ADXQMOD: Set Up a Second-Order Model

Macros for the Design and Analysis of Experiments

ADXQMOD: Set Up a Second-Order Model

%adxqmod(dsin, dsout, vlst ,ORDER2)

where

dsin names the SAS data set that contains the original design.

dsout names the output SAS data set that contains the original design and new variables for second-order terms in the model. The value of dsout can be the same as dsin; in this case, the original data set is replaced.

vlst lists the design factors. These must all be numeric.

ORDER2 adds the quadratic terms to the model. If you omit ORDER2, cross-product terms (interactions) are added, but quadratic terms are not.

The ADXQMOD macro adds variables that represent cross-product terms to a design data set, optionally adds quadratic terms to a design data set, and sets up the second-order regression model in the global macro variable ADXFIT.

The names of the new variables are formed by concatenating the names of their factors. For example, the name of the variable for the cross-product of A and B is AB. You should verify that this convention produces a valid variable name in all cases. To concatenate names:

Each combination of two factor names should have a length of 8 or less.
The constructed name for a cross-product should not conflict with a previously defined variable in the design.

For example, the variable names A and B meet these criteria. The variable names TEMPERAT and PRES do not.

The ADXQMOD macro makes it easier for you to use the REG procedure^* to analyze data for a second-order model. The REG procedure does not accept terms of the form X1*X2, so normally you need to create new variables for quadratic and interaction terms in a DATA step.

The following statements create a central composite design with the ADXCCD macro and output the design to a SAS data set named CENC1. Then the ADXQMOD macro adds new variables for quadratic terms and outputs the original design and new variables to a SAS data set named CENC2.

   %adxgen
   %adxff
   %adxcc
   %adxinit
   %adxccd(cenc1,6,32,14,2.3784)
   %adxqmod(cenc1,cenc2,t1 t2 t3 t4 t5 t6,order2)

Chapter Contents
Previous
Next
Top

dsin	names the SAS data set that contains the original design.
dsout	names the output SAS data set that contains the original design and new variables for second-order terms in the model. The value of dsout can be the same as dsin; in this case, the original data set is replaced.
vlst	lists the design factors. These must all be numeric.
ORDER2	adds the quadratic terms to the model. If you omit ORDER2, cross-product terms (interactions) are added, but quadratic terms are not.