Investigation into the Use of Normal and Half-Normal Plots for Interpreting Results from Screening Experiments

摘要

Daniel, in 1959, proposed the use of half-normal probability plots as a means of interpreting the results from unreplicated factorial experiments. Zahn, in 1975, suggested modifications to Daniel's approach and investigated some operating characteristics of the plots, along with techniques for estimating population sigma from the slope of the standardized contrasts plotted on half-normal probability paper. That work is reviewed here. This project examines some properties and the usefulness of both normal and half-normal probability plots when applied to the interpretation of screening experiments.