I constructed a linear mixed-effects model in Matlab with several categorical fixed factors, each having several levels. Fitlme calculates confidence intervals and p values for n-1 levels of each fixed factor compared to a selected reference. How can I get these values for other combinations of factor levels? (e.g., level 1 vs. level 2, level 1 vs. level 3, level 2 vs. level 3).
Thanks,
Chen
I can think of a few ways, although these may not be the best (I'm more familiar with lme in R rather than MATLAB):
betas = fixedeffects(lme);
t = ( betas(3) - betas(4) ) / sqrt( varcov(3,3) + varcov(4,4) - 2.*varcov(3,4) )
Thank you very much! Once I figured out how to change the reference level, re-running the model seemed to do the trick.
Hi Chen! I have exactly the same problem. I need the standard errors for different contrasts. I have added a sample code do you know how to calculate the related SEs?
Thanks!
- Badges
- Science topic