Citation

Miller I, Irick PE, Michael HL, Brown RN. Highw. Res. Board bull. 1953; 76: 31-51.

Copyright

(Copyright © 1953, National Research Council (U.S.A.), Highway Research Board)

DOI

unavailable

PMID

unavailable

Abstract

An appraisal was made of the sampling errors in the estimates of the trip frequencies for the various cells of an origin-destination traffic-survey tabulation. A mathematical exposition is given of the expected errors when the sampling is done at random, by time clusters, and by volume clusters. The paper also discloses the results of an empirical investigation into the sampling errors that actually arose when various sampling methods were applied to the origin-destination tabulations of the Lebanon and Kokomo, Indiana, surveys. These experimental results were in general accord with the theory. It is concluded that on the average, for the large number of estimates involved in any one survey, the theory of random sampling will satisfactorily explain the errors which arise from the various sampling methods proposed. The results given make it possible to predict the average errors and probability limits for these errors when a particular amount and type of sampling has been done. Conversely, one can determine the amount of sampling necessary to keep the sampling errors within specified probability limits. In the second part, the practical application of sampling to actual field conditions is presented. Station arrangement and sampling procedure for obtaining samples of 50 percent and 25 percent as developed during the conduct of an origin-destination survey at Richmond, Indiana, are discussed. Operational procedure for two sampling methods, volume cluster and time cluster, is given for two sample sizes and for locations with various roadway conditions. Sampling on two-lane, three-lane, and four-lane highways carrying traffic volumes ranging from 1,000 to 12,000 vehicles per day is shown to be both practical and economical. The advantages of each of the two sampling methods for various locations are enumerated. It is concluded that the use of systematic sampling in the taking of roadside interviews is not only practical and economical but that it has definite operational advantages. The conclusions indicate that a predetermined sampling procedure will place interviewing on a business- like basis that: (1) is recognized by the vehicle operator and the community, (2) promotes efficient operation of the interviewing stations, and (3) produces statistically sound results.