Page 1 of 1

using runmlwin with survey weights

Posted: Fri Sep 11, 2020 12:00 pm
by becciaa1992
Hello,

I am trying to fit a two-level multilevel logistic model in runmlwin using survey weights. The model runs fine using PQL2 estimation methods, however, I get the following error when trying to run the same model using MCMC methods: "Weights are only valid for univariate models estimated using (R)IGLS"

Here is the code I am using:

Code: Select all

* Fit model using PQL2
runmlwin outcome cons , /// 
  level2(group: cons) ///
  level1(id:, weightvar(weight)) ///
  discrete(distribution(binomial) link(logit) denominator(denominator) pql2) ///
  rigls maxiterations(100) ///
  nopause
  
* Fit model using MCMC
runmlwin outcome cons , ///
  level2(group: cons, residuals(u, savechains("u.dta", replace))) /// 
  level1(id:, weightvar(weight)) ///
  discrete(distribution(binomial) link(logit) denominator(denominator)) ///
  mcmc(burnin(5000) chain(50000) thinning(50) ///
    savechains("beta.dta", replace)) initsprevious /// 
  nopause 
I do have the most recent version of runmlwin. Any help would be appreciated, thank you!

Re: using runmlwin with survey weights

Posted: Fri Sep 11, 2020 1:29 pm
by ChrisCharlton
Unfortunately weighting is not currently implemented for MCMC within MLwiN, so you will be unable to fit this here. See http://www.bristol.ac.uk/cmm/software/s ... ights.html for more details on using weights with MLwiN.

Re: using runmlwin with survey weights

Posted: Fri Mar 12, 2021 2:39 pm
by ManuelDewez
Hello Chris,

I know that the post is from September 2020, but has there been any change on this topic, is weighting possible with MCMC in MLwin?

Thank you

Manuel

Re: using runmlwin with survey weights

Posted: Fri Mar 12, 2021 3:29 pm
by ChrisCharlton
Sorry, there is no change here, and this is not currently in our in our future plans.