I am using a three-level mixed-effects model in which:

  1. individuals (level 1) are nested in
  2. primary sampling units (PSUs) or enumeration areas (level 2) which in turn are nested in
  3. countries (level 3).

Using Stata's xtmixed command I get fixed effects for all my variables in standard regression coefficients and random effects for countries and PSUs as a sum of variance and co-variance parameters.


  • If my random-effects parameters are for country (in SD): 0.192*** (SE 0.030) and for PSU (SD): 0.620*** (0.020), how do I interpret them?
  • Should I just report what proportion of the variance is at the PSU level and at the country level?
  • Do these random-effects parameters say anything about the fixed-effect regression coefficients?
  • Can these parameters be used to predict the intercept of the DV?
  • Looking at the random-effects parameters can I justify my mixed-effects model building?

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.