Logistic Regression: A Self-learning Text, Third Edition (Statistics in the Health Sciences)

Theoroption can be used to obtain exponentiated regression parameter estimates.
The code follows (output omitted):

clogit mi smk sbp ecg, strata (match) or

Polytomous Logistic Regression

Polytomous logistic regression is demonstrated with the cancer dataset using the
mlogitcommand.

The outcome variable is SUBTYPE, a three-category outcome indicating whether
the subject’s histological subtype is Adenocarcinoma (coded 0), Adenosquamous
(coded 1), or Other (coded 2). The model is restated as follows:

ln

PðSUBTYPE¼gjXÞ

PðSUBTYPE¼ 0 jXÞ



¼agþbg 1 AGEþbg 2 ESTROGENþbg 3 SMOKING

whereg¼ 1 ; 2

Open the datasetcancer.dta. The code to run the polytomous logistic regression
follows:

mlogit subtype age estrogen smoking

Stata treats the outcome level that is coded zero as the reference group. The output
follows:

Multinomial regression Number of obs ¼ 286
LR chi2(6) ¼ 18.22
Prob>chi2 ¼ 0.0057
Log likelihood¼247.20254 Pseudo R2 ¼ 0.0355

subtype Coef. Std. Err. z P>jzj [95% Conf. Interval]

1
age .9870592 .4117898 2.40 0.017 .179966 1.794152
estrogen .6438991 .3435607 1.87 0.061 1.317266 .0294674
smoking .8894643 .5253481 1.69 0.090 .140199 1.919128
_cons 1.88218 .4024812 4.68 0.000 2.671029 1.093331

2
age .2822856 .3279659 0.86 0.389 .3605158 .925087
estrogen .1070862 .3067396 0.35 0.727 .7082847 .4941123
smoking 1.791312 1.046477 1.71 0.087 3.842369 .259746
_cons 1.203216 .3189758 3.77 0.000 1.828397 .5780355

(Outcome subtype¼¼0 is the comparison group)

STATA 657