http://spssx-discussion.165.s1.nabble.com/Advice-tp5430896p5431078.html
More clarification, please.
If I understand the table of data, you have about 5 faculty ratings
(4 to 6) and 25 (24 or 25) student ratings on each of 10 scenarios.
Is that correct?
Unless you have (what seems unlikely) 50 *different* faculty
members, and 247 *different* students, your suggestion of an
overall t-test is wrong. Proper testing will have to account for
each set of ratings done by each person.
An unbalanced ANOVA across scenes, identifying IDs, would
test whether the Attendings regularly scored higher or lower
than the Residents. That would not examine whether Residents
might be more varying in their responses.
Art also asked, "What hypotheses are you interested in?"
--
Rich Ulrich
Date: Wed, 25 Jan 2012 13:43:04 -0500
From:
[hidden email]Subject: Re: Advice - Follow-up
To:
[hidden email]
Advice
This is what an exploration of the data shows -- I'd thought I
could do ANOVA to test differences between faculty &
learners on each scenario - but my sense is that the numbers are too small to do
anything more than a global t-test? Am I on the mark? Thanks!
Jennifer
[snip, table; previous]