[S] Imputation strategy

Marc R. Feldesman (feldesmanm@pdx.edu)
Sat, 29 Aug 1998 19:06:08 -0700


This is *not* strictly an Splus question, but I am seeking opinions from
you on this topic.

I'm currently working with cranial measurements from 33 fossil "human"
skulls. These skulls, many quite incomplete, range in age from 200,000
years ago to about 40,000 years ago. Some are typically called
"Neanderthals" while others are called "Homo sapiens". All are formally
classified as Homo sapiens, but they are distinctly and uniquely different
from contemporary humans.

Many important measurements are missing from these specimens - probably
only 2 or 3 specimens have all 20 cranial measurements present and several
have only 10 measurements. To study the relationships of these specimens
to each other, and all (each) to modern humans, it would be useful to
"fatten" out the measurement set.

How to do this is the problem. Imputation and multiple imputation is the
obvious strategy, but approaching the problem in any systematic way is
problematic. I have a comparative modern human dataset with 2524 specimens
from 67 different populations upon which the full suite of 20 measurements
is available. If I try to impute the fossils' measurements from only their
own existing measurements, I don't have enough data (not to mention the
confounding effects of geography, time, and biology). If I try to impute
them from the modern human data, it isn't clear to me which groups to use
without biasing the results toward the very populations to which I would
later want to compare the fossils.

I've tried to simplify the issues for presentation here. Obviously the
problem is quite complicated, but I welcome any and all suggestions of how
to acquire the missing measurements. As with any study of fossils, I have
what I have. There isn't any more data and certainly no way to get more.

Thanks for allowing me to present this problem to you. Thanks in advance
for any ideas you might be willing to share.

Dr. Marc R. Feldesman
email: feldesmanm@pdx.edu
email: feldesman@ibm.net
pager: 503-870-2515
fax: 503-725-3905

"Don't know where I'm goin'
Don't like where I've been
There may be no exit
But, hell I'm goin' in" Jimmy Buffett

Powered by: Monstrochoerus - the 300 MHz Pentium II
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news