Mixed models

Discuss statistics related things
Post Reply
lucaselva
Posts: 2
Joined: Sun Jan 15, 2023 6:49 pm

Mixed models

Post by lucaselva »

I would like to run a mixed model to see the differences in variables between two groups: control (n=7) and test (n=12).
I then have 12 subjects numbered from 1 to 12 and 7 subjects numbered from 1 to 7.
My 2 factors are: time (0/15/30) and group (con/test).

The mixed model gives me results, however, when I change for example subjects 6 to subject 5 (and subject 5 to subject 6), the results (p) change. Whereas they should not change because there are no subject switching factors.

Do you have a solution so that the order of the subjects does not influence my results?

Thank you,
Luca
User avatar
mcfanda@gmail.com
Posts: 457
Joined: Thu Mar 23, 2017 9:24 pm

Re: Mixed models

Post by mcfanda@gmail.com »

That is an expected behaviour of the mixed model. The correct model should have each participant with an unique ID
lucaselva
Posts: 2
Joined: Sun Jan 15, 2023 6:49 pm

Re: Mixed models

Post by lucaselva »

Hello, thank you very much for this answer!

So, is it a problem or not if we compare 2 different populations with mixed model ? (the 12 test subjects being different from the 7 control subjects)
User avatar
mcfanda@gmail.com
Posts: 457
Joined: Thu Mar 23, 2017 9:24 pm

Re: Mixed models

Post by mcfanda@gmail.com »

if you put group as a factor in the model, you are testing that the two groups means come from different populations
Post Reply