Thank you very much, Svend, Michael, John, and Clive.
At least I can take solace in knowing that the question is not a trival one.
It looks to me like the residual variation is perhaps borderline (not
completeley sure what to do with that). And I have a little bit of trouble
reconciling what -mhodds- gives me (see below) as compared to doing
-by romi: logistic outcome zlog-
And I have a very interesting looking graph produced (slowly) by -inteff-
which I'd be happy to pass on off-list to anyone interested.
Vibl comes next, as does ordering my copy of the stata journal...
(if anyone is getting a big picture idea for how to describe this particular
interaction in a way that doesn't make a reviewer for a cardiolgy journal
have a seizure, I'd be most greatful)
*****
. complogit is_dead zlog, group(romi)
Number of obs =
20449
Wald chi2(2) =
334.45
Log likelihood = -3759.6533 Prob > chi2 =
0.0000
----------------------------------------------------------------------------
--
is_dead | Coef. Std. Err. z P>|z| [95% Conf.
Interval]
-------------+--------------------------------------------------------------
--
eq1 |
zlog | .8506914 .0536334 15.86 0.000 .7455718
.955811
romi | -.8017628 .3580006 -2.24 0.025 -1.503431
-.1000946
_cons | -1.667314 .0683411 -24.40 0.000 -1.80126
-1.533368
-------------+--------------------------------------------------------------
--
delta |
romi | .3088438 .1507395 2.05 0.040 .0133998
.6042878
----------------------------------------------------------------------------
--
(...output omitted...)
residual variation using -complogit-
----------------------------------------------------------------------------
----
Likelihood ratio test to reject null hypothesis of equal residual variation
----------------------------------------------------------------------------
----
Likelihood-ratio test LR chi2(1) =
4.75
(Assumption: two nested in allison_1) Prob > chi2 =
0.0293
. mhodds is_dead zlog
Score test for trend of odds with zlog
(The Odds Ratio estimate is an approximation to the odds ratio
for a one unit increase in zlog)
----------------------------------------------------------------
Odds Ratio chi2(1) P>chi2 [95% Conf. Interval]
----------------------------------------------------------------
2.538749 289.60 0.0000 2.280442 2.826315
----------------------------------------------------------------
. mhodds is_dead zlog ,by(romi) , if $bi | $slr
Score test for trend of odds with zlog
by romi
(The Odds Ratio estimate is an approximation to the odds ratio
for a one unit increase in zlog)
----------------------------------------------------------------------------
---
romi | Odds Ratio chi2(1) P>chi2 [95% Conf.
Interval]
----------+-----------------------------------------------------------------
---
0 | 2.789601 264.14 0.0000 2.46497
3.15699
1 | 6.212236 133.33 0.0000 4.55617
8.47025
----------------------------------------------------------------------------
---
Mantel-Haenszel estimate controlling for romi
----------------------------------------------------------------
Odds Ratio chi2(1) P>chi2 [95% Conf. Interval]
----------------------------------------------------------------
3.113895 375.37 0.0000 2.775876 3.493075
----------------------------------------------------------------
Test of homogeneity of ORs (approx): chi2(1) = 22.10
Pr>chi2 = 0.0000
.
*
* 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/
