of the difference).
Another option would be a "seemingly unrelated regressions" version for quantiles... but looking online I'm not sure anyone has done that (or at least I don't see a Stata package for it). I'd say the interaction method above would be good - you could compare means using Seemingly Unrelated Regression (sureg/suest in Stata) by doing two separate regressions and testing the coefficients after...but it sounds like you want the median.
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
