Hi Everyone,

I have survey data that uses brr for its error estimation. I would like to compare the coefficient of x in the first model to the coefficient of x at the second model. Specifically, I would like to see if the coefficient of x 'shrank' with the inclusion of z in the second model

Code:
regress y x z
predict a if e(sample)
recode a (.=.) (*=1)
svy, subpop(a): regress y x   //model 1
svy: regress y x z   //model 2
At this point I'm looking at manually calculating a t-test because I have both means and standard errors for x1 and x2, but I was wondering if there was a more elegant solution.

Cheers,

David.