|
Activity Number:
|
284
|
|
Type:
|
Roundtables
|
|
Date/Time:
|
Tuesday, August 5, 2008 : 12:30 PM to 1:50 PM
|
|
Sponsor:
|
Section on Quality and Productivity
|
| Abstract - #302116 |
|
Title:
|
Optimal Design of Split-Split-Plot and Strip-Plot Experiments
|
|
Author(s):
|
Bradley Jones and Peter Goos*+
|
|
Companies:
|
SAS Institute Inc. and Universiteit Antwerpen
|
|
Address:
|
Prinsstraat 13, Antwerpen, International, 2000, Belgium
|
|
Keywords:
|
split-plot ; optimal design ; strip-plot ; split-split-plot ; random effects
|
|
Abstract:
|
Complete randomization for designed experiments on multistep processes may be infeasible due to logistics of the flow of units through the process. In such cases, one approach is to split batches of units coming out of the first process so the split batches see different conditions in the second process. Strip-plot and split-split-plot designs have three variances to estimate. The discussion will identify the differences in the structure of these two designs and the trade-off between efficiency and convenience in the choice of one versus the other.
|
