A Repeated measures
ANOVA in StatView requires that the data be organized into a compact variable,
i.e. each row is a subject and you have one column for each level of each
factor. Recall that the factorial ANOVA can be done with either file
organization; this is not the case for the RM design.

Therefore,
if necessary you should take the time to review making compact variables,
e.g. using the Wine sample file, which has one factor with 7 levels.
(Though note that we will see later that there is a problem with doing
a RM ANOVA on such a dataset.)

When you
have an appropriate file with a compact variable open, go to Analyze - ANOVA
and t-tests - Repeated Measures ANOVA. The dialog box opens confusingly,
with the "between factor" boxhighlighted, ready for your entry. But
you do not have to have any between-subjects factors to run the analysis.
Instead ignore that box and just drag your "compact" variable into the "Repeated
measurement" box underneath, and run your analysis. Unlike in a factorial
analysis, you do not open up the compact variable. (Do
not put Subjects into the Between box, else the analysis will not run.)

Try the following
example:

-first do a 2-way factorial analysis using my file factorial.xls (so, 2 between-subjects factors, assuming 24 different subjects)

-then do the same
ANOVA, using the same data, but in a compact variable, in my file compact.xls
(the result should be exactly the same, right?)

I
hope this isn’t confusing to use the same data file for the two analyses
– obviously if these were real data, only one of these analyses would be
appropriate and the other not. This is just to show the mechanics of
doing these two kinds of analyses, and to compare the results.

result of RM:

result of factorial:

>>The
error terms, and the way they are displayed in the table (and also the associated
degrees of freedom) are different.

You'll see that
the p values are generally lower in the second table. If you do a factorial
analysis on RM data, then you are overestimating the significance of the
effects.

MIXED DESIGNS

If
some factors are within-subjects and some are between-subjects, then you
use the RM option. Your data file should have any Within
factors in a compact variable (these will be the columns) and any Between
factors, plus subjects, as the rows. Drag factors which
are not part of your compact variable into the Between box, and drag your
compact variable into the Repeated measurements box. The
StatView sample file Teaching Effectiveness is a straightforward mixed design,
with Time as the Within factor. Also try my files mixed.svd

and mixed2.svd.

* last updated
Dec. 2006 by P. Keating*

Back to the UCLA Phonetics Lab overall facilities page