4 questions concerning repeated measures analysis in JMP

[François Vézina <f.verzina@hetnet.nl>]

Hello there, I am presently struggling with my stat outputs from JMP (version 5 for mac).  I am an animal physiologist and I am looking at changes in time in body mass in birds exposed to three thermal treatments.

So I have two months of data (I will later extend it to a full year)
3 treatments (warm, cold and variable)
2 cages per treatment (6 birds per cage)

Here is what I do. In the Fit Model box I choose mass1 and mass2 as Y

Then,  I add as x:
Cage(treatment) ... cage nested in treatment...
treatment

I run it as a MANOVA and ask for Repeated Measures. It add the Time variable and then I get the output... and that is where I have troubles...

I think I understand the Between Subjects section... basically I'm only interested in the treatment effect there.
However, I have troubles with the Within Subjects section.

1) Is TIME only testing for an overall time effect independent from any treatment effect? (meaning that opposing trends in different treatments would result in no significant overall time effect)

2) What exactly is ALL WITHIN INTERACTIONS testing for? Is this testing for time effect within treatment? If yes how is this different from the "time x treatment" interaction term?

3) Repeated measure Manova gives the least square means but no SE? Why is that? Can I ask JMP to give me the SE as well?

4) People at my institute use the mixed approach with random variables in repeated measures designs... Using this approach gave me different results.... treatment effect is n.s. while the multivariate approach pick-up the difference (same least square means)... why is that? Is it normal to find different results with the two approach


Thanks for any help... I have been stuck with these problems for weeks and I am the only person using JMP here... I like it!

Francois

François Vézina Ph.D.
Post-doctoral fellow
Department of Marine Ecology and Evolution
Royal Netherlands Institute for Sea Research (NIOZ)
The Netherlands
+31 (0)222 369 383