SAS/STAT procedures for DOE analysis

[MarcTC]

I googled SAS DOE, only to find JMP and ADX. Are there any SAS/STAT procedures for analyzing the following designs:

Completely randomized designs
Randomized block designs
Full factorial designs
Fractional factorial designs
Response surface designs

[Doc_Duke]

If you search support.sas.com for the particular designs, you'll find the procedures to do them. For instance, searching for "randomized block design" yielded PROC FACTEX.

[MarcTC]

Thanks!

[lvm]

Also check out PROC PLAN and PROC OPTEX

[MarcTC]

I am new to SAS DOE tools. Is it true that PROC FACTEX is used for designing an experiment and PROC GLM is used for analyzing the DOE data?

[Paige]

PROC FACTEX is one way to create an experimental design. There are other ways.

PROC GLM is one way to analyze DOE data. There are other ways.

[MarcTC]

> PROC FACTEX is one way to create an experimental
> design. There are other ways.
>
> PROC GLM is one way to analyze DOE data. There are
> other ways.

Paige, could you suggest a few for "other ways"?

[Dale]

For response surface designs, you can use the RSREG (Response Surface REGression) procedure.

The MIXED procedure can be employed to analyze data collected for every one of the designs that you indicated in your first post. The MIXED procedure is frequently preferred over PROC GLM.

Recommendation of the procedures GLM, MIXED, and RSREG is predicated on the response being continuous, and inference from each of these procedures is conditioned on the residuals being normally distributed. There are a number of procedures (GENMOD, GLIMMIX, LOGISTIC, PROBIT) which can be employed to analyze data which have other distributional properties. As indicated by lvm, you really need to look at the introductory chapters to the SAS/STAT reference volume to get some idea of what procedures may be useful for problems of interest to you.

[lvm]

Since you are new to sas, you really should read the introductory chapters of the SAS/STAT 9.2 User's Guide (pdf available for download).