Samenvatting
Three separate measures of exposure to crashes were used to study collision data from the Ohio Turnpike for the period January 1966 through June 1970. The measures, consisting of induced exposure, vehicle miles, and an interaction model, yielded different results, and the assumption underlying these measures were examined. The author concludes that caution should be exercised to ensure that the appropriate measure is applied, because the exposure measurement has not been perfected and therefore, one may not conclude that one method of measurement is better than another for all circumstances.