while to apply to suicide. Refinements of Poisson regression models can be applied to low-base rate events and expand our understanding of suicide by sorting out the effects of age, race, and sex on clustered suicides (i.e., within counties) and improving our capability for assessment in prevention programs.

Because suicide has a low base-rate, studies need large populations to yield significant results. This problem is exacerbated by the poor reporting of suicides. Official suicide statistics capture completed suicides only and are fraught with inaccuracies. The numbers are inaccurate because of classification as death from undetermined causes and because of underreporting. It is estimated that most if not all of the undetermined cases are actually suicides. The quality of data on suicide attempts is even more tenuous than that of completed suicides.

Part of the reporting problem stems from the marked differences in the training and background of the person who by law certifies a death as a suicide. In the United States, the qualifications range from simply having an interest in the job (e.g., Indiana) to specialized training in forensic pathology (e.g., Oklahoma). Medico-legal officials may be elected, appointed or serve ex-officio (e.g., outside of the larger California counties, elected county sheriffs). Investigations may be centralized within a state (e.g., Rhode Island) or organized by each county (e.g., Utah). Internationally, differences in the organization and functioning of the officials monitoring suicide as a cause of death can produce artifactual differences even between similar countries such as England and Scotland. Most developing societies lack registries and expertly trained officials to record suicide. Further, there are cross-national differences in the underlying logic of classifications systems. In India, for example, the classification scheme focuses on social stresses and not psychopathology. Countries with religious sanctions against suicide tended to report suicide rates that were lower than for countries without sanctions. Furthermore, difference among countries may reflect the capacity of the emergency health care system to respond rather than differences in the intent of the individuals. The high rate of suicide among young Chinese women may result from their ready access to extremely toxic pesticides in the face of limited availability of emergency treatment. Each of these factors affects the nature, extent, and quality of the investigation and the classification of deaths as suicide.