Validation of self reporting accidents.

Author(s)
Transport and Road Research Laboratory TRRL TRL, Department of the Environment
Year
Abstract

It is common, in road safety research, to assess some of the results in terms of the numbers of accidents which different groups of motorists have had. This has been done in studies of the benefits of different driver training methods, in driver classification studies, and where different characteristics of road users are being investigated. However, in Great Britain, although national accident statistics are collected there is no record of involvement by named individuals. Police records include such information but only for accidents occurring within particular police areas. There are other types of records of named drivers, but only for limited populations. It has been necessary therefore in studies including a general population of drivers to ask subjects involved in such research studies for a statement of their own accident record over a certain period. It is obvious that the success of this method depends on the accuracy with which individuals report their own accident histories. Inaccuracies could result from deliberate omissions or from subjects' inability to remember details going back over a few years. There was therefore, a need to validate this technique of self-reporting and a study has been carried out to compare the accuracy of drivers own reports of their accidents with accident records. In this particular study a comparison has also been made between two methods of obtaining these reports - self completion postal questionnaires, and person to person interviews. The overall response rate for the two methods used was exactly the same - 83% for the interview method and 83% for the postal method. Also for both methods there was a poorer response from the drivers who were known to have had the greatest number of accidents but with neither method was this difference in response between those who had large or small numbers of accidents statistically significant at the 5% level. The results indicate that a group of drivers when interviewed will report 67% of their accidents and 85% when contacted in a postal survey, thus a postal survey produced more accurate replies. The reason may be that for some topics, like accidents, a postal survey being a more anonymous approach encourages people to give more information than they would when face to face with an interviewer. A postal questionnaire also allows the respondent more time to check accident details and to think back if necessary over a period of time. It has been thought that drivers might be particularly unwilling to admit to their more serious accidents, but in fact it is the less serious rather than the more serious accidents which were more often omitted. This seems to suggest that the main bias arises from a poor memory of insignificant incidents rather than from deliberate omissions. Accident records collected by questionnaire are sometimes used to categorize individuals into those with high or low numbers of accidents. This study has also been used to examine how many motorists would be misplaced if they were classified according to the number of accidents they reported. Several ways of making the grouping were tried and the most accurate classification was given by dividing those reporting no accidents from those reporting one or more. Good results can be obtained by both methods of survey and despite some individual inconsistencies the technique gives reasonably reliable data for group comparisons. However, the postal method was cheaper than the interview method and did produce more accurate replies in this study.

Publication

Library number
962369 ST [electronic version only]
Source

Crowthorne, Berkshire, Transport and Road Research Laboratory TRRL TRL, 1973, 2 p.; Leaflet ; LF 352

Our collection

This publication is one of our other publications, and part of our extensive collection of road safety literature, that also includes the SWOV publications.