Hi I used Linear mixed models since I have multiple evaluations of the same subject at each level of the other variables. I tested a Three-way interaction with two dichotomical and a continuous predictors.
The three.way interaction is significant, and I puctured it with graphs in the paper.
I commented the interactions based on graphs.
The reviewers asked me to prove the interaction significance. How is the best way to proceed in this case?
Since you have a three way interaction you have a term in your model of the form, bX1*x2*X3. The reviewers are apparently asking for the p-value on b. Simply quote that. with the usual additional information. Best wishes, D. Booth
Hi Maria,
If you have the p-value for the interaction in your paper, and still need to prove the interaction, one way would be to compare models with and without that interaction using likelihood ratio test. If this test is significant, then the interaction can not be removed from the model. Otherwise, the interaction is not adding nothing to the model, and can be removed.
Good luck,
Eliana
Eliana Ibrahimi You are correct,of course, but that F test is equivalent to the t-test on b that I recommended with F=t^2.. At this point, it is possible that what they are asking is what does having that term in the model mean in terms of the question the model answers in which case they would not be using significance in the statistical meaning but rather the scientific one I suppose. In such a case, I would have no recourse except to explain the plot further or ask the editor what s/he thought they meant. In that case I would cite the obvious sections in: https://www.google.com/search?q=montgomery+douglas+c.+design+and+analysis+of+experiments&oq=Montgomery&aqs=chrome.1.69i57j69i59j0l5j69i60.9978j0j7&sourceid=chrome&ie=UTF-8 Best wishes, David Booth
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS