Hi! I am new to SPSS, so some of the questions may seem obvious, but please bear with me! I have performed hierarchical logistic regression with dichotomous independent and dependent variables. I have 3 questions regarding significance:
1. The step/block/model are all significant, but Hosmer and Lemeshow Test is also significant, which means the model is a poor fit. How do I judge this situation? The sample is 1100+ participants.
2. Some dependent variables are significant (<.01), but the odds ratio (Exp(B)) is below 1, meaning that saying "Yes" in dependent variable is (1/Exp(B)) times higher when saying "No" in independent variable. For some of these variables it makes little theoretical sense. Is there any other way to interpret the odds ratio data?
3. Which one is more important to the model: significance of individual contributor variables or the odds ratio? For now I thought to focus on the significance and mention the odds ratio in the text.