A re-examination of the accident proneness concept

Abstract

The method of studying accident proneness by demonstrating that small percentages of people have large percentages of the accidents is unsound and fallacious. When these percentages are compared with a theoretically computed equal liability distribution, the accident distribution obtained is in accordance with chance expectancy. Examples from several studies using the percentage method are discussed. A method for analysis of the variances of accident records of two component variances is suggested, one component attributable to differences in accident liability (personal characteristics and stable environmental conditions), the other to unpredictable chance factors. The method is applicable only when the obtained distribution resembles a composite of Poisson distributions. 13 references. (PsycInfo Database Record (c) 2025 APA, all rights reserved)