Help with use of contrasts in logistic regression

Moderators: statman, Analyst Techy, andris, Fierce, GerineL, Smash

Posts: 1
Joined: Tue Sep 06, 2011 12:05 pm
Location: Sweden

Help with use of contrasts in logistic regression

Postby sindre76 » Tue Sep 06, 2011 12:20 pm

Dear all, I would appreciate help on use of contrast in logistic regression analyses. I am using two binary variables in the model (1 predictor and 1 dependent). Both have the same coding (0s an 1s). The purpose is to predict the probability of converting from healthy to ill on the basis of group membership. When I consult my SPSS cook book it would seem that I should use the indicator contrast with the "first" option. I assume that this is were my problem lies: in the classification table for both blocks I only get cases listed that fit in the first category those with no membership that do not convert. It lists the same percentage correct for both classification tables but as the omnibus test of coefficients is highly significant and as there is indeed significant results for the variable in question it would seem that something went well. I have also checked the results using chi-square (which I believe would give similar answers) and the results, except for the classification table, seem to be correct (e.g. the Haentzel-Manzel results are in agreement with those listed in the variables in the equation table).

Thank you in advance for helping me!

Kind regards Sindre

Who is online

Users browsing this forum: No registered users and 1 guest