A Useful Statistical Test

Abstract

THE method described below gives a useful criterion of the significance of the difference between an observed and a hypothetical frequency distribution. Since it provides an exact test, it is particularly useful in the case of small samples, in which the expectation in some frequency classes may be very small. It is only applicable when the number of possible samples containing the same number of observations as the observed sample is determinable. The test answers the question: Is this a reasonably likely sample to be obtained in a random draw? It does not give the proportion of less likely samples which may be expected to arise, as does the χ²-test.