note: 0b.HX_HTN != 1 predicts failure perfectly
0b.HX_HTN dropped and 14 obs not used
note: 0b.HX_DM != 1 predicts failure perfectly
0b.HX_DM dropped and 5 obs not used
When performing logistic regression, the categorical variable that perfectly predicts failure of binary outcome is dropped, thus reducing the corresponding observation.
​​​​​​​
Is there any option or setting to conserve observation from dropping?