Hi listers,

I am looking to see whether emotional well-being scores (continuous) are predictors of hospitalisation (DV; yes/no). I am using a logistic regression adjusting for IMD and education (both are categorical with 10 and 5 levels, respectively).

logistic hospital bl_well i.IMD i.edu

I would then like to obtain the OR for those who reported no emotional well-being (scored 0) and those who scored 100 (max score on the questionnaire). I thought I could use margins . I thought something like but I am not sure:

margins , at(bl_well(0 100)) atmeans

Is at means OK to use with categorical covariates?

The OR for well-being on hospitalisation is 0.80 (0.68 - 0.94) but when I look at the margins results they don't seem to be comparable.

_at 1: 0.019 (0.18- 0.19)
_at 2: 2.24e-13 (-3.55e-12 - 4.00e-12)

Where am I going wrong?