SPSS (version 6.1 and later) uses the command GENLOG to fit
loglinear models.
There are two important parts to the command:
A definition of the table being modelled:
genlog housing vote fsex
and a definition of the model being fitted to it:
/design housing vote fsex vote by fsex.
This example defines the table to be analysed as the
housing by vote table, and fits a model to it
which contains the three `main effects' and the
vote*fsex interaction:
It produces an intimidating amount of output, including
graphs. Much of it is useful but for the moment we will ignore
most of it, except the model fit statistics (look for
Likelihood Ratio towards the end of the output.
