The lecture material instructs the students not to interpret the main effects when a significant interaction is present, and the course material goes beyond that and says not to even report that. It might be a school-of-thought kind of thing, but I was taught (and later taught myself) that this depends on the type of interaction - in ordinal interactions (when the simple effects all point in the same direction), there is no real issue, because the underlying conclusion doesn't change, only its magnitude. In disordinal interactions (when at least one or more simple effects point to a different direction or are equal to zero) it is more important to make sure you are not conveying the wrong interpretation. But at any rate, in both cases, results should be reported, which is also something that I believe is important from a research integrity perspective. So in my opinion, there's nothing wrong in testing interactions first in order to not be mislead by the main effects, but I would suggest changing some of the language we use in instructing the students what to do with that information.
I think it is worthwhile to include within the course material something about simple effect analysis as the next logical step to do once you find a significant interaction. This could potentially be in the form of the "bonus" material that just signposts the students who might find it relevant to read more about it, without including any code examples on it, but I think it is important they know that the simple effects need to be investigated similar to a post-hoc analysis on main effects.
Related to the two above points, it might be useful to teach students about ordinal and disordinal interactions, which is something you would need simple effect analysis to test at the inferential level (though you could still talk about it at the descriptive level without that). Again, might be enough as a bonus material, but I find that terminology useful. Can't say how frequently it is being used within biological/medical sciences, though, outside cognitive neurosciences.
It might be worthwhile to add a few slides that show the students how they can read the main effects from the interaction plots. You could of course plot these separately, but if they come across this sort of plot in a paper, it could be handy to know that. This could be done e.g. around slides 188-191, simply by adding a few animations and visual elements to one of the plots should suffice.
On slide 209 it says that the course material will include Two continuous predictors, though I see that is only being covered in CS5. Also related - might be useful if we show them they could visualise two continuous predictors as a 3d surface.
The lecture material instructs the students not to interpret the main effects when a significant interaction is present, and the course material goes beyond that and says not to even report that. It might be a school-of-thought kind of thing, but I was taught (and later taught myself) that this depends on the type of interaction - in ordinal interactions (when the simple effects all point in the same direction), there is no real issue, because the underlying conclusion doesn't change, only its magnitude. In disordinal interactions (when at least one or more simple effects point to a different direction or are equal to zero) it is more important to make sure you are not conveying the wrong interpretation. But at any rate, in both cases, results should be reported, which is also something that I believe is important from a research integrity perspective. So in my opinion, there's nothing wrong in testing interactions first in order to not be mislead by the main effects, but I would suggest changing some of the language we use in instructing the students what to do with that information.
I think it is worthwhile to include within the course material something about simple effect analysis as the next logical step to do once you find a significant interaction. This could potentially be in the form of the "bonus" material that just signposts the students who might find it relevant to read more about it, without including any code examples on it, but I think it is important they know that the simple effects need to be investigated similar to a post-hoc analysis on main effects.
Related to the two above points, it might be useful to teach students about ordinal and disordinal interactions, which is something you would need simple effect analysis to test at the inferential level (though you could still talk about it at the descriptive level without that). Again, might be enough as a bonus material, but I find that terminology useful. Can't say how frequently it is being used within biological/medical sciences, though, outside cognitive neurosciences.
It might be worthwhile to add a few slides that show the students how they can read the main effects from the interaction plots. You could of course plot these separately, but if they come across this sort of plot in a paper, it could be handy to know that. This could be done e.g. around slides 188-191, simply by adding a few animations and visual elements to one of the plots should suffice.
On slide 209 it says that the course material will include Two continuous predictors, though I see that is only being covered in CS5. Also related - might be useful if we show them they could visualise two continuous predictors as a 3d surface.