However, it is one thing to show associations
between accidents and behavioural characteristics
and another to show that the behavioural characteristics
contribute to accident risk. Ideally, one could
establish that behavioural characteristics contributed
to accident risk by manipulating the former in
an experiment and observing an effect on the latter.
In the case of driving speed, it is well known that
interventions to reduce speed in particular locations
can reduce accident rates. However, a causal link
between a general tendency of drivers to drive fast
and a higher accident rate has not been tested in
this way. Thus many drivers report that they are
“fast but safe.” Indeed, faster drivers tend to be
more confident and have a higher opinion of their
driving skill. They believe that they can handle the
speeds at which they drive. The question is whether
this is, in fact, the case or whether fast drivers are
at increased risk because of the speeds at which
they drive. Similarly, although many accidents are
obviously caused by blatant violations of the highway
code or the law, a tendency to commit violations
may be associated with an increased risk of accidents
of other types. Given that a large proportion
of accidents do not involve deliberate lawbreaking,
it would be important to know whether this was the
case.
Experimental tests of the possible contribution
of behavioural characteristics such as fast driving,
tendency to commit violations, or careless decision
making to accidents would be extremely expensive
and difficult to mount. However, a weaker test of a
possible causal link consists of examining whether
behavioural characteristics are associated with particular
types of accident for which a plausible causal
mechanism can be established. For example, if the
characteristic of fast driving is a direct cause of
increased accident risk, its effect should be detectable
specifically on accidents in which excess speed
could have been a contributory factor. If it turned
out that fast driving was linked with accidents in
which no plausible link with speed could be inferred,
this would imply that fast driving was but a marker
of some other characteristic that increased risk of
accident.