How to Visualize a Binary Moderator?

[AdrianWorkingWithPLS]

Hello everybody,

I currently developed a research model that includes a binary moderator (with the values 0 and 1). For reporting my results, I seek to visualize the moderation effect using a simple slope plot. However, as the standard slope plot implemented in Smart PLS 3.2.6 includes three levels of the moderator (–1 SD, Mean, +1 SD), I think that it does not appropriately reflect its binary nature. Rather, a slope plot that focuses on the two levels of the moderator (0 and 1) would be more helpful. Can anybody give me some hints how to visualize a binary moderator in this regard?

Up to now, I read through the web pages provided by Jeremy Dawson and Kristopher Preacher, but I feel that I’m not able to use their tools as the requirements of the data input seem not to fit the PLS output (e.g., regarding the lack of values for the intercept/constant or for the unstandardised regression coefficients required). If anybody could provide some clarifications regarding this issue, I would be greatly thankful.

Thank you!
Adrian

[jmbecker]

With the standardized results that you get from the normal PLS it is quite hard to generate a plot for a binary moderator as the binary variable is also standardized and thus not in its original 0/1 format anymore.
However, you could use the IPMA which unstandardizes the PLS results and thus gives you the information that you need.

[AdrianWorkingWithPLS]

Dear Dr. Becker,

Thank you very much for your quick answer. Ok, I understood that I can get the unstandardized path coefficients from the IPMA. Besides, I guess I can get the mean / SD values of my independent variable by just copying the unstandardized latent variable scores, and computing mean and SD values in Excel. Still, I struggle to find the intercept / constant in the IPMA. Can you tell me where I can find it?

Finally, I wonder how my control variables might influence the plot as Jeremy Dawson wrote: "if you have control variables in your regression, the values of the dependent variable displayed on the plot will be inaccurate unless you also standardise (or centre) all control variables first (although the pattern, and therefore the interpretation, will be correct)". So, I guess the output provided by the IPMA still contains this issue, right?

Thank you again for clarifying!
Adrian