For the two binary responses, if you would settle for a probit link
function then Stata's bivariate probit command -biprobit- should give
you what you need to make the comparison, allowing for dependence
between mstair_1 and mstair2.
. biprobit h1 h2 = x5
Bivariate probit regression Number of obs =
15
Wald chi2(2) =
1.75
Log likelihood = -16.681705 Prob > chi2 =
0.4170
------------------------------------------------------------------------
------
| Coef. Std. Err. z P>|z| [95% Conf.
Interval]
-------------+----------------------------------------------------------
------
h1 |
x5 | .0459239 .0417967 1.10 0.272 -.0359961
.1278439
_cons | -1.271098 1.134072 -1.12 0.262 -3.493839
.9516428
-------------+----------------------------------------------------------
------
h2 |
x5 | -.0239751 .0509927 -0.47 0.638 -.1239191
.0759688
_cons | -.241399 1.313751 -0.18 0.854 -2.816303
2.333505
-------------+----------------------------------------------------------
------
/athrho | .5238399 .5659559 0.93 0.355 -.5854134
1.633093
-------------+----------------------------------------------------------
------
rho | .4806582 .4352019 -.5265889
.9265008
------------------------------------------------------------------------
------
Likelihood-ratio test of rho=0: chi2(1) = .984849 Prob > chi2 =
0.3210
. test [h1]x5=[h2]x5
( 1) [h1]x5 - [h2]x5 = 0
chi2( 1) = 1.44
Prob > chi2 = 0.2298
Al Feiveson
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Sara Mottram
Sent: Wednesday, September 13, 2006 10:38 AM
To: [email protected]
Subject: st: Testing equality of coefficients in different logistic
models
I have an ordinal response with three levels and I am hoping eventually
to fit an ordinal regression model, possible a partial proportional odds
model using -gologit2-. However, to explore my data and get an idea of
where I may and may not have proportionality of odds, I have created two
dichotomies and would like to fit two binary logistic regression models,
testing for the equality of the slope coefficients, much as the
-autofit- option does in -gologit2-.
i.e. in the output below, is 1.86 equal to 1.96?
. logistic mstair_1 edu
------------------------------------------------------------------------
------
mstair_1 | Odds Ratio Std. Err. z P>|z| [95% Conf.
Interval]
-------------+----------------------------------------------------------
-------------+------
edu | 1.861951 .0870595 13.29 0.000 1.698903
2.040648
------------------------------------------------------------------------
------
. logistic mstair_2 edu
------------------------------------------------------------------------
------
mstair_2 | Odds Ratio Std. Err. z P>|z| [95% Conf.
Interval]
-------------+----------------------------------------------------------
-------------+------
edu | 1.960462 .1124576 11.74 0.000 1.751988
2.193743
------------------------------------------------------------------------
------
I suspect that there is an Stata command to carry out this test, but I
have been unable to find it, either in the manuals or in the FAQ. From
the manual, I get the impression that the Wald test performed by -test-
can compare coefficients within the same model or compare all of the
coefficients in two models?
I would also value opinions on whether there is any value in performing
two binary regressions, given that -gologit2- has an automatic fitting
procedure in -autofit-. (I am basing my analysis plan on a fairly out of
date paper, which suggests that binary regressions are the only way to
choose where non-proportionality is needed as there were no programs
able to do this at the time).
Best wishes
Sara
--
Sara Mottram
Research Assistant: Biostatistics
Primary Care Musculoskeletal Research Centre Primary Care Sciences Keele
University Staffordshire, ST5 5BG
Tel: 01782 584711
Fax: 01782 583911
*
* 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/
*
* 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/