Thanks for your quick response, but in the book of survival analysis for Medical Research by Collete, he emphasizes that the common log-likelihood for the goodness of fit of cox-model with frailty terms is not acceptable because marginal log-likelihood avlue is obtained by integrating the log-likelihood over the frailty term, which is not the case with the usual cox model in that we obtain the log-likelihood values, The question am asking is when I have these two values, can I compare them? If yes, why?   The AIC can not be used for the frailty term as it is fitted using the penalized likelihood , i don't know whether am right?     ... View more

Yes there are multiple measurements per subject ... View more

I tried running the cox regression using the frailty with penalized likelihood and this code frailtyPenal(Surv(days_events,OUTCOME)~cluster(id)+resqualintra,n.knots=1, kappa=100,data=base) Moreover, I tried running a log-normal survival model and this is code (brack.AFT.GF <- flexsurvreg(Surv(days_events, OUTCOME) ~ resqualintra, data=base, dist="lnorm")) log-normal survivalmodel Cox Frailty with penalized likelihood model As we observe, the parametric model I have a negative estimate and for the Frailty model I have a Positive estimate; SO this is my issue,What could be an explanation to this? ... View more

Thanks 😊 for your reply and concern ... View more

actually i have access to this article but it does not contain the source code for  GEEGOR.mac that is why am asking the file,. this is the code I found in the article, so it shows I need the the GEEGOR.mac file   % "c:\geegor.mac"; %multrpm(dataset=longil_cat,id=id, margin=1, co=1,y=bmi_cat,x=time **bleep**e sex smoking,numcov=5,xcorr=**bleep**e sex ,pc=1, maxit=50,criteria=0.00001); ... View more

Thanks for your reply,  I have tried the code, below and I have very strange results like very high standard errors, what might be the case? because I sude glimmiw for equal slopes and the results look attractive   proc nlmixed data=longil_cat noad qpoints=30; title 'BMICAT, PROC NLMIXED, ordinal, adaptive, q=30'; parms int1=2.5080 int2=4.5586 d= 0.07258; eta1 =int1+ b+beta11*time + beta12***bleep**e + beta13*smoking + beta14*sex + beta15*time***bleep**e + beta16*time*sex; eta2 = int2 + b + beta11*time + beta22***bleep**e + beta13*smoking + beta14*sex + beta25*time***bleep**e + beta26*time*sex; if bmi_cat=1 then z = (1/(1+exp(-eta1))); else if bmi_cat=2 then z = (1/(1+exp(-eta1))) - (1/(1+exp(-eta2))); else z = 1 - (1/(1+exp(-eta2))); if z > 1e-20 then ll = log(z); else ll = -1e10; model bmi_cat ~ general(ll); random b ~normal(0,d**2) subject= id out=EB; estimate 'var(d)' d*d; run; ... View more

But it does not take into consideration the random effect ... View more