Interaction effect with continuous criterion variable, dummy predictor and dummy moderator as linear plot?

[Ava Cerrid]

Hello everyone,

I have a question regarding the adequacy of using a linear visualisation plot for an interaction effect of a continuous criterion variable, a dummy predictor variable and a dummy moderator variable. Since the association between these variables is not linear (because of the dichotomous nature of the predictor and moderator), is there an alternative plot to visualise this? Apparently, many people use a linear graph for such purpose, at least this is what my research pointed towards. In some cases boxplots were used, but I find the interpretation of this graph less intuitive and understandable.
I calculated an OLS regression and used marginsplot to visualize my results, if this is important to know.

I would be very thankful for any advice or insight on this matter.

Kind regards,
Ava.
__
PS: I added an example of my plot so you know how it looks at the moment.

[Phil Bromiley]

You didn't get a quick answer. You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex. When you say criterion - does that mean the dependent variable? Knowing exactly what you ran would make it easier to help you.

With a dummy variable moderated by a dummy variable, you only have 4 conditions. Whether graphing that helps or not is not clear. You could use post to save the margins results for the four conditions and then make up a table or bar graph.

[Ava Cerrid]

You didn't get a quick answer. You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex. When you say criterion - does that mean the dependent variable? Knowing exactly what you ran would make it easier to help you.

With a dummy variable moderated by a dummy variable, you only have 4 conditions. Whether graphing that helps or not is not clear. You could use post to save the margins results for the four conditions and then make up a table or bar graph.

Thank you very much Phil for your answer! Unfortunately I cannot share my Statacode, since the data is confidential, but thanks for the advice on using sample data. I did not consider that. Indeed I mean the dependent variable with "criterion". I thought it is a know term, I am sorry.
I was advised to visualise the interaction effect results and also thought about a bar graph. By this I would avoid the problem with the non-linear association. I just wondered whether it is ok to use the above shown graph since I have seen it many times when looking up graphs for interactions effects with this specific variable constellation.

I highly appreciate your input on my question and on general Statalist procedure!