Citation

Horbachov P, Makarichev O, Svichynskyi S, Ivanov I. Transportation (Amst) 2022; 49(1): 115-136.

Copyright

(Copyright © 2022, Holtzbrinck Springer Nature Publishing Group)

DOI

10.1007/s11116-021-10168-6

PMID

unavailable

Abstract

Travel surveys in cities remain the main source of information for obtaining people's trip characteristics and developing transport models that serve to predict the performance indicators of transport systems. Despite the large number of travel surveys that have been conducted, there is still no universal methodology for survey design and no strict instructions for choosing a certain methodology for a particular case. Moreover, existing guidelines for sample size definition do not provide any estimates of the accuracy of obtained results. A new approach is proposed for determining the number of sample observations necessary for obtaining travel characteristics--such as travel time or distance--with a given precision. It takes into account the probabilistic nature of sample estimates and does not depend on the trip characteristic being studied. An example of the application of this approach demonstrated that existing guidelines for sample size definition allow trip characteristics to be defined with an error of 20-40%. Additionally, it was determined that while maintaining an average error below 30%, there is a possibility to decrease the required number of observations by up to 56% compared to the number recommended in the Handbook of Transport Modelling (HTM). The study also shows that it is possible to significantly decrease the sample size when the distribution law of the trip characteristic being investigated is known. Another outcome of the study was a substantiated reference size for a pilot survey sample.

Language: en