Dear Experts,
I've a statistic question for you.
I've a 2x2 within subjects design A (a1, a2) x B (b1, b2) so the design
matrix has four regressor a1b1, a1b2, a2b1, a2b2, and I'm interesting in
main effect of A, main effect of B and interaction AxB. I'd want to run the
analysis in an alterantive way regarding the typical way so I apprecciate
any comment and advice.
1) TYPICAL WAY: at 1-st level I run a t-contrast for main effect of A, a
t-contrast for main effect of B and a t-contrast for interaction AxB; after
at 2-nd level I run one sample t-test for each of these three effects in
both directions.
2) ALTERNATIVE WAY: at 1-st level I run a t-contrast for each regressor
a1b1, a1b2, a2b1, a2b2 as [1]; [0 1]; [0 0 1]; [0 0 0 1]; after at 2-nd
level I run one-way anova within subjects with four groups, one for each
regressor, and I explore the three effects in both directions using the t
contrast manager, for example I set [1 1 -1 -1] and [-1 -1 1 1] for main
effect of A and so on for main effect of B and interaction AxB.
Is the alternative way correct?
Which is the most efficent/powerful way?
In typical and alterantive ways is the varinace pooled or partitioned?
Thank you.
--
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.1.405 / Virus Database: 268.12.7/454 - Release Date: 21/09/2006
|