On the Use of Stagewise Regression in Random Balance Screening Experiments.
DESMATICS INC STATE COLLEGE PA
Pagination or Media Count:
Random Balance RB is a design technique that may have much to offer the researcher planning a factor screening experiment. The RB concept is most useful, however, in the design of supersaturated screening experiments. An experiment is supersaturated when the number of factors i.e., design variables under investigation exceeds the number of runs available. As it is, screening experiments are often handicapped by the scarcity of experimental runs because of time, budget, andor resource limitations. The authors are concerned with the supersaturated situation. A major concern with RB experimental design is that there are no specific or unique statistical techniques for analyzing RB designs. The purpose of this technical report is to investigate the use of stagewise regression SR as a method of analysis for RB screening experiments. The authors approach is to determine the efficiency of the first two stages in order to obtain an indication of what can occur between consecutive stages. In doing so, the SR method is compared with an individual F-test approach. Finally, two Monte Carlo case studies are conducted.
- Statistics and Probability