the red "down" arrow for my fixed effects variable in my repeated measures regression is "greyed out" and not functioning - what am I doing wrong?

Question

I am conducting a mixed effects, repeated measures linear regression with JMP Pro 18 Student version. I have one random effect and one fixed effect with full factorial modelling. The model included effects leverage with the restricted maximum likelihood method. When I run the model, I get no errors and my overall results look sensible....

... but when I click on the down arrow for the  fixed effects variable to look at the LS means table, plot, etc....... that down arrow menu is "greyed out" and non functional.

What am I doing wrong?

LogitOddsElk541 · Answer

You are both right and very helpful.

After more trial and error, here is what I learned:

If I change my fixed effect to ordinal, and then do a full factorial with my random effect, then the program will not run. The first new attachment shows that by doing a full factorial with an ordinal variable that "ordinal terms are not supported for random effects"

So, next I ran the program with fixed effects as ordinal, but I did not do the full factorial, and this was successful in proceeding with Tukey HSD ( see second and third new attachment)

Hooray -  but what do I lose by not performing the full factorial? Is the overall model still legit? Is the Tukey HSD still legit if the full factorial was not performed?

ordinal terms not supported for random effects.png

Running the model without the full factorial of the fixed and random effects and fixed effects are ordinal.png

Tukey HSD Table.png

MRB3855 · Answer

Hi @LogitOddsElk541 . It appears @Dan_Obermiller  is correct. Your third screenshot shows the fixed effect is continuous; notice the parameter estimates…that value -2.443512 is a slope. If “weeks of follow up a fixed effect” were nominal that value would be associated with a particular level of “weeks of follow up a fixed effect”. Or am I misunderstanding?

LogitOddsElk541 · Answer

Hi Dan,

You are very kind to try to help me... thanks!

I have tried various permutations of characterizing the fixed and random independent variables as continuous, or ordinal, or nominal.... but I remain flummoxed. You will note in some of the screen shots that the drop down menu for the fixed effect is greyed out - not allowing me to perform Tukey HSD

I am attaching some screen shots as well as a data set that mirrors the one I have been using.

Any help is appreciated.

Vaughn

Screenshot of model set up.png

ordinal terms not supported for random effects.png

Screenshot of model output when fixed effect is nominal - note fixed effect arrow box is greyed out.png

Screenshot of model output when fixed effect is ordinal and random effect is nominal - note fixed effect arrow box is greyed out.png

Screenshot with random as nominal and fixed as continuous.png

Practice data set for repeated measures.xlsx