From: John Sorkin <jsorkin_at_grecc.umaryland.edu>

Date: Thu 12 May 2005 - 14:50:43 EST

R-help@stat.math.ethz.ch mailing list

https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide!

http://www.R-project.org/posting-guide.html

R-help@stat.math.ethz.ch mailing list

https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html Received on Thu May 12 16:33:42 2005

Date: Thu 12 May 2005 - 14:50:43 EST

Darren is of course correct, but I hope the following, brief,
intentionally non-technical explanation will help:
Repeated measures analyses are needed only when there are three or more
measurements from a given experimental unit and the two or more
measurements are both on the right hand side of the equals sign, i.e.
are independent variables. When there are only two observations, any one
of the following models can be used, none of which have two measurements
from the same experimental unit as independent variables.:

change (i.e. post-pre)=pre

post=pre

You will note that in both models only one observation from a given
subject is on the right hand side of the equals sign, i.e. pre. When you
have three observations from a given subject you generally need to have
two or more observations on the right and side of the equals sign
(unless you are doing something like a Markov Chain, but Markov Chains
are beyond the scope of this Email.) and so need to consider repeated
measures techniques.

John

John Sorkin M.D., Ph.D.

Chief, Biostatistics and Informatics

Baltimore VA Medical Center GRECC and

University of Maryland School of Medicine Claude Pepper OAIC

University of Maryland School of Medicine
Division of Gerontology

Baltimore VA Medical Center

10 North Greene Street

**GRECC (BT/18/GR)
**

Baltimore, MD 21201-1524

410-605-7119

**- NOTE NEW EMAIL ADDRESS:
**

jsorkin@grecc.umaryland.edu

>>> Darren Weber <darrenleeweber@gmail.com> 5/11/2005 6:40:09 PM >>>

With respect to calculating the epsilon index of sphericity for ANOVA,

discussed on pp. 45-47 of:

http://www.psych.upenn.edu/~baron/rpsych.pdf

It notes that epsilon is not required for a repeated measures design
with

only k=2 levels, as the minimum value of epsilon (e) is given by:

e = 1/(k-1)

so for k=2, we have e = 1 (ie, no correction of the F test df; see p.
46).

These notes apply to a univariate F test.

How do we estimate the minimum value of epsilon for a 2 factor ANOVA?

I have an experiment where we measure brain activity from the left and
right

hemisphere, for two experimental conditions, in each subject. I
consider the

measures from each hemisphere a repeated measures factor (2 levels) and
the

experimental conditions is also a repeated measure (2 levels). The
question

now is, how do we calculate epsilon for this 2 factor study and is it
possible that epsilon could be anything < 1 when each factor has only 2

levels?

[[alternative HTML version deleted]]

R-help@stat.math.ethz.ch mailing list

https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide!

http://www.R-project.org/posting-guide.html

[[alternative HTML version deleted]]

R-help@stat.math.ethz.ch mailing list

https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html Received on Thu May 12 16:33:42 2005

*
This archive was generated by hypermail 2.1.8
: Fri 03 Mar 2006 - 03:31:42 EST
*