Statistical Intervals, 2nd Edition

Preface to First Edition

Engineers, managers, scientists, and others often need to draw conclusions from scanty data. For example, based upon the results of a limited sample, one might need to decide whether a product is ready for release to manufacturing, to determine how reliable a space system really is, or to assess the impact of an alleged environmental hazard. Sample data provide uncertain results about the population or process of interest. Statistical intervals quantify this uncertainty by what is referred to, in public opinion polls, as “the margin of error.” In this book, we show how to compute such intervals, demonstrate their practical applications, and clearly state the assumptions that one makes in their use. We go far beyond the discussion in current texts and provide a wide arsenal of tools that we have found useful in practical applications.

We show in the first chapter that an essential initial step is to assure that statistical methods are applicable. This requires the assumption that the data can be regarded as a random sample from the population or process of interest. In evaluating a new product, this might necessitate an evaluation of how and when the sample units were built, the environment in which they were tested, the way they were measured—and how these relate to the product or process of interest. If the desired assurance is not forthcoming, the methods of this book might provide merely a lower bound on the total uncertainty, reflecting only the ...