On 6 Sep 2003 04:48:13 GMT, Eric Bohlman <[EMAIL PROTECTED]>
wrote:

> "David Heiser" <[EMAIL PROTECTED]> wrote in
> news:[EMAIL PROTECTED]: 
> >   Our term "regression" for fitting an equation to data, comes from
> >   his observation that the heights of offspring is on the average less
> >   than the heights of either parent. In other words, there is a
> >   regression of the parents height. 
> 
> Not "less than."  Less *extreme* than, i.e. closer to the mean.  Less in 
> absolute value if you mean-center the heights.  Taller-than-average or 
> shorter-than-average parents are likely to have children with closer-to-
> average heights, *not* children whose heights are correspondingly on the 
> other side of the mean, e.g. taller-than-average parents are *not* 
> especially likely to have shorter-than-average children.  In fact their 
> children are likely to be taller-than-average as well, just not by as much 
> as their parents.

And, of course that fruitful paradox always 
should be mentioned:  The converse also is true.

While selected parents have children whose heights are 
less extreme on the average, selected children have parents 
whose heights are less extreme on the average  

-- 
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
"Taxes are the price we pay for civilization." 
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to