A Field Guide to Experimental 
Designs

Example SAS code and other packages

Very few (perhaps zero) statistical computer applications produce straight forward and appropriate output for designs anymore complex than a few simple ones. Most of the applications will produce the numbers you need, but you will have to summarize them and put them into the proper format yourself. Furthermore, the terminology used by these applications often does not match that normally used by statisticians.

To help, we have provided sample SAS code to analyse each design. One of the worst ANOVA formats is that produced by the SAS System GLM procedure. However, SAS GLM is also one of the few that can usually generate the numbers you need.

To obtain the proper results from a computer package, you must know before hand (1) what design you are using, (2) what formulas are used to test the hypotheses from that design, (3) and how to generate those tests with your statistical software.

To check the appropriatemness of your analysis you should generate the ANOVA table for your experiment. The values should be consistent with your design: sums of squares and degrees of freedom should add up as you believe they should, and F-tests should use the terms that you expect.

A table with sources of variation and degrees of freedom should be prepared at the time of design and before the experiment is executed in the field. Seat of the pants designs often cannot be analysed because care was not taken. The ANOVA table is an important check before and after!

 

Go to...

Washington State University
Tree Fruit Research and Extension Center
1100 N. Western Ave.
Wenatchee WA 98801

phone: 509-663-8181
fax: 509-662-8714


Copyright © Washington State University Disclaimer
WSU Electronic Publishing and Appropriate Use Policy

Comments concerning this page to webservant@tfrec.wsu.edu

Monday, August 14, 2000