I'm performing a logistic regression and I have one categorical variable as an explanatory variable (Hab). Sometimes I get really big S.E for some levels of the independent factor (as you can see from the attached picture. When I checked the data I found out that the dependent variable has only 1 unique value for all the observations corresponding to the factor that has the issue. I want to know how to remove this problem, do we have to make some conditions about the data in order to ensure that all the time the Y has a mixture between 0 and 1 for all the levels of the independent variables? I even performed an LRT for the significance of the categorical variables and it still in the regression even all the levels are not significant.
)
