Precision and linearity

[K. Shannon]

Bart_Corsaro_at_USLRMG01 at INTERNETMAIL.PR.CYANAMID.COM wrote:
> 
> I agree.  However, I think you are all wrong.  The best way to draw a straight
> line is with one point.  Just think how easy it is to fit all experimental data
> to that line. And as long as you are working with only one unknown at a time,
> you can always get a correlation coefficient of 1.0.  Now that's the kind of
> data for grants.
> 
> Bart Corsaro


It is theoretically impossible to get an r squared value of 1.  The only reason 
that it may happen from time to time on a standard curve when performing 
linear regression is due to computer neglection of error.  I f one should 
disagree, they might check out a book on entropy.
-- 
++++++++++++++++++++++++++
Kevin Shannon,
kmshannon at uoknor.edu
shannon.kevin at oklahoma.va.gov
VAMC Microbiology Lab,
Oklahoma City, OK

"in situ  de viri vivescere "
"Viruses survive in nature."
++++++++++++++++++++++++++