A Flexible Statistic for Detecting a Preponderance of Short or Long Term Variation in a Series of Events

Abstract

A series of independent, normally distributed events may be arbitrarily intervalized on some a priori basis. The within interval variation is taken as a measure of short term variation in the series of events. The between interval variation is taken as a measure of long term variation in the series of events. The a priori intervalization is under the experimenter's control so that he may define short and long term variation for his own experimental situation. A conservative F statistic is developed for detecting a preponderance of short or long term variation.