mixed effects modelling

Discuss statistics related things

by Lucie » Wed Mar 03, 2021 10:21 am

I am looking for some advice...

In terms of mixed effects modelling on GAMLj, how do you know whether to include a random slope in your model or just look at random intercepts?

Thanks
Last edited by Lucie on Thu Mar 11, 2021 11:13 am, edited 1 time in total.
Lucie
 
Posts: 7
Joined: Thu Feb 25, 2021 5:31 pm

by mcfanda@gmail.com » Thu Mar 04, 2021 9:55 am

In general, you can try to put as random coefficients all the coefficients that may vary from cluster to cluster. If you find variability, you keep them in the random component, otherwise, you remove them. In other words, if an effect can be computed within each cluster, then it may be different from cluster to cluster, then it should be tested as random
User avatar
mcfanda@gmail.com
 
Posts: 245
Joined: Thu Mar 23, 2017 9:24 pm


Return to Statistics