The best statistical distributions of unreported accidents in Malaysia

The effectiveness and efficiency of any road safety policy measure depends directly on the availability and accuracy of the reported accident data on the treated sites. When a large portion of road accidents remains unreported, vital information concerning the characteristics of those accidents will not be available. This paper reports on the distribution of unreported accidents among Malaysian drivers. Based on survey conducted to 2000 Malaysian drivers, we measure the under-reporting information through three variables which are count, proportions and ratio of the unreported accidents. Our aim is to find the best distribution that describes the features of unreported accidents for three variables studied. Results shown that for the count data, the best fitted distribution is Negative Binomial whereas for the proportion, Beta distribution is found to be the best. Lastly, for the ratio data, the best fitted distribution is found to be Lognormal.