I am using the PROCESS Macro to conduct a moderation analysis. To visualise the significant interaction, I copied and pasted the necessary text and created a new syntax. The scatterplot was produced but I am unable to add fit lines at the subgroups (please see screenshot of error message attached).
Does anyone know why this might be? I have had a browse online and have seen a lot of mention about re-coding categorical variables when faced with this problem. However, in my case, the variables included in the creation of the scatterplot are all continuous.
You may wish to look at the material from the attached search. The plots in R should be fairly easy if that's all you need to do. Best wishes, David Booth
Hello Holly,
So the moderator and DV are continuous, and you want to plot the relationship separately for each group? In SPSS, go to Graphs / Regression Variable Plots, put in DV as vertical var (Y-axis), moderator as horizontal var (X-axis), and use the grouping variable as "Color by" option. Click "Options" then check the "Scatterplot Fit Lines / Linear" box, then click on the "Grouping / Fit line for each categorical color group" radio button. That will get you your respective subgroup regression lines on the same plot.
Using syntax, it would look something like this (vars are "dv" "mod" and "group":
STATS REGRESS PLOT YVARS=dv XVARS=mod COLOR=group
/OPTIONS CATEGORICAL=BARS TITLE="DV regressed on moderator" GROUP=1 BOXPLOTS INDENT=15 YSCALE=75
/FITLINES LINEAR APPLYTO=GROUP .
Good luck with your work.
@Holly. I don't want to criticize my good friend David Morse. But I would suggest looking at the University of Virginia materials prior to trying the SPSS approach. I did my first time ever approach to this with total time spent for the complete analysis of about 30 minutes using the Virginia materials. This was a few years ago now and the Virginia stuff is completely revised and easier. With the free download from the z-library I recommended you should have no trouble in doing the entire analysis in less than about one hour including plots. Plus you are introduced to the most powerful stat package in existence. Try it out and see. With apologies to my dear friend David Morse, Best wishes, David Booth
Oops I forgot you can run R from your smartphone too just google this to see how to do it. Best wishes, David Booth