I ran logistic Regression and noticed that though the Hosmer and Lemshow test indicates a model fit with an alpha greater than .05, more than 40% of the expected frequencies in the HL contingency table are less than 5 cases, and others have less than 1 (i.e., 0.867). I did some online research and I noticed that most people are not reporting the sample size table. What is the appropriate way of handling the sample size issue? I am planning on using the GLM procedure as an alternative, but if there are other suggestions, that would be great. Thank you.