The FACTEX Procedure

Features

There is no inherent limit to the number of factors and the size of the design that you can construct with the FACTEX procedure. Instead of looking up designs in an internal table, the FACTEX procedure uses a general algorithm to search for the construction rules for a specified design.

You can use the FACTEX procedure to generate designs such as the following:

  • factorial designs, such as $2^3$ designs, with and without blocking

  • fractional factorial designs, such as $2^{4-1}_\mr {IV}$, with and without blocking

  • split-plot and fractional split-plot designs

  • three-level designs, with and without blocking

  • mixed-level factorial designs, such as $4 \times 3$ designs, with and without blocking

  • randomized complete block design

  • factorial designs with outer arrays

  • hyper-Graeco-Latin square designs

You can also create more complex designs, such as incomplete block designs, by using the FACTEX procedure in conjunction with the DATA step.

You can save the design constructed by the FACTEX procedure in a SAS data set. After you have run your experiment, you can add the values of the response variable and use the GLM procedure to perform analysis of variance and study significance of effects.

The FACTEX procedure is an interactive procedure. After specifying an initial design, you can submit additional statements without reinvoking the procedure. After you have constructed a design, you can do the following:

  • print the design points

  • examine the alias structure for the design

  • modify the design by changing its size, changing the use of blocking, or specifying the effects of interest in the model again

  • output the design to a data set

  • examine the confounding rules that generate the design

  • randomize the design

  • replicate the design

  • recode the design from standard values (such as $\pm 1$) to values appropriate for your situation

  • find another design