Plotting beyond 3 variables in RM ANOVA

General help and assistance with jamovi. Bug reports can be made at our issues page: https://github.com/jamovi/jamovi/issues . (If you're unsure feel free to discuss it here)

by PoLyGLoT » Tue Oct 30, 2018 2:19 pm

Hi all,

First, some praise. I absolutely love jamovi. I love the overall design and simplicity of the program. It is far and away the best stats program I have ever used (and I have tried just about all of them). So hats off to everyone for making such an excellent program. You will probably see me around here a lot as I will be following future developments closely. :grinning:

Now to my question. I really like the default plots in jamovi, but is there any way to go beyond 3 levels of plotting in a RM ANOVA? I'm referring to the plots generated in the "Estimated Marginal Means" section.

Problem: Currently, you can put 3 variables in the box, but no more.
What I want: I want to put 3 variables in the box, but then plot those variables split by another variable that puts each new "level" on a new row (e.g., 1 row for "Low", 1 row for "Medium" and 1 row for "High").

I am extremely new to R, but I think what I want to do is basically introduce a "facet" like in ggplot2 for the variables in the EM means box.

I downloaded the Rj module, and I can see the code plotting the graphs. I have no problem working with the syntax if need be, I am just curious as to what the command would be.

Alternatively, if I could somehow plot this using the filter command (e.g., filter to only plot "Low", then filter to plot "Medium" etc.) I could then combine them all myself.

Also, is there any way to change the plots that come out of Rj module to the "Dark2" theme like I get in the settings of jamovi, with the same size / settings?

Thanks again for any assistance and thanks again for making this program available to all of us.
PoLyGLoT
 
Posts: 1
Joined: Tue Oct 30, 2018 2:00 pm

by Ravi » Tue Oct 30, 2018 10:19 pm

Hi PoLyGLoT,

At the moment you are limited to three variables in the estimated marginal means. I assumed this would cover most use cases, as it seems quite impossible to interpret a four-way interaction anyway. But if you'll make an issue on github, I'll see whether I can extend it at some point: https://github.com/jamovi/jamovi/issues.

Filtering wouldn't give you the same answer, as the estimated marginal means are based on the model that includes the four-way interaction. By filtering you would fit three separate models and therefore the plots could come out differently.

So you can add the jamovi theme with the "Dark2" color palette to ggplot2 plots as follows:
Code: Select all
jmvcore::theme_default(scale='discrete', palette='Dark2')


However, you cannot add themes directly to the jmv plot object at the moment :upside_down_face:

Cheers,
Ravi
User avatar
Ravi
 
Posts: 140
Joined: Sat Jan 28, 2017 11:18 am

by reason180 » Wed Feb 13, 2019 3:56 pm

I believe a four-way interaction can be interpetable, depending on what the data pattern looks like (see the example below. So it would be nice for jamovi to take it one step further to accommodate four-way interactions in marginal means plots.

Image
reason180
 
Posts: 27
Joined: Mon Jul 24, 2017 4:56 pm


Return to Help