Review article
Statistical analysis of accidents
B. Petz
; Institut za medicinska istraživanja Jugoslaoenshe akademije znanosti i umjetnosti, Zagreb
Abstract
The autor points out the complexity of the problem which is far from being definitely solved. Since the time of the first investigators (GREENWOOD, WOODS, YULE, NEWBOLD) knowledge about this problem has advanced very little. The explanation is given of the comparisons between the observed distribution of accidents and (a) the Poisson and (b) negative binomial distribution. The basic facts concerning the negative binomial distribution are presented. It is specially emphasized that conclusions from these comparisons should be drawn very carefully. Further difficulties confronting the study of accidents are as follows: (1) inconsistency of exposure, (2) no knowledge of how long the exposure should last to give a comprehensive evidence of all important factors, (3) insufficient knowledge of the influence of past accidents on future ones, ( 4) the loss in a sample of accidents of the people who leave the job or are killed, (5) decrease in (he exposure time by sickness absence usually following accidents, (6) no precise definition existing so far of the term accident upon which the value of statistical analysis depends a great deal. The vagueness of the term »accident proneness» is discussed and the term »accident liability« is recommended instead.
Keywords
Hrčak ID:
184184
URI
Publication date:
15.3.1957.
Visits: 817 *