Interpreting results from Lincom command

[Maria Carrasco]

Hi colleagues,

I would immensely appreciate help to interpret the results of running the lincom command with the syntax below.

My interpretation is that clients of sex workers (Client=1, non-client=0) who are married have an odds ratio of 2.96 in terms of not being medically circumcised (vmmc_rev=1 not being medically circumcised, vmmc=0, being medically circumcised) compared to non clients of sex workers who are single.

In advance, many thanks for your help with this.

*********

. do "C:\Users\MARIAA~1\AppData\Local\Temp\STD28dc_0000 00.tmp"

. svy: logistic vmmc_rev client##i.marital_2 i.age_10 i.education_2 i.religion_3
(running logistic on estimation sample)

Survey: Logistic regression

Number of strata = 20 Number of obs = 12,454
Number of PSUs = 721 Population size = 12,339.107
Design df = 701
F( 10, 692) = 30.00
Prob > F = 0.0000

----------------------------------------------------------------------------------------
| Linearized
vmmc_rev | Odds Ratio Std. Err. t P>|t| [95% Conf. Interval]
-----------------------+----------------------------------------------------------------
client |
Yes | 1.299014 .2211201 1.54 0.125 .9299729 1.814501
|
marital_2 |
married/partnered | 1.613687 .1533006 5.04 0.000 1.339106 1.944569
|
client#marital_2 |
Yes#married/partnered | .8945178 .2872448 -0.35 0.729 .4761908 1.680339
|
age_10 |
25-34 | 1.047437 .1128328 0.43 0.667 .8477647 1.294137
35-44 | 1.248614 .1393511 1.99 0.047 1.002919 1.554499
45-54 | 1.554565 .2410289 2.85 0.005 1.146583 2.107718
55-59 | 1.556044 .3371461 2.04 0.042 1.01689 2.381056
|
education_2 |
secondary-higher | .4055958 .0327936 -11.16 0.000 .3460607 .4753732
|
religion_3 |
2 | .1449861 .0541511 -5.17 0.000 .0696402 .3018511
3 | .7283261 .2297458 -1.00 0.315 .3920625 1.352996
|
_cons | 7.510644 .6838538 22.14 0.000 6.281164 8.980783
----------------------------------------------------------------------------------------
Note: _cons estimates baseline odds.

.
end of do-file

. do "C:\Users\MARIAA~1\AppData\Local\Temp\STD28dc_0000 00.tmp"

. lincom 1.client + 1.marital_2, eform

( 1) [vmmc_rev]1.client + [vmmc_rev]1.marital_2 = 0

------------------------------------------------------------------------------
vmmc_rev | Odds Ratio Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
(1) | 2.096201 .441758 3.51 0.000 1.385921 3.170496
------------------------------------------------------------------------------

[Clyde Schechter]

No, that's not right. In fact, 1.client + 1.marital_2 does not correspond to an odds ratio for any group of people in your study. To get an odds ratio for people who are both married and clients you need

lincom 1.client + 1.marital_2  + 1.client#1.marital_2, eform

[Maria Carrasco]

Thank you so much. I have been struggling with this for a long time, reading through various threads and could not come with the right answer. This is very much appreciated.

[Maria Carrasco]

A quick follow up question to make sure I am on the right path, in terms of the interpretation of the results. Does the lincom command give me the odds ration for people who are both married and clients compared to people who are not-married and not-clients (note we only had two categories in each variable). Thank you!

[Clyde Schechter]

Yes, exactly.