The CAUSALTRT Procedure

OUTPUT Statement

  • OUTPUT <OUT=SAS-data-set> <keyword=name …keyword=name>;

The OUTPUT statement creates a new SAS data set that contains all the variables in the input data set and optionally contains predicted values that are used to estimate the treatment effect. If the estimation of the causal effect requires that a regression model for the outcome variable be fit, then you can request that the predicted potential outcomes be included in the output data set. You can request that the predicted propensity scores and inverse probability weights be included if the estimation of the causal effect fits a model for the treatment assignment.

The output data set contains estimated potential outcomes and propensity scores for all observations in which the explanatory variables for the model are not missing, regardless of whether the outcome is missing.

This behavior enables you to predict potential outcomes and propensity scores of "new" observations without affecting the original model fit. These new observations must have nonmissing explanatory variable values and treatment assignment, but their outcome values are missing in the input data set. This way, these new observations cannot be used for model fitting but their predicted values are computed in the OUT= data set.

You can specify the following options:

OUT=SAS-data-set

names the output data set. If you omit this option, an output data set is created and given a default name that uses the DATAn convention.

keyword=name

specifies a statistic to be included in the output data set and assigns the specified name to the new variable that contains the statistic. Specify one or more of the following keywords and names (connect each keyword and name by an equal sign):

IPW=name

requests the predicted inverse probability weight.

POCNT=name
PREDCNT=name

requests the predicted potential outcome for the control condition.

POTRT=name
PREDTRT=name

requests the predicted potential outcome for the treatment condition.

PSCORE=name
PS=name

requests the predicted propensity score.

If you do not specify any keyword=name options, the output data set contains only the original variables.

Last updated: December 09, 2022