1
$\begingroup$

I have two data sets collected from two different sets of participants on their behaviour. EDIT: Both have the same response variables (Propensity to behave in a certain way - Yes or No). But they have some common (sex, age) and some different explanatory variables (education, origin etc) because they were two different surveys and two different set of participants. I have fitted generalised linear mixed effects models (GLMM) with participant ID as random effect separately for these data sets.

Why I would like to apply model averaging here is because I would like to take all the available information into my model from both data sets rather than fitting separate models for the two data sets.

Can I apply model averaging for the best Generalised Linear Mixed-effects models selected from the two different data sets or is it applied only for nested models?

Edit: What I mean by model averaging is averaging across model parameters instead of selecting one best model so that the model uncertainty issue can be addressed. (Eg: AIC and BIC-based model averaging - where weights are applied to the parameters, (Bates and Granger 1969), Bayesian model averaging (BMA) (Raftery et al. 1997))

Thanks!

$\endgroup$
  • $\begingroup$ Please clarify any acronyms you use in your posts. What GLMM stands for in your post might stand for something entirely different for someone else. $\endgroup$ – StatsStudent Jan 31 '19 at 4:00
  • $\begingroup$ @StatsStudent I have edited that. Thanks for pointing out $\endgroup$ – Jessie Jan 31 '19 at 4:09
  • 3
    $\begingroup$ Define "model averaging". $\endgroup$ – user2974951 Jan 31 '19 at 7:20
  • $\begingroup$ @user297451 What I mean by model averaging is averaging across model parameters instead of selecting one best model so that the model uncertainty issue can be addressed. $\endgroup$ – Jessie Feb 1 '19 at 0:40
  • $\begingroup$ Could you still clarify the nature of the data a bit: do the sets of participants differ somehow? Is the response "yes and no" the same question? Why are the covariates partially different between the datasets? (These details might matter) $\endgroup$ – Juho Kokkala Feb 2 '19 at 8:07

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.