When fitting a GLMM with
lme4, I understand that the coefficients for the estimates of fixed effects in the summary() output must be exponentiated to be back on the scale of the original data.
- are the standard errors for fixed effects in the summary() output also logged and need exponentiatiating? I have seen conflicting answers on other questions, some saying the standard errors are on the original data scale and others saying they have also been logged like the estimates
- do the random effect standard deviations/variances need exponentiating or are they unaffected by the log link so are still on the original data scale anyway?