DEVELOPMENT OF MEASURES FOR A DRIVER LICENSING PROGRAM IN THE STATE OF NORTH CAROLINA. PHASE I. ANALYSES OF CURRENT LICENSING TESTS

This study is to examine the measurement properties of the driver licensing tests used in North Carolina, in order to determine their suitability for assessing driver knowledge and skill. The initial licensing rules test is drawn solely from "Traffic Law and Highway Safety" (N.C., 1970). Passing is 70% or more of the items answered correctly. The renewal licensing rules test is similar to the former and based on "Driver's Refresher Handbook of Traffic Laws and Highway Safety" (N.C., 1969). The Supplementary Data Form was administered voluntarily to a sample of license renewal applicants and provided personal background, driving experience, and miles driven. Recorded accidents and violations were available from driver records. Samples consisted of 21,671 applicants who completed the Initial Licensing Rules test in Spring 1969 and 8,187 renewal applicants who took the test in January 1970. Data analyses are to deal with rules test characteristics, rules test validation, and background correlates. The adequacy of the rules test leaves room for considerable improvement. There is too large a proportion of items that are exclusively easy for the license applicants, with serious effects on item and test variance. Only one renewal test form (C) showed any degree of potential value (although slight) by virtue of its significant correlation with violation convictions for females and with accident occurrences for males. Only with the addition to the rules test scores of driver background data, were any reasonable levels of relationship with the accident-violation criteria found. Individual background variables in general were found more consistently and more highly correlated with accidents or violations for males than for females. There is a need to construct test items with more defensible measurement characteristics than were found in the present rules tests. Measures should be built on more than reading ability, or verbal skills as the single component. A wider range of driver characteristics and driving experience should be incorporated for investigation in any further predictive studies of licensing tests. There is a need to develop more readily available standardized measures of driver performance that can serve as defensible criteria, as opposed to depending entirely on recorded accident and violation data.