This report presents a study aimed at exploring the reliability of self-reported data in analyses of accident liability. Data from two comparable postal questionnaire surveys of accidents carried out in 1987/8 and 1990/91 has provided the opportunity to compare accident and exposure data collected from the same respondents on two occasions three years apart. The report also includes some overall comparisons between the accident data collected in these surveys with similar data collected during the course of two later accident liability surveys. The study shows that although in an overall statistical sense the annual mileage reported by drivers in the first survey is proportional to that reported by the same drivers in the second survey, there is a great deal of variability in the individual mileage estimates. In contrast, providing the accident reporting period is constant from survey to survey and memory loss effects can thereby be avoided, the study shows that considerable consistency in the accident liabilities resulting from the surveys of this kind can be expected. The similarity of the average accident frequencies reported in the '87 and '91 surveys with those obtained in two more recent studies confirms the reliability of the self-report method for use in accident studies. (A)
Samenvatting