DF Problem in Goodness-of-Fit: Grouped Data in Nominal Logistic and GLM/Binomial/Logit Platforms

bwalters1

 I have data that I am creating a binary logistic model from k x 2 x 2 grouped data. I know how to create a new data table that lists every observation but I get a really long data table. How can I use the count data that my professor gives us in the GLM platform? Also, how do I calculate the model degrees of freedom once that model is created? 

 

 

 

 

Mark_Bailey

The likelihood ratio test of the whole model is based on the degrees of freedom for the model, in this case 2. That is, model DF for 2x2 cross-tabulation is (n1 - 1) + (n2 - 1) = (2 - 1) + (2 - 1) = 2.

 

The degrees of freedom for the deviance test are based on the hypothesized 'errors' from the model, not the model parameters, that are used to fit the saturated model, in this case 1117. So 2 + 1117 + 1 = 1120 sample size.