On Tue, 9 Mar 1999, Stephen Black went:
> Specifically, does anyone know how I can do a test for a difference
> between two trends on SPSS?
In SPSS, no.
Based on the way you describe the question, I _think_ you might find
some general guidance from this paper, available in pdf format at
www.educ.msu.edu/units/groups/lammp/psu1.pdf
Stephen W. Raudenbush. Toward a coherent framework for comparing
trajectories of individual change.
It's about a family of procedures known variously as hierarchical
models, mixed models, latent-growth models, multilevel models,
random-effects models...the list of terms goes on, and Raudenbush
argues that people's preferred term is usually dictated by the
software they use. SAS does this stuff with PROC MIXED (and the
GLIMMIX macro if you have a repeated binary DV). SPSS, as I said,
doesn't.
I've been trying hard to understand all this. It ain't no t-test.
Actually, if there's anyone out there who can help me intrepret the
the output of the GLIMMIX macro, I'd be awfully appreciative.
--David Epstein
[EMAIL PROTECTED]
