Hello.....I am wondering if there is a way to do this without
violating various assumptions about independence and variability.
I have run subjects in two versions of an experiment, with partially
overlapping conditions. In each, 30 subjects participated in all
conditions (fully crossed within design).
version 1 version 2
a
b
c c
d d
e
f
So in total, cell sizes are:
a-30
b-30
c-60
d-60
e-30
f-30
Besides the fact that a and b are more independent of e and f than a,
b, c, and d (or c, d, e, and f) are from each other, since different
subjects participated in 1 and 2, the resulting cell sizes are
different.
How would the result be in error in simply finding a mean for a, b, c,
d, e, and f and performing an ANOVA on these, assuming they can be
viewed as levels of a factor?
Thanks
Jim
--
Watch out for spam block
List of online jazz guitar instruction websites:
http://www-personal.umich.edu/~jimkk/jazzfast.html
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
