There are examples where safety engineering has had considerable success, but it has been difficult to prove that a particular measure is effective in reducing accidents. Sites are also chosen not at random but because of recent high number of accidents. Exact behaviour error is also difficult to pinpoint. Road users try to turn engineering modifications to their advantage. The aim of the studies is to establish the range of behavioural changes that occur following installation of a selection of measures, the possible links between behaviours and accident risk, and the possibility of "fine tuning" measures to take account of behavioural responses. Behaviours of primary interest are those which (1) indicate road users' perceptions of road environment; (2) violate traffic law; (3) cause conflicts between road users; (4) are related to accident occurrence; and (5) have some prospect of being improved by design modifications. Most reliable quantitative measures are: speed, deceleration and braking, following distance, overtaking rates, gap acceptance, failure to give way, pedestrian crossing strategies, choice of crossing location. Measurements rely heavily on automatic data collection and video techniques. Main aims of engineering modifications are: (1) to reduce error making by removing difficult tasks, making tasks easier and slowing speeds; (2) to make road users more alert/observant; and (3) to make road users more aware of each other.
Abstract