Effects of small sample origin-destination data on transportation study results

This paper discusses the effects of using small-sample origin-destination survey data as the basis for urban transportation planning models. The study was based primarily on home-interview survey data collected in the 1969 San Antonio-Bexar County Urban Transportation Study. The analysis demonstrates the ability of small sample origin-destination data to produce travel estimates that are in close agreement with model results obtained by using traditional large-sample survey data. The survey data for 12,477 dwelling units (i.e., a 5 percent sample) were used as a data base from which repeated geographically stratified random samples of 6400, 3200, 1600, 800, 400, and 200 observations were drawn. Two samples for each sample size, representing the 10th and 90th percentiles of distributions of sample estimates of total automobile-driver travel, were selected for evaluation. This procedure provided a 0.8 probability that samples of similar size will produce travel estimates as good as or better than those obtained here. The selected samples were used to develop inputs to trip-generation, trip-distribution (gravity), and traffic-assignment models. Samples of 400 or more dwelling units were found to produce acceptable results. /Author/