This pilot study has demonstrated that the NASS (National Accident Sampling System) concept is a feasible means of providing a national representative accident data base. Overall missing data rates and non response rates for major field tasks reached acceptively low levels by the end of the pilot study. It was concluded that the urbanized areas have a missing data rate approximately twice as high as the rural areas with respect to important NASS variables. The error and missing data rates, and the variability of the latter were found to be sensitive indicators of training deficiencies, coding procedure ambiguities and weakness in data definition.
Abstract