Abstract
This paper reviews the probable characteristics of the populations the engineer is likely to encounter, and considers the dangers inherent in reliance upon analysis based on a characteristic not truly representative of the population. For comparison between two or more samples, an inappropriate choice of population characteristic may not matter a great deal; but any estimate of extreme values must depend vitally upon our confidence, on grounds quite independent of the samples themselves, that the samples are indeed drawn from populations having the same general characteristic, albeit one different from that on which the comparison is based.
Get full access to this article
View all access options for this article.