give hyperstat to people and post hoc tests

I have a query regarding post hoc tests.
When we carry out, say, a 2 way mixed anova and we come up with a significant interaction, how do we go about exploring it further?
Can we use Confidence Intervals for all factors? I mean, it seems that they apply better to between - Ss factors, right?
Is it preferable to perform paired t tests for the within-Ss factor and independent t tests for the between-Ss factor? Is there any rule of thumb that indicates what you do when you need to analyze an interaction?

One last thing, which, in fact, is what worries me most... I've been trying in vain to log onto Hyperstat for a week!
Thanks to it, unlike most of my classmates, when it came to stats, I never felt like crying, leaving college and work the land, or hated myself, life and mankind. So, please somebody tell me what's going on?!!!

Thanks to you all!!!!



TS Contributor
When doing post-hoc tests for a significant interaction in a repeated-measures ANOVA, use either:

(1) paired t-tests and a Bonferroni correction for the within factors

(2) or Games-Howell

In terms of "rules of thumb" for testing interactions - if your hypotheses suggest an interaction(s) exists, then don't worry about increasing Type I error rates - use a powerful test. If you're just doing a bunch of post-hoc tests (unplanned), then be conservative.

As for hyperstat, I was able to get there a few minutes ago:

-although they took down our links without telling us, so we're not too happy with them.....but they do cover this topic as well:
Honestly thank you

Thank you very much.

Helping out somebody that you don't know is a big thing.

Ps. I can't figure out why I can't access hyperstat. Crazy thing is that I am able to connect to the home page through some other web address, other than the one you gave me, and every link you provide there works just fine, but still I can't access any of the contents of the book. Anyway, I have most of the chapters saved!:)