On 14 Mar 2000 20:28:17 GMT, Ron Bloom <[EMAIL PROTECTED]>
wrote:
>
> I think I made the formulation too wordy in previous
> post.
>
> Let me try this simple question:
>
> When one wishes to do a (multi)linear regression on a set of
> observed data, and one is in the (unusual) position of possessing
> a set of sample standard deviations (of varying degrees of f.)
> at each value of the "explanatory" variable, how does one
> determine whether one ought or ought not to solve the weighted
> least squares problem using those sample standard deviations?
>
> What is the usual decision test for "heterscedasticity" *before* one
> solves the regression system? What do people do in practise?
- If weighting says that, mostly, ONE sample will matter, then you
have to decide if you want to effectively ignore all the rest. Or,
how you will weight. Or else, you can be ignorant, and never notice
that you have made that your weighting makes a choice. I have seen a
meta-analysis that pretended to combine one study with N=2000 and
several studies with Ns mostly under 50, and the manner of combining
pretended each subject was equal. So, only one sample mattered.
The easiest solution, if it works out, is to figure your answer and
reach your happy conclusion in several ways, so that you can show that
the conclusion is robust, and the choice of analysis does not matter.
Heterogeneity. Outliers. Flaw in the data collection? or valuable
datum, or crucial insight into the structure of the problem?
In biostatistics, our collaborators have enough trouble understanding
regular multiple regression, and we statisticians worry about keeping
the clinicians away from the temptations of "stepwise". I have never
faced regression "weighting" for real data; I would try to clean up my
data or justify it, instead of weighting. But there are other areas
(certain survey research) where the use of weighting -- for various
reasons that probably do not match yours -- is the standard option.
Maybe someone else will add to this....
--
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
===========================================================================
This list is open to everyone. Occasionally, less thoughtful
people send inappropriate messages. Please DO NOT COMPLAIN TO
THE POSTMASTER about these messages because the postmaster has no
way of controlling them, and excessive complaints will result in
termination of the list.
For information about this list, including information about the
problem of inappropriate messages and information about how to
unsubscribe, please see the web page at
http://jse.stat.ncsu.edu/
===========================================================================
