[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: -listcoef- and returning results
At 12:33 AM 9/14/2006, you wrote:
Date: Wed, 13 Sep 2006 15:57:04 -0500
From: Richard Williams <Richard.A.Williams.5@ND.edu>
Subject: Re: st: logistic regression and outreg: how to display
standardized factor change
I'll add that Long & Freese's listcoef does not return or ereturn its
results. Ergo, you will have to program the calculations
yourself. fitstat does return the estimated variance of y*, which is
needed for some of the standardizations done by listcoef.
I'm reluctant to post here where I am somewhat ignorant, but here's a
thought that could be passed on: Given that -listcoef- has done all
the heaving lifting, couldn't one just modify it with an rclass
designation, and stick -return scalar- (or matrix) at the end of the
code? This of interest to me because I am doing a bunch of work on
logreg models for student retention, and I'm wanting to do lots of
standardized comparisons (on y) so as to have meaningful comparisons
of effects across models?
Incidentally, I am starting to look toward the MSS/N. Central
meetings in Chicago next spring, and hoping to have a good quant
methods session. (I'm not organizing, but trying to help out.) If
you and/or any students might be interested, that would be good.
Just got a email recently from Matt Larner, referred by Mike Davern,
because Matt had moved to Colorado and was looking for professional
work. I think we found some good leads for him, at least I hope
so. He looked like a competent guy.
Mike Lacy, Assoc. Prof.
Soc. Dept., Colo. State. Univ.
Fort Collins CO 80523 USA
* For searches and help try: