Post hocs in repeated measures?
Dave Voelker
Anyone know a way to do post hocs with a repeated measures ANOVA?
I've got a dataset where each subject rated six objects, and I want to
know which pairs of objects are significantly different. In a simple
ANOVA, I'd run Scheffes to do this, but that's not an option for SPSS's
repeated measures procedure. The built-in orthonormalized contrasts won't
do. Anyone know any other options?
--Dave Voelker
dvoe...@inav.net
Steven J. Pierce
Dave Voelker wrote:
> Anyone know a way to do post hocs with a repeated measures ANOVA?
Why not just take the appropriate error term from the ANOVA table and do
the post hoc tests by hand or in a spreadsheet?
--
Steven J. Pierce
Master's Student in Experimental Psychology
Mississippi State University
E-Mail: sj...@ra.msstate.edu
WWW: http://www2.msstate.edu/~sjp2/index.html
David Nichols
In article <3248DB...@ra.msstate.edu>,
Doing this is very dangerous. If the sphericity assumption is even slightly
violated, the coverage levels and power of individual comparisons is badly
compromised. Unless you're certain about sphericity (in which case you can
take summary data from the ANOVA table and descriptive statistics and use
MATRIX DATA to input the information into ONEWAY), you should use what are
equivalent to paired t-tests, generally with some sort of a general
probability based correction. I have written a set of macros implenting
this method for the one way repeated measures case:
The macros are available for downloading from our FTP site
(ftp.spss.com). They can be accessed via traditional anonymous
FTP, in the directory pub/spss/statistics/nichols/macros. See
the Index file for a listing of available macros. A Web browser
can also be used to access the FTP site, or to access the macros
from our Web site (www.spss.com). Once you reach the SPSS Web
site, click on Support, then under Ftp Archives, click on Macros
and scroll until you find the ones of interest to you. All files
are ASCII text, and each macro or set of macros is accompanied
by a readme file.
There are versions for LSD (uncorrected), Bonferroni and Dunn-Sidak
(corrected) comparisons. If you don't want to mess around with macros,
use T-TEST for each unique pair, and cut your alpha down according to
your favorite version (e.g., Bonferroni).
--
-----------------------------------------------------------------------------
David Nichols Senior Support Statistician SPSS, Inc.
Phone: (312) 329-3684 Internet: nic...@spss.com Fax: (312) 329-3668
-----------------------------------------------------------------------------