Therefore, although there is considerable scope
for error and bias in using drivers’ brief written reports
to determine accident type, the above considerations
suggest that with large samples, it should
be possible to detect specific patterns of relationship
between behavioural characteristics and accidents
of different types. Preliminary support for this view
comes from a study in which self-reported fast driving
speed as a general behavioural characteristic was
shown to be specifically associated with what were
judged independently to be “speed-related” accidents
(West et al. 1991).