A STATISTIC FOR RANDOM MODEL TESTING.

The article presents information about a statistic for random model testing. Very often testing for the random model becomes important in the social sciences. For instance, if one has three categories of classification, he would be interested to know whether the chance of an individual falling into these categories is the same. Conventionally the test statistic used for this type of analysis is the contingency x². Authors show here that the notion of variance, if applied to relative frequencies, yields a simple statistic for random model testing. Further-more, it is shown that the few statistic, viz. variance of relative frequencies, is related to the contingency x². As indicated, the social scientist frequently encounters in his research the problem of testing for the random model as well as for the degree of dispersion of responses or other attributes of individuals and aggregates involving nominal data. The variance of relative frequencies can be of significant aid to the researcher in reducing his data to a convenient single measure expressing dispersion and a test for randomness. The following case should exemplify the utility of the statistic. One of the important dimensions of the division of labor in science including the social sciences, is the pattern of scientific foci or interest within different scientific disciplines.