ABSTRACT
Statistical distributions deal with the scatter of numbers, the statistician is often called upon to analyze non-numerical patterns in the classification of events. A good statistician should come to the data with an ability to generalize beyond the classifications presented. To the statistician, the contingency table is nothing but a collection of counts. The statistician should be alert to spot other internal structures in the tables that might be ignored by "standard" statistical tests if they were applied blindly to the structure presented. Thus, a good statistician must be skeptical about the form in which a contingency table is presented. The power of A statistical test of hypothesis can be improved by restricting the alternative hypotheses against which the null hypothesis is tested. The usual test of row and column independence calculates a "distance" from the observed point to the nearest point in the four-dimensional space defined by the hypothesis of row and column independence.
