Page 1 of 1

Problem Using Sampling Weights

Posted: Fri Mar 09, 2018 8:04 pm
by SoumavaBasu
Hello,
I am using the runmlwin command in STATA 13. the simple model I am trying out is working fine when no sampling weights are given.

. runmlwin x_out cons if gender~=. & pwt1~=., level2(group: cons) level1(ind:) discrete(distribution(binomial) link(logit) denominator(cons) pql2)
> maxiterations(30)

However, when I introduce weights, MLwiN stops working:

. runmlwin x_out cons if gender~=. & pwt1~=., level2(group: cons, weightvar(pwt2)) level1(ind:, weightvar(pwt1)) discrete(distribution(binomial) l
> ink(logit) denominator(cons) pql2) maxiterations(30)

Following error is shown in STATA:
The model did not run properly in MLwiN. You most likely clicked the 'Abort Macro' button in MLwiN, rather than the 'Resume Macro' button.

r(198);

Kindly let me know if I did anything wrong here.

Thanks,
Soumava

Re: Problem Using Sampling Weights

Posted: Fri Mar 09, 2018 10:08 pm
by ChrisCharlton
There was a bug in MLwiN 3.01 where the missing indicator column wasn't recalculated after specifying weights, causing a potential crash. Could you try updating to version 3.02 (available from https://www.cmm.bristol.ac.uk/clients/softwaredownload/) to see whether that fixes it for you?

Re: Problem Using Sampling Weights

Posted: Fri Mar 09, 2018 11:21 pm
by SoumavaBasu
Hi Chris,
This seems to have worked. Thank you so much.
Soumava