Suicide from the Golden Gate Bridge: do they cluster over time?

Abstract

Epidemics of suicidal behavior have been reported in hospitals (Crawford, 1966) and military bases (Hankoff, 1961). The present analysis concerned data collected by the second author during a visit to the Medical Examiner's office in San Francisco for 47 jumpers from the Golden Gate Bridge for a two-year period ( 1976-1977) and investigated whether they conformed to a random process. The null hypothesis was that the Golden Gate Bridge suicides formed a Poisson process with a mean arrival rate for the two-year period given by the mean of 46/729 per day. A statistical test was designed with two parameters No and L as follows. Take a sample of size N of waiting times and let Z = the number of values No. A program was written to generate significance tables for various values of No and L. In the data, there were four waiting periods of 4 1 (a = 0.31). There were six waiting periods 4 2 (a = 0.47). There were nine waiting periods 4 3 (U = 0.40). There were 12 waiting periods 4 4 (a = 0.32). There were 13 waiting periods 6 5 (a = 0.61). Thus, we could not reject the null hypothesis. It seems, therefore, that jumpers from the Golden Gate Bridge are randomly distributed over time and that no discernible clustering effect was apparent.