I have a two-year panel data set of patients who all received a treatment at the end of year 1. The panel is split into eight 3-month periods, four before and four after. Patients are coded as having poor kidney function, diabetes and cardiovascular disease - all as dummy variables. The dependent binary variable is a 0 if they were not admitted to hospital in the quarter, and a 1 if they were. I am not interested in controlling for the diseases listed, I am interested in understanding how these different patient groups reacted to the treatment. If I code my treatment as a dummy variable, my input is:
Admitted to Hospital = age + sex + i.kidney#i.diabetes#i.cardiovascular#i.treatment.
I have run a RE xtlogit (using or command) setting patient as the id code and quarter as the time code. My output is quite long - but I understand that all interpretations are related back to the base category. My output says that compared to the base category of "no disease in pre-treatment year", my category of "no disease in post-treatment year" is 1.33 (significant at the 5%). Can I say that these patients were 1.33 times as likely to be admitted to hospital in each quarter, in the year after treatment? I would deeply appreciate any help anyone has to offer.
0 Response to -xtlogit- Four-Way Dummy Variable Interaction Term Interpretation
Post a Comment