In article <[EMAIL PROTECTED]>,
Kai Arzheimer <[EMAIL PROTECTED]> wrote:
>[EMAIL PROTECTED] (Rodney Carr) writes:
>> The problem I am having is that I'm not sure what estimating method
>> to use. EQS implements a number of different methods (Maximum
>> Likelihood, Least Squares, GLS, etc). Unfortunately they give quite
>> different results. Actually, LS gives fit indices that are fairly
>> high, but none of the others do (so I'd like to use the LS method!).
>> But I can't find any references that explain which method should be
>> used. Please, do you have any ideas for where I might look for
>> advice?
I did not notice the earlier article. The question is what
is wanted, and why.
If one wants to come up with estimates of quantities based
on current values of other quantities, least squares and
related methods are quite appropriate. If one wants to
understand what is happening structurally, least squares
is likely to give excessively high fits.
A VERY old example is that of estimating the consumption
function, C = \alpha + \beta * Y + error, Y being income.
Now if one wants to come up with an estimate of this year's
consumption from this year's income under unchanged
conditions, least squares is fine. But if one wants to
estimate the effect of a government making grants to people,
the structural value of \beta, not the regression value of
the LS coefficient \gamma, is what is wanted.
--
This address is for information only. I do not claim that these views
are those of the Statistics Department or of Purdue University.
Herman Rubin, Dept. of Statistics, Purdue Univ., West Lafayette IN47907-1399
[EMAIL PROTECTED] Phone: (765)494-6054 FAX: (765)494-0558
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
