[R-sig-ME] logistic regression on posttest (0, 1) with pretest(0, 1)*Group(Treatment, Ctrl) interaction

[Sorkin, John]

Souheyla,

Souheyla,

I don't know every R package so I can't make a blanket statement. Searching the internet makes it appear that  using lm(y~x*z) will produce a model with the main effects and the interaction, i.e. y=f(x,z,x*z). Nevertheless, I believe it to be best coding practices (especially when sending your code to a third part) to always expressly give all terms in a model and never leave the main effects out of a model with an interaction. I am sorry I can't be of any additional help with the more fundamental aspects of your original posting.

John


John David Sorkin M.D., Ph.D.
Professor of Medicine
Chief, Biostatistics and Informatics
University of Maryland School of Medicine Division of Gerontology and Geriatric Medicine
Baltimore VA Medical Center
10 North Greene Street
GRECC (BT/18/GR)
Baltimore, MD 21201-1524
(Phone) 410-605-7119
(Fax) 410-605-7913 (Please call phone number above prior to faxing)


________________________________
From: Souheyla GHEBGHOUB <souheyla.ghebghoub using gmail.com>
Sent: Sunday, April 21, 2019 2:07 PM
To: Sorkin, John
Cc: Ren�; r-sig-mixed-models using r-project.org
Subject: Re: [R-sig-ME] logistic regression on posttest (0, 1) with pretest(0, 1)*Group(Treatment, Ctrl) interaction

Dear John,

Thank you for your email.
I used to think that  posttest ~ pretest*Group will automatically give you the main effects of group and pretest without having to set them again separately in the syntax? Please correct me if I am wrong.

Thank you again,
Souheyla


On Sun, 21 Apr 2019 at 19:02, Sorkin, John <jsorkin using som.umaryland.edu<mailto:jsorkin using som.umaryland.edu>> wrote:
Souheyla,

It is both difficult and dangerous to add a comment to a thread that one has not followed, and in doing so possibly making an inappropriate suggestion. Please forgive what may be an not fully informed thought.

The model you suggest, posttest ~ pretest*Group  (ignoring random effects) is unusual. In a model that contains an interaction,  I would expect to see, in addition to the interaction, all main effects included in the interaction, i.e.
posttest ~ group+pretest+pretest*Group

John


John David Sorkin M.D., Ph.D.
Professor of Medicine
Chief, Biostatistics and Informatics
University of Maryland School of Medicine Division of Gerontology and Geriatric Medicine
Baltimore VA Medical Center
10 North Greene Street
GRECC (BT/18/GR)
Baltimore, MD 21201-1524
(Phone) 410-605-7119
(Fax) 410-605-7913 (Please call phone number above prior to faxing)


________________________________
From: R-sig-mixed-models <r-sig-mixed-models-bounces using r-project.org<mailto:r-sig-mixed-models-bounces using r-project.org>> on behalf of Souheyla GHEBGHOUB <souheyla.ghebghoub using gmail.com<mailto:souheyla.ghebghoub using gmail.com>>
Sent: Sunday, April 21, 2019 4:57 AM
To: Ren�; r-sig-mixed-models using r-project.org<mailto:r-sig-mixed-models using r-project.org>
Subject: [R-sig-ME] logistic regression on posttest (0, 1) with pretest(0, 1)*Group(Treatment, Ctrl) interaction

Dear Rene, and any member of the list who is willing to read this : )

I have decided to use the interesting model that you have structured for me,
To refresh your memory here is what you said :

*Anyway, there is a second possibility to define your model, depending on
> how you want to interpret it. In the previous model you can say something
> about the type-of-change likelihoods depending on the treatment group.
> But you could implement the model as binomial as well (i.e. logistic
> regression) mod2 <- brm(posttest ~ pretest*Group + (1|Subject) +
> (1+Group|Word),...)  And what you would expect here would be an interaction
> between pre-test and Group. For instance; if pretest=0 & treatment 1 then
> posttest larger than with pretest=0 & treatment 2; but not when pretest=1
> (because this is a plausible no gain situation). And so on...*


But I found the interpretation of the Pretest*Group interaction results to
be tough. I still can't grasp how to claim there is an effect of Group on
posttest outcome when considering pretest. The pretest slope does not come
with 0 or 1 in the output, the Group does come with one category, but its
confusing what the intercept and slope estimates refer to in this case?

Thank you very much,
Souheyla

        [[alternative HTML version deleted]]

_______________________________________________
R-sig-mixed-models using r-project.org<mailto:R-sig-mixed-models using r-project.org> mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models

	[[alternative HTML version deleted]]