Split-Plot Analysis and Design Experiments When Observations Are Correlated

In split-plot experiments, most of articles dealt with subplots of same size. However, in practice, this might not be the situation. There might have subplots of different sample sizes. In such cases, the analysis could be difficult. We expect that the analysis of the same size can be extended to that of different sizes. If not possible, we have to find another approach. This is the focus of the first year. In designing experiments, most of articles dealt with designing experiments when resulting data are uncorrelated. However, in practice, it might have the situations when the resulting data are correlated in some way. In such cases, the difficulty of design increases. We expect to find design so that estimates of all the factorial parameters have the same precision and correlations. This is an interesting research topic. This is the focus of the second year.

Project ID:PA9506-0321
External Project ID:NSC95-2118-M182-002