lincom

Lincom is short for 'linear combinations of estimators" which should give you a good idea of what it does. Basically, following any estimation command, lincom allows you to manipulate the coefficients to better understand their relationships to one another. Additionally, you can use lincom to calculate odds ratios and incidence-rate ratios between sets of covariates. The below example based out of [R]lincom should give you a slightly better idea of what that means and how to pull it off.

First, pretend you ran a regression as follows regress y var1 var2 var3 Now that you have coefficents, you want to know the difference between var2 and var3. This is a perfect use for lincom, via which the command would simply be lincom var3 -var2 where var# is whatever you called the corresponding variable used in the regression.

Note: lincom will handle any liner expressions (including multiplying coefficients) but you will need nlcom to do things like divide coefficients.

See the UCLA Stta FAQ for a detailed example of how you might use lincom (in this case to explore a three-way interaction. Additionally, help lincom brings up a list of full commands to access Stata's sample data sets and corresponding examples of how you might use lincom with them.

Back to Estimation