I am running a 3 factor (one random) REML model with all interactions. Standard errors of LS means are not calculated for levels of the random effect. Is there a way for this type of model to estimate these standard errors?