1
$\begingroup$

I would like to conduct a GEE analysis (repeated measures logistic regression) but I am not sure whether my data allows me to run a GEE Binary Logistic Model or if I should use GEE Binomial Identity Model (custom).

My participants took a pre-test then learned words and were tested on the same words again. My dependent variable = score on posttest/word (0 or 1) (words already known on pre-test are left empty) My subject variable = participant My within-subject variable = words Predictors: two categoricals and one scale (that characterized words) + 1 scale (that characterized participants). Here is what my data looks like: enter image description here

I want to know whether words that have the value 1 for the predictors are more likely to be learned than words that have value 0... Can someone help please? Is it allowed to use a Binary Logistic in this case? Thanks!

Improve this question
$\endgroup$
3
  • $\begingroup$ Hi, what is a “binomial identity model”? $\endgroup$
    – T.E.G.
    Commented Aug 22, 2020 at 14:11
  • $\begingroup$ Binomial is used when the Y variable is dichotomous (0 or 1) and the identity link is sometimes used for binomial data to yield a linear probability model. $\endgroup$
    – Fievez Isabeau
    Commented Aug 24, 2020 at 12:41
  • $\begingroup$ @FievezIsabeau Not exactly. Binomial does not actually require the Y is dichotomous - it can model proportions. There are other models (cloglog, poisson) to consider in case Y is dichotomous. Specifying binomial in this case creates a mean-variance structure such that the variance is p * (1-p) where p is the mean. $\endgroup$
    – AdamO
    Commented Feb 22 at 17:10

1 Answer 1

Reset to default
1
$\begingroup$

There's no difference in the data to which these two models can be applied. Both are designed for binary response data. Technically, a linear probability model can't be correct with binary data, since it generally predicts probabilities outside of the 0-1 range of the response, but as a practical matter the results from the two are often pretty similar.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.