> would the following follow-ups via glt be appropriate if I wanted to know look at all of the main effects
> and interactions with and without trauma severity (PCL5; see script below)?

The question does not make sense to me. Once you incorporate a quantitative variable (e.g., PCL5 in your case), you can make inferences by pretending the absence of its effect. Instead, you can only control the effect of the quantitative variable at a particular value. For example, the first 7 post-hoc tests in your script are conducted with PCL5 fixed at the center value. I'm not so sure whether the next 7 post-hoc tests are really what you wanted, but they give you the difference of the PCL5 effects between the two levels of a factor.

> I am a little confused about glf. The description in the help page of glfCode makes it seem like it is
> typically used for testing the main effects and interaction -but isn't that what 3dLME is already doing?

They are meant to provide an F-test for a subset of a main effect or interaction. For example, the typical main effect for a factor with 4 levels tests whether there are differences among all the 4 levels, but you can use -glf specification to test whether there are differences across, for example, the first 3 levels.

Gang



Edited 1 time(s). Last edit at 03/07/2019 09:37PM by Gang.