An Application of Scaling to an Industrial Problem

tioning the validity of many of the tests based on the principles of mathematical probability to be found in the literature of sociological research. They are mistaken if they suppose that all such tests are necessarily unsound because of the nature of sociological data. As a general rule, it can be said that whenever simple random or stratified samples can be correctly taken from well defined sociological universes, probability theory can be applied to the resulting sampling errors with just as much validity as in throwing dice or drawing with replacements from a well-shuffled deck of cards. ideal conditions for such tests are realized in sampling completely enumerated existent universes, e.g., a list of all the married couples in New York City divorced and notdivorced in I944. greatest difficulties arise when the universe is hypothetical, or composed of events that have not yet happened, e.g., the same married couples on January i, I944 who will or will not be divorced during the year; and this is the kind of situation with which sociologists will most often be concerned. amount of validity that can be claimed for the usual tests of significance in many such cases is not clear, but a great deal more can be done to increase their validity than has been customary, especially if empirical probabilities are obtained from larger and more homogeneous classes, and if more controlled experiments are undertaken. In our opinion the proper attitude for sociologists to take is expressed in the following sentences from the well-known writer on mathematical probability quoted above: The fact that many applications of probability are based on belief or faith should not discourage us; for it is better to do something, though it may be not quite reliable, than nothing. Only we must not be overconfident about the conclusions reached under such circumstances

Geometry · Scaling · Computer Science · Mathematics · Opinion Dynamics and Social Influence

• "Randomness of Error" in Reproducible Scales

  Open Access•Robert G Smith•Educational and Psychological…•1951

  Add to list