Journal of Political Economy Vol. 25 No. 8 1917
Some Advantages of the Logarithmic Scale in Statistical Diagrams
Abstract
Statisticians have long been aware that for some purposes of analysis and graphic representation the logarithmic scale has special advantages which the ordinary natural scale does not offer. But with that awareness they have mostly rested content. Com? paratively few have worked with the logarithmic method and really come to know it in its applications. The logarithmic scale, it has been assumed, would be unintelligible save to experts. Because of its unfamiliarity it has been neglected; through neglect it remains unfamiliar, and thus a serviceable statistical tool lies virtually idle. It is the thesis of this article that the logarithmic scale is too useful as a statistical auxiliary to be disregarded simply because it is not yet generally understood; and, further, that the best way to secure for it a more general understanding and appreciation is to use it intelligently at every appropriate opportunity. This is not by any means to say that the logarithmic scale should prevail? ingly replace the natural scale in ordinary graphic work. Often, however, where only one method is to be followed, the logarithmic construction is clearly superior. Even if such occasions are as yet comparatively infrequent, it is still true that figures drawn to the logarithmic scale will usually afford an informing contrast and supplement to diagrams in the more conventional form. For demonstrating the characteristics of logarithmic diagrams such a contrast is quite the most promising plan. Mr. Bowley, in his brief but excellent discussion of logarithmic diagrams,1 has recommended this procedure. "It would be useful," he remarks, "to offer several diagrams on both scales; for in many series of figures the differences exhibited by the two methods are very instructive." Mr. Bowley felt himself constrained by the limits of space to forego acting upon his own proposal, but the suggestion was good. In the following pages space is deliberately devoted to 1 Elements of Statistics, pp. 188 ff.
- DOI
- 10.1086/253026
- Volume
- 25
- Issue
- 8
- Pages
- 805-841
- Language
- en
- Sources
- semanticscholar openalex crossref