Effect size two groups with time additive model

questions concerning analysis/theory using program MARK

Effect size two groups with time additive model

Postby constant survivor » Sun Dec 20, 2020 1:51 pm

Hello,
I am considering two groups: Group W and Group E.
31 Occasions.
The model with lowest AIC is phi(t/g+t)p(.) --> TSM model. 'g' is for group.

I am interested in calculating the effect size of the factor group.

Up to now, I worked around it by building model phi(t/g)p(.) and followed the instructions in chapter 6 for calculating the effect size. I built the design matrix analogous to the example in the book so that beta2 reflects the effect size. I used identity link function as well as logit link function and was able to get the same results for effect size, SE and CI from both of them and cross checking it with the real parameter estimates... all worked fine.

If I apply the same technique to model phi(t/g+t)p(.) I get a very similar result for the effect size. I just wonder how one would calculate an SE and CI for that because due to the time factor in the model, this now obviously depends on which time interval is considered.

I thought of giving a range for the SE and report the lowest/highest confidence limits to account for this.
But I guess this would just be a workaround.

So how would this be done correctly if I wanted to use model phi(t/g+t)p(.) for this? Is there anyway a need for the time factor or should I simply consider the model that drops time out of the euqtion?

Thanks
Hannes
constant survivor
 
Posts: 36
Joined: Wed Dec 11, 2019 12:20 pm

Return to analysis help

Who is online

Users browsing this forum: Google [Bot] and 9 guests