Chapter Contents |
Previous |
Next |

Macros for the Design and Analysis of Experiments |

ds | gives the name of the SAS data set that contains the design.
The design must be in coded form, either as originally
constructed or after recoding with the ADXCODE macro. |

vlst | lists the design factors. |

nv | gives the number of design factors. By default, nv
is constructed from vlst. |

res | gives the resolution for the design. |

nb | gives the number of blocks in the design. |

NOPRINT | suppresses listing the alias structure. This is useful when you want only the output data sets. |

By default, if *ds, vlst, res,* or *nb* are omitted,
the values from the last design constructed are used.

In orthogonally confounded fractional factorial designs, certain
effects are confounded or *aliased* with each other. The
alias structure of the design is the pattern of this confounding.

- lists the alias structure and stores it in the ADXALIAS data
set. The alias structure is listed as a series of
*alias chains*of effects. Two effects are listed as equal to each other if they are aliased with one another in the design. Effects are listed as equal to "(BLOCKS)" if they are confounded with the block effect. Two aliased effects cannot be simultaneously estimated, so the model constructed in the ADXFIT variable (see below) contains one effect from each alias chain and no effect from a chain that is aliased with blocks. - creates the ADXEFF data set, which contains the design and additional variables for cross-product terms. This data set is generally used only by other macros in the collection.
- creates the macro variables ADXFIT and ADXNFIT. ADXFIT is the right-hand side of the appropriate model to use to analyze the design with PROC REG (in SAS/STAT software) when ADXEFF is the input data set to the REG procedure. ADXNFIT is the number of terms for the model given by ADXFIT.

It is a good idea to examine the alias structure of your design before decoding it, and to name factors so that effects you want to estimate are estimable and not aliased with one another. Alternatively, you can use the FACTEX procedure to construct and decode the design; in this procedure you can directly specify which effects to estimate and which effects to consider nonnegligible. See Part 3, "The FACTEX Procedure," for more details on using the FACTEX procedure.

Suppose you create a fractional factorial design for 5 factors in 16 runs and 4 blocks and you output it to the data set DESIGN with the following statements:

%adxgen %adxff %adxinit %adxffd(design,5,16,4)To examine the alias structure of this design, submit the statement

%adxalias(design,t1 t2 t3 t4 t5,5,4,4)

The statement above uses the default variable names for treatment factors. If you call the ADXALIAS macro immediately after constructing the design, you can leave all parameters blank. Thus, the following statements produce the same output as the two sets of statements above:

%adxgen %adxff %adxinit %adxffd(design,5,16,4) %adxalias()

Chapter Contents |
Previous |
Next |
Top |

Copyright © 1999 by SAS Institute Inc., Cary, NC, USA. All rights reserved.