Graphical Statistics and Common-Sense Applications

Analogs in statistics and mechanics can give the engineer a commonsense meaning of arithmetic means, standard deviations, and lines of mutual regression by least squares. Even so, numbers often have less vivid impact than graphs in the interpretation and projection of data which vary. Examples show how data can be analyzed graphically using medians and quartile deviations as measures of averages and variations. Such graphical methods lead to inferences equally valid and usually more meaningful than corresponding numbers arrived at by the method of least squares. A scale of probable deviations is presented in terms of percentage frequency and quartile deviations which, for normal distributions, leads graphically to the same probability estimates as those derived numerically by least squares, less abstractly and in less time. A list of papers on probability and statistics as indexed in Transactions, ASCE from 1901 through 1950 is appended.