On 13 Aug 2003 05:47:50 -0700, [EMAIL PROTECTED] (Donald Burrill) wrote:
> I'm not familiar with the "Chow test". But your problem is logically
> equivalent to fitting the model
> [ snip, rest]
"Chow test" is the name sometimes used for the usual test
for nested models. I learned that name from these groups.
I've used that name a number of times, assuming it was standard -
but I recently noticed that it was not so much a standard as
I had assumed. Now I note that DB, who is widely read,
doesn't recognize that name.
Google in Groups will show that it is not infrequent, but
I'm one of the most frequent people citing it, in the Usenet
groups.
============ from google.
From: Alexander Tsyplakov ([EMAIL PROTECTED])
Subject: Re: comparing two betas
View: Complete Thread (4 articles)
Original Format
Newsgroups: sci.stat.edu
Date: 2000-10-09 08:53:33 PST
Let me quote from my own old message sent to sci.stat.math:
The appropriate method is given by using dummy (0 / 1)
variables in regression.
If added constructed variables are not significant then
coefficients are the same (statistically).
If you test for *all* coefficients then this method will
give you Chow test.
Some references:
Pagan, A.R., and D.F.Nicholls. "Estimating Predictions,
Prediction Errors
and Their Standard Deviations Using Constructed Variables,"
Journal of Econometrics, 24 (1984), 293-310.
Chow, G.C. "Tests of Equality between Sets of Coefficients
in Two Linear Regressions," Econometrica, 28 (1960),
591-605.
A. Tsyplakov
Novosibirsk State University
===========
--
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
"Taxes are the price we pay for civilization." Justice Holmes.
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
