Dear all, I'm running a model with one fixed factor which has four groups called "species", and a clustering factor called "nest". My dependent variable (timeto) is "ttm" (time to moult) which is number of days perindividual<http://r.789695.n4.nabble.com/parfm-frailty-model-and-post-hoc-testing-td4672712.html#>, and the Status-variable is called "moulted_final". The code and its results are as follows. library(parfm)> Moult=read.table(file="HSBS R moult2.txt",header=T) >modelMoult=parfm(Surv(ttm,moulted_final)~species,cluster="nest",data=Moult,dist="weibull",frailty="possta") Execution time: 12.72 second(s)> anova(modelMoult)Analysis of Deviance Table Parametric frailty model: response is Surv(ttm, moulted_final) Terms added sequentially (first to last) loglik Chisq Df Pr(>|Chi|) NULL -346.61 species -341.35 10.514 1 0.001184 ** --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 As you can see there are significant differences among species and I would like to know how to obtain these. I'm used to using linear models in which post hoc testing gives you pairwise p-values, but I'm not sure if that is how parfm works. On a side note, all my samples have moulted so "moulted_final" has the same state (1) for all samples. Thanks in advance, Raoul [[alternative HTML version deleted]]