Is the time of suicide a random choice? A new statistical perspective.

Abstract

The problem of the time distribution of suicide has aroused the interest of many authors, but their findings often appear uncertain and contradictory. These inconsistencies can be partially explained by both the choice of the sample and the statistical method used for the analysis. In this study, in order to reveal expected and unexpected periodicities in the time distribution of suicides, we analyzed a small but complete and very homogeneous sample of suicides which occurred in the Mountain District of the Metauro River Valley in Italy in the period 1960-1994. A recent circular statistics technique, the "maximization of mean vector length" was used. We found no significant weekly and lunar patterns, but significant seasonal and intraseasonal cycles, with peaks in March and at the beginning of each season. The superposition of these two significant cyclic trends was also studied.