Thanks again and your logic is reasonable. Indeed you describe the process
accurately. I will give your ideas a go and let you know how it turned out.
From: Peter Flom [mailto:flom@NDRI.ORG]
Sent: Wednesday, November 05, 2003 11:09 AM
Subject: Re: Transforming data
If the residuals are normally distributed, then there is no statistical
reason to transform the dependent variable. There may be substantive
reasons to do so, however. I know nothing about this substantive area,
but there are certainly many physical traits where a log scale is more
re mixed model
I'm not entirely sure I understand, but I think what you're saying is
the design was something like
baseline (measure both eyes)
surgery on eye number 1
surgery on eye number 2
If this is the case, then I think you DO need a mixed model. You also
need to have some way of representing whether each eye had had surgery.
How to do this is a substantive question, but one idea is to simply add
a variable (for each eye, at each time) 'postsurgery'; if all the people
eventually got surgery on both eyes, then you could eliminate the time
This may not be ideal, but I do not see any way to avoid a mixed model;
as I said, I don't know the substantive area,
but I woud be amazed if there were not some effect of the PERSON as
well as the EYE.
Peter L. Flom, PhD
Assistant Director, Statistics and Data Analysis Core
Center for Drug Use and HIV Research
National Development and Research Institutes
71 W. 23rd St
New York, NY 10010
(212) 845-4485 (voice)
(917) 438-0894 (fax)
>>> Thompson Bill T Contr USAFSAM/FEC <Bill.Thompson@BROOKS.AF.MIL>
11/5/2003 11:55:46 AM >>>
You are correct in your assumptions and I will follow your lead
the residuals, skew and kurtosis. If the residuals are "normally"
distributed would you suggest NOT transforming the data?
One issue that has come up regarding "eye" is that these are "fellow"
from the same person (humans). Since the results are "highly"
baseline it was suggested we just pool the eyes and eliminate that
Unfortunately, subjects received surgery on one eye at a time with the
possibility of several weeks between surgeries. As a result the
data is not as highly correlated because of different surgical outcomes
different eyes. Hence, we are trying to decide if we should analyze
eye separately or use a mixed model.