I have a question about a specfic output from GENLINMIXED in the newer versions of SPSS.
When I run a glmm I will get a "predicted vs observed values" plot. Now often these turn out quite differently.
Say I have longitudinal data and want to account for repeated measures. I can do this by either using the REPEATED statement or the RANDOM statement with an Intercept for each subject.
I then find that although both options are legitimate tools and the AIC is usually not very different the predicted vs observed plots are completely different. I wonder if that can be traced back to how the models work so if its an model inherent property or what else causes the difference here. Does this say anything about the quality of the models or should I just look at the AIC? Is it sometimes useful for interpretation (if you actually want to do predictions with your model) and sometimes not (if you just want to test and get the correct p-values but you dont want to predict any future outcomes?). The Model with the REPEATED statement outputs a almost horizontal scatter plot for observed vs predicted whereas the model with the RANDOM statement outputs a almost diagonal scatter plot.
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
For a list of commands to manage subscriptions, send the command