Dear all,
I have recently received a review for a paper where one of my reviewers asks
whether we can give "a global significance" of two categorical variablesthat
are entered as fixed effects in the model.
The model is a multilevel logistic regression where use of personal
protective equipment (PPE) is the outcome variable yes/no.
The use of the multilevel feature is dictated by the fact that I have panel
data of a cohort of about 10,000 workers measured at different point in time
(between 1 to 25 measurements per worker), not necessarily at the same time
and who work in 10 different companies.
The model has two random effects (random intercepts): one for the worker and
one for the company.
The variables that the reviewer is interested in are the job (has 14
different categories) and the department (has 10 different categories).
I am not aware of any command or option that gives a p-value or what the
reviewer calls a "global significance" which I guess would be the magnitude
and effect of job and department as a whole rather than the betas for each
categories.
Your help would be immensely appreciated
Many thanks,
Hind
|