How to estimate representative point in non-ellipsoidal multivari ate data

["Byers, Bob" <rhb1@cdc.gov>]

Dear listers,

I have estimated 3 parameters 1000 times using bootstrap samples.
With the brush() function I can see that the triplet formed by
the means is not near any realized points. This is because
the cloud of points is shaped like a spoon.
 I would like to estimate a triplet that is representative of
the population, but how?
Thanks,
Bob 

Robert H. Byers, Jr. PhD
Statistics & Data Mgmt.
Div. of HIV-AIDS Prevention, CDC
bbyers@cdc.gov
(404) 639-2025