The Detection of Individual Differences in Accident Susceptibility

Publisher Website
Google Scholar
Abstract
Most accident distributions look a lot like the Poisson curve. They also look like the negative binomial. It is hard to differentiate between these two theoretical distributions; and each of them may, in any event, be generated by at least two conflicting hypotheses. Univariate study of accident distributions leads almost inevitably to inconclusive results. If bivariate methods (involving two or more time periods) are used, the situation is only a little better. Bivariate Poissons and negative binomials share the major disadvantages of their univariate analogues. Correlational methods imply the Poisson model, while disavowing it explicitly. Conclusions from bivariate study of accidents are conflicting and unclear. Investigations of time periods between successive accidents have produced some promising, but highly tentative, results. The logical premises of this type of investigation are still somewhat ambiguous. If it is assumed that individual differences in accident susceptibility constitute a fruitful area of research, improvements are needed in the available methods, in the development of new methods, and in the availability of adequate data to which the methods may be applied.

This publication has 1 reference indexed in Scilit: