Implementing Moderation analysis using fitrm when i have a 2 within subjects factors

Question

Ketan Jaltare 2020-12-9

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/688434-implementing-moderation-analysis-using-fitrm-when-i-have-a-2-within-subjects-factors

回答： Aditya Patil 2020-12-24

I have a (A)2x(B)2 within subjects design and a continuous moderator variable (X). I want to test whether X moderates the relationship between A and B. How can I impliment this in matlab using the fitrm function?

I have tried the following :

rm = fitrm(Datamatrix,'A-B ~1 + X','WithinDesign',within);

However I am not sure if this is the right thing to do.

Should i be using the fitlm function instead? If so, can someone explain how i should construct my design matrix for it?

Thanks in advance!

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Aditya Patil 2020-12-24

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/688434-implementing-moderation-analysis-using-fitrm-when-i-have-a-2-within-subjects-factors#answer_583362

The idea behind repeated measures is that we measure the response variable over time or under different conditions.

If A and B are somehow the same, for example measurements of the same variable or of comparable variables with the same units, then you may choose to conduct a repeated measures analysis. For example, you may do the analysis on left arm and right arm of subjects.
If A and B are fundamentally different things, then you may choose not to use repeated measures analysis. For example, If A were age and B were weight of some subjects, and they are of very different scales, then you may decide fitrm is not appropriate.

As a workaround, it would be possible to model A and B separately. An advantage is that they would not be constrained to have the same functional relationship. For example, A could be a linear function of X and B could be a quadratic function of X. It would also be possible to study the residuals from the two models in order to decide the degree of correlation between A and B after adjusting for the between-subject models.