Hi!
I ran an experiment where participants took a test of 16 questions, yes/no (binary).
They were tested at either 7days or 28days  two groups between subjects.
Analayzed as subjects, their results for the test are higher at RI7.
RI7: 5.5
RI28: 3.6
ttests confirm that this difference is sig  t = 2.171, p = .032
Analyzed as items, RI7 scores are still higher.
RI7: 515 incorrect  269 correct
RI28: 596 incorrect  172 correct (more incorrect, fewer incorrect).
However
When put into a GLMM (binomial, logit), RI7 scores come out sig lower, no matter what I do and what other variables I put in or take out.
An example from one model:
RI effect: F = 28.542 p = <.001
RI7>RI28 coefficient .625, OR .535, p = <.001
EMMs: RI7: .668; RI28: .790
(more detailed stats in attachment)
Can anyone explain how this is possible or give ideas for something I can check?
Thanks in advance!!
I ran an experiment where participants took a test of 16 questions, yes/no (binary).
They were tested at either 7days or 28days  two groups between subjects.
Analayzed as subjects, their results for the test are higher at RI7.
RI7: 5.5
RI28: 3.6
ttests confirm that this difference is sig  t = 2.171, p = .032
Analyzed as items, RI7 scores are still higher.
RI7: 515 incorrect  269 correct
RI28: 596 incorrect  172 correct (more incorrect, fewer incorrect).
However
When put into a GLMM (binomial, logit), RI7 scores come out sig lower, no matter what I do and what other variables I put in or take out.
An example from one model:
RI effect: F = 28.542 p = <.001
RI7>RI28 coefficient .625, OR .535, p = <.001
EMMs: RI7: .668; RI28: .790
(more detailed stats in attachment)
Can anyone explain how this is possible or give ideas for something I can check?
Thanks in advance!!
Attachments

65.9 KB Views: 2
Last edited: