In proc regress, proc rlogist and proc survival, you can use a * between two variables (such as two categorical variables or one categorical and one continuous variable) to create an interaction term on the model statement. However, you cannot do this with two continuous variables; you need to create the interaction term in a data step before running the model. For example, srsex and racehpra are categorical variables. In the example below, we create the interaction term between srsex and racehpra.
proc regress data=temp1 filetype=sas design = jackknife; weight rakedw0; jackwgts rakedw1--rakedw80 / adjjack=1; model ab23 = srsex racehpra srsex*racehpra; subgroup srsex racehpra; levels 2 4; run;Number of observations read : 55428 Weighted count: 23847415 Observations used in the analysis : 1000 Weighted count: 466228 Denominator degrees of freedom : 80 Maximum number of estimable parameters for the model is 8 Weighted mean response is 42.854796 Multiple R-Square for the dependent variable AB23: 0.081618 Variance Estimation Method: Replicate Weight Jackknife Working Correlations: Independent Link Function: Identity Response variable AB23: AB23 ---------------------------------------------------------------------- Independent P-value Variables and Beta T-Test Effects Coeff. SE Beta T-Test B=0 B=0 ---------------------------------------------------------------------- Intercept 47.87 1.81 26.51 0.0000 SRSEX MALE 1.63 2.41 0.68 0.5011 FEMALE 0.00 0.00 . . RACEHPRA LATINO -9.67 2.04 -4.75 0.0000 PACIFIC ISLANDER 4.32 6.03 0.72 0.4757 AIAN -3.47 2.68 -1.29 0.1992 ASIAN 0.00 0.00 . . SRSEX, RACEHPRA MALE, LATINO 3.51 3.09 1.14 0.2596 MALE, PACIFIC ISLANDER -5.64 7.27 -0.78 0.4398 MALE, AIAN -0.39 3.49 -0.11 0.9108 MALE, ASIAN 0.00 0.00 . . FEMALE, LATINO 0.00 0.00 . . FEMALE, PACIFIC ISLANDER 0.00 0.00 . . FEMALE, AIAN 0.00 0.00 . . FEMALE, ASIAN 0.00 0.00 . . ---------------------------------------------------------------------- ------------------------------------------------------- Contrast Degrees of P-value Freedom Wald F Wald F ------------------------------------------------------- OVERALL MODEL 8 960.50 0.0000 MODEL MINUS INTERCEPT 7 8.59 0.0000 INTERCEPT . . . SRSEX . . . RACEHPRA . . . SRSEX * RACEHPRA 3 1.12 0.3457 -------------------------------------------------------
In the next example, we will create an interaction term using a categorical variable, racehpra, and a continuous variable, ae13.
proc regress data=temp1 filetype=sas design = jackknife; weight rakedw0; jackwgts rakedw1--rakedw80 / adjjack=1; model ab23 = ae13 racehpra ae13*racehpra; subgroup racehpra; levels 4; run;Number of observations read : 55428 Weighted count: 23847415 Observations used in the analysis : 322 Weighted count: 158239 Denominator degrees of freedom : 80 Maximum number of estimable parameters for the model is 8 Weighted mean response is 40.890349 Multiple R-Square for the dependent variable AB23: 0.058683 Variance Estimation Method: Replicate Weight Jackknife Working Correlations: Independent Link Function: Identity Response variable AB23: AB23 ---------------------------------------------------------------------- Independent P-value Variables and Beta T-Test Effects Coeff. SE Beta T-Test B=0 B=0 ---------------------------------------------------------------------- Intercept 46.99 3.16 14.87 0.0000 AE13 -0.15 1.13 -0.14 0.8917 RACEHPRA LATINO -5.02 3.64 -1.38 0.1715 PACIFIC ISLANDER -9.37 7.35 -1.28 0.2057 AIAN -4.19 4.37 -0.96 0.3396 ASIAN 0.00 0.00 . . AE13, RACEHPRA 1, LATINO -0.89 1.36 -0.66 0.5113 1, PACIFIC ISLANDER 4.09 1.65 2.49 0.0150 1, AIAN -0.47 1.45 -0.33 0.7435 1, ASIAN 0.00 0.00 . . ---------------------------------------------------------------------- ------------------------------------------------------- Contrast Degrees of P-value Freedom Wald F Wald F ------------------------------------------------------- OVERALL MODEL 8 257.15 0.0000 MODEL MINUS INTERCEPT 7 4.35 0.0004 INTERCEPT . . . AE13 . . . RACEHPRA 3 0.93 0.4289 AE13 * RACEHPRA 3 4.76 0.0042 -------------------------------------------------------
In the example below, we try to use two continuous variables, ae13 and ae14, to create the interaction term. The error message that was displayed in the log is shown below. If you want to include this type of interaction term in your model, you will need to create it in a data step before running the procedure and include it on the model statement.
proc regress data=temp1 filetype=sas design = jackknife; weight rakedw0; jackwgts rakedw1--rakedw80 / adjjack=1; model ab23 = ae13 ae14 ae13*ae14; run;SEMANTIC ERROR : (Message 452) At most one continuous variable is allowed in each term of the RHS of a MODEL statement