Hi, FYI..
there is fast algorithm ... "lstand"...that computes beta weights, as well as
unstandardized coeffs and odd rations..it only applies to the logistic command.
/emmanuel
=============================================
Emmanuel F. Koku
[email protected]
Ph.D Candidate and Research Associate
Centre for Urban and Community Studies
University of Toronto
455 Spadina Avenue
Toronto, Canada M5S 2G8
Tel: 1(443)465-4653
Fax: 1(208)977-0232
========================================
Quoting [email protected]:
Emmanuel:
Perhaps you can mention to the Statalist that you found the answer in lstand.
For some reason I cannot send a message, or reply, to Statalist. It may help
others as well.
Thanks, joe Hilbe
In a message dated 11/21/2004 11:21:43 AM US Mountain Standard Time,
[email protected] writes:
Joe,
the Lstand did work...I think i was getting no observation error because
during
conversion from spss, stata mistakingly recorded some of my numeric variables
as
string. I got an "destring" ado that was able to convert the data to
numeric
format, and thereafter, everything worked fine.
I must say, I am very impressed with this program, and wondered why i had
stayed
with sas/spss for so long ... it has tons of features, and the user-produced
algorithms/extensions is a plus. I'm using version 6 at the moment - my
student
budget cannot afford version 8 at this time, but i can't wait to get a hold
of
one and take check it out.
thanks for pointing me to lstand..it does what i wanted, and i've already
produced my table. One question though about building hierarchical models..
how
can i check the chi-sq test for the full and restricted models (i.e, if i
want
to find out if it would be statistically significant to include a set of
interaction terms..in a fuller model, do i go by the value of LR chi^2 (df)
and
Prob> chi^2 values printed by the logistic command?
sincerely, Emmanuel
=============================================
Emmanuel F. Koku
[email protected]
Ph.D Candidate and Research Associate
Centre for Urban and Community Studies
University of Toronto
455 Spadina Avenue
Toronto, Canada M5S 2G8
Tel: 1(443)465-4653
Fax: 1(208)977-0232
========================================
Quoting [email protected]:
Emmanuel:
I just tried running lstand following a logistic regression. See the output
below. It does exactly what I intended. Perhaps your copy is not good, or
(the
likely reason) when you copied it there were lines that were too long and
got wrapped. These have to be manually corrected. I'll send you my copy if
you
still have problems.
Best, Joe
========================
. use auto
. logistic foreign mpg weight length turn trunk, nolog
Logistic regression Number of obs =
74
LR chi2(5) =
43.42
Prob > chi2 =
0.0000
Log likelihood = -23.322345 Pseudo R2 =
0.4821
------------------------------------------------------------------------------
foreign | Odds Ratio Std. Err. z P>|z| [95% Conf.
Interval]
-------------+----------------------------------------------------------------
mpg | .8377264 .0814023 -1.82 0.068 .6924533
1.013477
weight | .9965262 .0021876 -1.59 0.113 .9922477
1.000823
length | 1.043064 .0714009 0.62 0.538 .9121028
1.19283
turn | .655959 .1120318 -2.47 0.014 .4693526
.9167567
trunk | 1.011491 .1220834 0.09 0.925 .7984084
1.281442
------------------------------------------------------------------------------
. lstand
Table of Predictor Estimates:
Standardized Coefficients and Partial Correlations
No. Var Coef OR St.Coef PartCorr Prob(z)
=======================================================================
0 Constant 20.8919
1 mpg -0.1771 0.8377 -0.5648 -0.1211 0.068
2 weight -0.0035 0.9965 -1.4911 -0.0755 0.113
3 length 0.0422 1.0431 0.5176 0.0000 0.538
4 turn -0.4217 0.6560 -1.0227 -0.2132 0.014
5 trunk 0.0114 1.0115 0.0269 0.0000 0.925
=======================================================================
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
