nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

How do I create parallel plots to include classes with no members?

TCM

Hi,

I want to illustrate the change in responses before and after a stimulus presentation using a parallel plot. I have a subset of the data; however, this subset does not reflect the entire range of responses. The full set of responses are Extremely similar, Very similar, Somewhat similar, Not very similar, Not at all similar.

Is there a way to include all responses to both sides in order to more clearly demonstrate the change after the stimulus?

Here is the table (col 2 is Before, col 3 is After)

Accepted answers

Jeff_Perkinson

I think the image below is what you're looking for.

2020-11-02_16-23-35.353.png

I made it by adding rows that represent the missing levels in each column. To get them to show on the last column I had to turn on the Combine Sets option on the Parallel element in Graph Builder.

2020-11-02_16-23-58.510.png

Community329055.jmp

All comments

Thierry_S

Hi,
I'm not clear on what you are trying to achieve here: the entire response range from Column 2 / "Before" is covered by the only a subset of responses in column 3. Hence, if I understand correctly, there is nothing to plot for those missing column 3 categories.
Conceptually, it is rather challenging to imagine how you would represent the fact that a response was not observed.
Best,
TS

TCM

Conceptually, one could denote the 5 ordered classes on both posts (before, after) . This just serves the purpose of making clear that there are 5 classes and some classes are not populated.

I remember there is a parallel plot add-in. I'll see what I can do with that and the original dataset.

Thanks for confirming my initial impressions.

tcm

Kevin_Anderson

Hi, TCM!

I think I know what you're after. Please attach an anonymized dataset with which to experiment. Thanks!