Sampling Procedures for Roadside Interviews in Origin-and-Destination Traffic Surveys HR-8

(1953) Sampling Procedures for Roadside Interviews in Origin-and-Destination Traffic Surveys HR-8. Transportation, Department of

Preview

PDF 76-004.pdf File Size:801kB

Abstract

In the origin-and-destination traffic study of the main routes at Ames, Iowa, in the fall of 1949, the roadside interviews, 7 a. m. to 11 p. m., were recorded in time intervals of 2 min. It is possible, therefore, to reconstruct the traffic flow through any of the interview stations in road sequence of vehicles, except within the 2-min. intervals. Five samples were drawn from the IBM cards. The resulting trips between pairs of origin-and-destination zones were compared with the trips found in the 100-percent interviews. Systematic samples of 10 and 20 percent were drawn by selecting for each direction of travel every 10th and every 5th trip in sequence throughout the 16-hr. period for the four external stations on US 30 and US 69 at the city limits. Statistically, these two samples adhered to the characteristics of a normal distribution and are of value in predicting the probable maximum percent errors to be expected from comparable samples. The third sample was drawn from Station 3 only on a time-controlled basis by taking all vehicles reaching the interview station in the first 2 min. of each 10-min. period. This sample had about the same characteristics as the systematic 20-percent sample. Since these three samples were selected by a procedure difficult to maintain at a roadside station, the fourth sample was taken on a time-and-size-controlled procedure that could be easily performed at the roadside. This sample was satisfactory and of about the same quality as the time-controlled sample. These third and fourth samples were e:q)anded to the lO-mm. , 1-hr., and 16-hr. 100- percent traffic volumes. The 10-min. and the 16-hr. basis gave better results than the 1-hr. expansion. The fifth test was applied to traffic in one direction only. The results indicate that for these stations at Ames, Iowa, acceptable results could be obtained by sampling one direction only. These brief results are insufficient, even for Ames, to determine the best basis of taking an origin-and-destination sample at roadside-interview stations. The results do show, however, that the time-and-size-controlled sample of about 20 percent of the total traffic should yield acceptable results. When the traffic by direction balances each other in interchanges between pairs of origin-and-destination zones, the unidirectional method of sampling could be used.