Hi
On Tue, 19 Nov 2002, Annette Taylor wrote:
> It sounds to me like you want to do post-hoc comparisons. I
> was taught, in graduate school, so it was some time ago, to
> do all the necessary comparisons I want with t-tests and then
> use a bonferroni correction for alpha. According to Simes
> (1986) you can rank order the p-values from smallest to
> largest for all the t-tests you performed and then determine
> the critical value by taking each value in order as follows:
> for the first critical value it is calculated as (1 times
> .05)/total number of comparisons made. The second critical
> value is (2 times .05)/ total number of comparisons made.
> Continue in this fashion for each comparison. As soon as you
> reach a nonsignificant comparison you stop. All the others
> after that are nonsignificant regardless of the math. But
> there are other bonferroni corrections you may be able to
> find and that make more sense to you.
My reading of Rob's post was that he was doing simple effects
following a significant interaction, and I posted the MANOVA
commands to perform that analysis. I do not think that a
correction is needed. The commands were:
manova score by eth(1 5) cit(1 2)
/design cit w eth(1) cit w eth(2) cit w eth(3)
cit w eth(4) cit w eth(5) eth
This is essentially an alternative partitioning of the 9 df for
this design. That is, rather than
df eth = 4 = 5-1
df cit = 1 = 2-1
df eth = 4 = (5-1)(2-1)
the df are
df cit w eth(1) = 1
df cit w eth(2) = 1
df cit w eth(3) = 1
df cit w eth(4) = 1
df cit w eth(5) = 1
df eth = 4
A simple effects analysis groups the SS for one of the main
effects and the SS for the interaction (and their corresponding
dfs) and then re-allocates them to the simple effects. It is
closer in spirit to planned comparisons than to post hoc tests.
Best wishes
Jim
============================================================================
James M. Clark (204) 786-9757
Department of Psychology (204) 774-4134 Fax
University of Winnipeg 4L05D
Winnipeg, Manitoba R3B 2E9 [EMAIL PROTECTED]
CANADA http://www.uwinnipeg.ca/~clark
============================================================================
---
You are currently subscribed to tips as: [EMAIL PROTECTED]
To unsubscribe send a blank email to [EMAIL PROTECTED]
