STD variable=value <variable=value …>;
You can use this statement to specify the standard deviations of the treatment variable, mediator variable, and covariates. The values for standard deviations must be nonnegative.
For example, the following statement specifies the standard deviations of the variables x1 and x2:
std x1=2.3 x2=4.1;
The STD statement is not required. When the standard deviations of variables are used in evaluating specific causal mediation effects, PROC CAUSALMED can compute the standard deviations from the input raw data automatically. When you specify the STD statement, it overwrites those computed values of standard deviations. Respecifying standard deviations of variables can affect the evaluation of causal mediation effects only when both of the following conditions are true:
The treatment variable, mediator variable, or covariate that you specify in the STD statement is a continuous variable.
The standard deviation of the specified treatment variable, mediator variable, or covariate serves as the scale of the respective variable level that is used to evaluate causal mediation effects.
In other words, specifying the STD statement is effective only when you use the SD keyword to define the level of a continuous treatment variable, mediator variable, or covariate for computing specific sets of causal mediation effects that are defined in the EVALUATE statement.
A practical situation in which you might want to use the STD statement is when you input sampling weights by using the WEIGHT statement. PROC CAUSALMED computes the weighted standard deviation by a formula that is consistent with the formula used in PROC MEANS. However, this formula might not be appropriate in every instance. In such cases, you can compute your own standard deviations by using the appropriate formulas and input them into the procedure by using the STD statement.
For more information about the default weighted standard deviation formula that the CAUSALMED procedure uses, see the WEIGHT statement.