The analysis of event rates using intervals

Abstract

Event interval analysis had historical antecedents in the past century, but the analysis of rates of events has been largely performed using counts of events. When the information content of intervals and counts of the same events are compared, it is clear that the information content of counts is sensitive to the number of events in a counting interval. The reduced information content of counts where the number of events in a counting interval is small may affect the analysis of event rates. Both simulated and historical data are used to illustrate such effects. It is concluded that event interval analysis may be more appropriate for the analysis of event rates when the events in question are few in the counting intervals.