Page 1 of 1

Hosmer and Lemeshow how to allocate blocks to groups

Posted: Thu Jun 12, 2014 5:37 pm
by gysc
Hello, I am interested to know how ties are handled in the SPSS implementation of the Hosmer and Lemeshow test in logistic regression. I realise that some of my observations will have the same risk (predicted value) and will therefore belong to the same block. I also realise that SPSS will not split members of the same block amongst different groups (there are usually 10 groups). The question is, if a block spans a group boundary, how does SPSS decide which group to put the entire block into?