Abstract

[This abstract is based on the author’s abstract.]Seventy-five years ago it was demonstrated that the key concepts of randomization, blocking, and replication make it possible to conduct experiments on processes that may not be in a state of statistical control. Yet even today confusion remains about the need for statistical control as a prerequisite for conducting valid industrial experiments. This article reviews and extends the original argument, reusing the original examples to discount the need for statistical control as a prerequisite in industrial experiments. Clarifying this issue may encourage quality practitioners to find wider use for designed experiments.