Statistical Methods in Research and Development

Many engineers have heard of the recent advances in the field of applied statistics. It is the purpose of this paper to introduce and demonstrate one of the many statistical tools that can be of considerable aid in reducing trial and error experimentation for the design, development, or research engineer. Some variability enters into every experiment that may be undertaken. Some of the variability is due to chance errors of measurement and some is due to definitely assignable causes operating on the factors producing the data. Separation of the various assignable causes entering the experiment and their comparison with experimental error is known as the "analysis of variance." In order to most efficiently determine the significance and magnitude of the assignable variabilities, experimental error must be minimized. Methods of accomplishing this end are discussed. Following a discussion of some basic concepts, a computational example of an analysis is shown. The paper concludes with a summary of steps used in carrying a development problem to completion.