IDENTIFY effect-specification;
The IDENTIFY statement specifies the causal effects to be analyzed. The effect-specification has the following form:
treatment-variables right-arrow outcome-variables < | adjustment-variables >
Each of the treatment-variables, outcome-variables, and adjustment-variables contains a list of variables in any of the following forms:
variables
{variables}
[variables]
(variables)
The use of braces, brackets, or parentheses for grouping variables is optional but highly recommended because it clearly distinguishes the roles of different types of variables within the effect-specification.
The right-arrow that indicates the causal direction can be of different forms. For information about specifying arrows in the CAUSALGRAPH procedure, see the section Arrow or Edge Specification.
The following is an example for analyzing the effect of a single treatment variable on a single outcome variable without any prescribed adjustment variables:
identify x ==> y;
The following examples are equivalent for specifying a joint causal effect analysis that includes a prescribed set of adjustment variables:
identify {x1-x3} ==> {y1 y5} | {c1-c4 z};
identify {x1 x2 x3} ==> y1 y5 | {c1-c4 z};
identify x1 x2 x3 ==> y1 y5 | c1 c2 c3 c4 z;
You must use an IDENTIFY statement if you use a TESTID statement or if you use the LIST option in the PROC CAUSALGRAPH statement. You cannot specify more than one IDENTIFY statement.
A causal effect consists of one or more treatment variables and one or more outcome variables. The treatment variables and the outcome variables are each specified as a list of one or more valid SAS variable names. The two variable lists are separated by a single right arrow. Every treatment variable and every outcome variable must be measured or observed. This means that you cannot include any treatment or outcome variable in the UNMEASURED statement.
If you specify more than one treatment variable or more than one outcome variable, PROC CAUSALGRAPH attempts to identify the joint effect. For more information about the interpretation of joint treatment effects, see the section Identifying Joint Treatment Effects. However, you cannot specify multiple treatment or multiple outcome variables if you specify the METHOD=IV option in the PROC CAUSALGRAPH statement.
You cannot use the same variable as a treatment variable and an outcome variable. Every treatment variable and every outcome variable must appear in at least one MODEL statement, or else the effect is ill-defined. In that case, the missing variable is ignored. In addition, if you use an IDENTIFY statement, then every model must contain at least one treatment variable and at least one outcome variable.
If you want to prescribe a set of adjustment variables in the effect-specification, use a vertical bar (|) after the list of outcome variables and then specify the list of adjustment variables. This set of prescribed adjustment variables is included in every adjustment set that is tested or constructed by the CAUSALGRAPH procedure. The set of prescribed adjustment variables is ignored if you specify the METHOD=IV in the PROC CAUSALGRAPH statement.
An adjustment variable cannot also be a treatment variable or an outcome variable. Every adjustment variable must appear in at least one MODEL statement, or else it is not defined and is ignored. Every adjustment variable must be measured or observed. This means that you cannot include any prescribed adjustment variable in the UNMEASURED statement.