I have built a 2-compartment models. after the stepwise program, and then I got the final model with some covariates.However, in my theta output, I am not getting any stderr or %CV estimates. What should i do next.
1. I known i can change the stderr method such as fishor, but this method has low accuracy, some parameters' cv% was much high, and confidence interval including 0, so when i write article, this result is acceptable?
2.on the other hand, when i "accpeted all fixed and random" in the final model, and then run again untill i got the cv%, this method was corrected?(because this model including covariates,so maybe the OFV was changed, which different in final model )
3.if i can changed the algorithm in the final model？(such as QRPEM), and If I do this, do I need to run from the basic model?
4.the CV% and 95%CI in the bootstrap（1000 simulation，can it be substituted as CV% in the final model when I write article, Is there literature support?
I really need help,
have a good day.