🚀 go-pugleaf

Geoff Selig has hit the nail on the head. The
notion of building models that predict data is central in
statistics; and with the current vast increase in EDA packages,
we should be shifting from teaching about 'descriptive stats'
to data models - helping people to look for patterns in data.
The mean isn't just a descriptor, it's a predictor.
Geoff's description of degrees of freedom in terms of model
building moves df from the realm of superstition to the realm of
the understandable - sure, the concept if hard, but it's also
vital!
I have to admit a major influence, that of Judd and McClelland's
excellent book "Data Analysis - A Model Comparison Approach"
(Harcourt Brace Jovanovich, 1989 ISBN 0-15-516765-0). Their
account of degrees of freedom in terms of model-building parameters
(pp 86-7) is a breath of fresh air on the subject.


R M Conroy, B.Atchelor (my own degree of freedom...)

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ "I was sad because I had no shoes until I met a man with no +
+  feet. 'Great', I thought, 'He won't be needing his shoes.' +
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Ronan Conroy
Lecturer in computing and         Royal College of Surgeons in Ireland
biostatistics                     123, St Stephen's Green, Dublin 2.
                                  (353)-(1)-784422 [fax:780934]