• Journal of Korean Society for Quality Management
• /
• v.33 no.3
• /
• pp.156-160
• /
• 2005

We consider practical experimental design strategies and analysis to find out whether a modified method give better results than the standard method. The most practical design strategy is for experimenter to make r successive runs under the current standard method and then, change the standard method to a modified method to make another r successive runs under a modified method. To test a statistically significant difference between the population mean of the standard method and a modified method, additional recent data for sufficient number of consecutive responses under the standard method is needed to construct external reference distribution(Box, et al., 1968). Upon those informations unavailable, the practical design strategy is to run the experiment by split plot designs. In this paper, two types of split plot designs are proposed and how to determine efficiently the number of repetition within a given method and replication of those two methods are discussed based on results of the level of significance ${\alpha}$= 0.05 and the power being at least 0.9 at the detectable difference of ${\mu}_2-{\mu}_1=1.5{\times}{\sigma}$.