Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, is already up and running.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Stata command to compare linear combinations of predicted values after regression

From   Ari Samaranayaka <>
To   <>
Subject   st: Stata command to compare linear combinations of predicted values after regression
Date   Sun, 2 Dec 2012 18:02:15 +1300

Hi folks,
Does anyone know a stata command to get incidence rate ratio (IRR) of one group relative to another group after poisson regression when none of the groups are directly represented in the model, but both groups can be defined by linear combinations of groups in the model?

The model is in the form (a simplified version) :
xt:poisson y i.x
where x is a variable with mutually exclusive categories.

The groups I want to compare are different linear combinations of categories of x. These groups cannot be directly included in the model because they are not mutually exclusive. For example, in the table below P Q R S are categories in variable X; groups I want to compare are A and B; numbers in the table are the number of observations.

            P     Q     R     S
Group A    25    30    35    40
Group B    50    10    15    60

observations in group A includes 25 from P, 30 from Q, 35 from R, 40 from S.
observations in group B includes 50 from P, 10 from Q, 15 from R, 60 from S.
Groups A and B are not mutually exclusive.

I know lincom command can be used to get IRR for one linear combination of regression coefficients relative to another linear combination of coefficients. But none of the A or B groups are linear combinations of coefficients, they are linear combinations of number of observations corresponding to coefficients from the model. Although we can do hand calculations to get IRR by comparing predicted values in groups A and B, I am looking for a stata command for this because in actual analysis I have very large number of combinations to deal with.
Thanks for any help.

*   For searches and help try:

© Copyright 1996–2016 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index