Elevated road fatality rates at night involve multiple variables, each
of which contributes to an unknown degree. In order to determine the
relative contributions of reduced visibility and drivers' consumption
of alcohol, we performed two quasi-experiments analyzing data recorded
by the U.S. Fatal Accident Reporting System from 1980 through 1990. T
he results confirmed that both variables play a major role in nighttim
e road fatalities and revealed new evidence that they contribute diffe
rentially to two classes of fatal collisions: Degraded visibility in l
ow illumination is associated primarily with collisions involving pede
strians and pedalcyclists, whereas drivers' consumption of alcohol pla
ys a larger role in other fatal collisions.