Hosmer Lemeshow goodness of fit test - grouping


New Member

I need to perform Hosmer-Lemeshow test for logistic regression, but I'm not sure how to group data (continuous variable - interval 0-1).
Basically there are two options - divide the interval into equal sections OR divide data into same populated groups.

Which of these two approaches is methodically better? Results strongly depend on the grouping type and number of groups.

Thanks for an answers