Monitoring passenger use measures such as boardings, revenue, passenger-miles, and load can be an expensive task for transit systems. One way to reduce data collection costs is to measure only one of these items, the auxiliary item, and then apply multiplicative factors that are estimated from a small joint sample to its estimated mean to estimate the means of the other items. Statistical aspects of the conversion factor approach are presented, including sample size estimation and determination of the accuracy of both the conversion factors and the inferred estimates. An optimal sampling plan that minimizes the combined cost of estimating the conversion factors and estimating the mean of the auxiliary item is determined. Cost savings between 0 and 75 percent of the cost of direct estimation are obtained for various situations. This paper appeared in transportation research record no. 1144, Transit management, marketing, and performance. For covering abstract see irrd no 818469.
Abstract