I'd like to estimate relative risks in addition to odds ratios.
Why do I get the same estimates from -or- option after -logit- as from -rrr-
after -mlogit-, but different estimates from -relrisk-?
When using -mlogit- or -svymlogit- and option -rrr-, are the estimates
actually odds ratios? If so,
how can I get relative risks from multinomial regression?
In the output below, I'm just using a binary outcome variable as an example.
Actually, I want to use -svymlogit- for a 4-level outcome variable.
Thank you for your help.
Ssu Weng
[email protected]
. logit APNCB PP_5X1 PP_5X2 PP_5X3 PP_5X5, or
Logit estimates Number of obs =
5632
LR chi2(4) =
66.44
Prob > chi2 =
0.0000
Log likelihood = -3661.7963 Pseudo R2 =
0.0090
----------------------------------------------------------------------------
--
APNCB | Odds Ratio Std. Err. z P>|z| [95% Conf.
Interval]
-------------+--------------------------------------------------------------
--
PP_5X1 | 1.865625 .1609324 7.23 0.000 1.575426
2.209279
PP_5X2 | .9794411 .1370441 -0.15 0.882 .7445222
1.288484
PP_5X3 | 1.395263 .0915377 5.08 0.000 1.226908
1.586719
PP_5X5 | 1.54276 .1649583 4.05 0.000 1.251078
1.902446
----------------------------------------------------------------------------
--
. relrisk
Po Rel Risk Lower Limit Upper Limit
RR PP_5X1: 0.350 1.432 1.311 1.552
RR PP_5X2: 0.368 0.987 0.822 1.165
RR PP_5X3: 0.353 1.224 1.136 1.314
RR PP_5X5: 0.361 1.290 1.147 1.435
. mlogit APNCB PP_5X1 PP_5X2 PP_5X3 PP_5X5, rrr
Multinomial regression Number of obs =
5632
LR chi2(4) =
66.44
Prob > chi2 =
0.0000
Log likelihood = -3661.7963 Pseudo R2 =
0.0090
----------------------------------------------------------------------------
--
APNCB | RRR Std. Err. z P>|z| [95% Conf.
Interval]
-------------+--------------------------------------------------------------
--
1 |
PP_5X1 | 1.865625 .1609324 7.23 0.000 1.575426
2.209279
PP_5X2 | .9794411 .1370441 -0.15 0.882 .7445222
1.288484
PP_5X3 | 1.395263 .0915377 5.08 0.000 1.226908
1.586719
PP_5X5 | 1.54276 .1649583 4.05 0.000 1.251078
1.902446
----------------------------------------------------------------------------
--
(Outcome APNCB==0 is the comparison group)
** APNCB is a common outcome
. tab APNCB
APNCB | Freq. Percent Cum.
------------+-----------------------------------
0 | 3592 62.90 62.90
1 | 2119 37.10 100.00
------------+-----------------------------------
Total | 5711 100.00
*
* 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/
