Thread of Original Message:
----------------------------------------------------------------------------
-----
----- Original Message -----
From: Jan de Leeuw <[EMAIL PROTECTED]>
To: David A. Heiser <[EMAIL PROTECTED]>
Sent: Wednesday, December 29, 1999 6:56 PM
Subject: Re: Question: Bivariate Regression
> http://www.amazon.com/exec/obidos/ASIN/0965336239/qid=946522539/sr=1-2
> /102-0639577-4869610
>
> This is a photocopy of the first edition, sells for 34.50 at Amazon.
>
> --- Jan
>
> >----- Original Message -----
> >From: Jan de Leeuw <[EMAIL PROTECTED]>
> >To: Dan Ryan <[EMAIL PROTECTED]>; <[EMAIL PROTECTED]>
> >Sent: Wednesday, December 29, 1999 9:26 AM
> >Subject: Re: Question: Bivariate Regression
> >
> >
> > > At 10:22 AM -0500 12/29/99, Dan Ryan wrote:
> > > >I am hoping someone can point me in the right direction.
> > > >
> > > >I am interested in finding literature on about a bivariate
regression
> >where
> > > >the parameters of the bivariate distribution are estimated.
> > > >
> > > >If anyone has any ideas, they would be greatly appreciated.
> > > >
> > > >Best Wishes for the New Year...
> > > >
> > > >Dan
> > > >
> > > >Daniel A.J. Ryan
> > > >"You have to kick at the darkness
> > > >Research Statistician
> > > >till it bleeds daylight"
> > > >Atlantic Food and Horticulture Research Centre
> > > >Bruce Cockburn
> > > >Agriculture and Agri-Food Canada
> > > >Phone: (902) 679-5599
> > > >Fax: (902) 679-2311
> > > >Email: [EMAIL PROTECTED]
> > >
> > > F. Galton
> > > Natural Inheritance
> > > London, MacMillan and Co, 1889
> > > No ISBN
> > >
> > > ===
> > > Jan de Leeuw; Professor and Chair, UCLA Department of Statistics;
> > > US mail: 8142 Math Sciences Bldg, Box 951554, Los Angeles, CA
90095-1554
> > > phone (310)-825-9550; fax (310)-206-5658; email:
[EMAIL PROTECTED]
> > > http://www.stat.ucla.edu/~deleeuw and
http://home1.gte.net/datamine/
> > >
----------------------------------------------------------------------------
-------------------------------------------
Got the book and am reading it. Galton writes in a really good
understandable style. Very direct, straightforward and refreshing. His words
and structure really is very contemporary, considering it was from England
in the 1880's. Quite different from the ponderousness of R.A.Fisher's
writtings. However like all things it represents what was known in the
1880's. There are a lot of other things too,
His observation on page 100 (and the work of Hamiton Dickson in appendix B)
leads to the bivariate distribution. However it is not explicitly presented
here. You have to look at Dickson's equation 4 and "see" the bivariate
distribution in it. His description does not have that clarity that one
would like to have from a discoveror, where the entire concept is shown in
blinding light (ureka!).
That appears later. My earliest encounter is in Fisher's book, where he
takes Galton's data (Statistical Methods, page 46 and table 2), but does use
it to illustrate the bivariate distribution. In chapter VI, Fisher uses
other stature data and uses it to show correlations and the bivariate
normal distribution. I would guess that the form evolved during the 35 years
between Galton's work and Fisher's book. (Note that K. Pearson and R.A.
Fisher worked in the lab that Galton started)
I would suspect that there are many contemporary stat books that do a very
good job of explaining the bivariate distribution, and why the statisitics
are so complicated.
There are a lot of good quotes in the book.
"Fortunately for us, our ignorance of the subject will not introduce any
special difficulty in the inquiry on which we are now engaged." (page
16)(How any thesis' would be accepted if they began with this statement?)
"A serious complexity due to sexual differences seems to await us at
every step when investigating the problems of heredity. Fortunately we are
able to evade it altogether by using an artifice at the outset..... The
artifice is never to deal with female measures as they are observed, but to
employ their male equivalents in place of them." (page 6) (It is basically
about human height, but it still makes a great quote)
For the structural equation modlers (the SEMer's), Galton brings out
latent variables in chapter II.
"It is the triteness of these experiences that makes the most varied
life monotonous afte a time, and many olf men as well as Solomon have
frequent occaision to lament t6hat there is nothing new under the sun."
"I had to collect all my data for myself, as nothing existed, so far as
I know, that would satisfy even my primary requirement." (page 71) (Very
interesting on how he went about getting the stature data in chapter VI.
Just like we do today, go out and pay people (or give gifts) to have them
give you personal information. He even has quality control on his data. He
gives the names of all those who gave him their data! Impressive.)
(Regards the inheritance of the artistic faculty, page 155). "I began by
dividing my data into four classes of aptitudes; the first was for music
alone; the second for drawing alone; the third for both music and drawing;
and the fourth includes all those whose artistic capacities a discrete
silence was observed. After prefatory trials, I found it so difficult to
separate aptitude for music from aptitude for drawing, that I determined the
throw the first three classes into the sisangle group of artistic. This and
the group of the Nonartistic are the only two divisions now to be
considered." (Nowdays the SEMers would try and go into great detail and
attemts to develop many latent endogenous variables that relate to
"artistic".)
"Also, because from such calculations as I have made, the hereditary
influences of the two sexes in respect to art appear to be pretty equal: as
they are inrespect to nearly every other characteristic, exclusive of
diseases, that I have examined." (page 158)
On page 107, he has a really neat analog computer to forecast stature.
Another interesting side, would be to take his tables of "Temper in
Marriage" and "Temper in Families" (page 230 and 231), do a SEM and do some
alternate stat, and compare your conclusions with his.
DAHeiser
