Dear Statalist,
I am running a logistic model where I have discovered interactions between:
1) Gender [male] and Age [age_65; dichotomous]; and
2) Angioplasty [idx_angio] and Age [age_65]
My confusion is in interpreting the simple effects of age at each level of
the interactions.
I have tried using [email protected]_angio with i.male*i.age_65 also present in
the model, but was not able to interpret the results.
For example:
. xi3: logistic outcome i.male*i.age_65 [email protected]_angio
i.male _Imale_0-1 (naturally coded; _Imale_0 omitted)
i.age_65 _Iage_65_0-1 (naturally coded; _Iage_65_0 omitted)
g.idx_angio _Iidx_angio_0-1 (naturally coded; _Iidx_angio_0
omitted)
note: _Iag1Wid1 dropped due to collinearity
Logistic regression Number of obs =
15877
LR chi2(5) =
403.64
Prob > chi2 =
0.0000
Log likelihood = -10093.725 Pseudo R2 =
0.0196
----------------------------------------------------------------------------
--
Outcome | Odds Ratio Std. Err. z P>|z| [95% Conf.
Interval]
-------------+--------------------------------------------------------------
--
_Imale_1 | 1.312062 .0609901 5.84 0.000 1.197807
1.437215
_Iage_65_1 | 1.277411 .2022554 1.55 0.122 .9366081
1.742221
_Ima1Xag1 | .7752202 .0525718 -3.75 0.000 .6787355
.8854206
_Iidx_angi~1 | 2.334687 .2707346 7.31 0.000 1.860038
2.930459
_Iag1Wid0 | 1.603602 .249047 3.04 0.002 1.182772
2.174164
----------------------------------------------------------------------------
--
Where _Iag1Wid1 = over65 vs. under65 at idx_angio==1; and _Iag1Wid0 = over65
vs. under65 at idx_angio==0
Why was _Iag1Wid1 dropped due to collinearity? Also is the OR for _Imale_1
(1.31) for "males versus females AT under 65 (age_65==0)"? --Or is it "males
versus females AT under 65 AND no angioplasty (idx_angio==0)"?
Also, I have tried the 3 way interaction: i.male*i.age_65*i.idx_angio, but
none of the terms were significant.
What am I missing here??
Gregory Daniel
Health Core, Inc.
4735 Ogletown-Stanton Road
Suite 3201
Newark, DE 19713-2094
*
* 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/
