Evaluation of Origin-Destination Data for an External Cordon Line Survey

The primary travel characteristic in question is the mirroring of the magnitude and distribution of the inbound trip patterns to the patterns of the outbound trips. The linear regression analysis for the internal-external (IE) trips plotted by direction and zone volume provided a correlation of 0.09885 for all survey stations combined. The association of the regression lines with the expected 45 degree line when plotting inbound versus outbound trips by zone is a second indicator, and showed an average intercept of about +2.6 trips with a slope of 0.989. The standard error of the estimate was calculated to by 37 trips for all stations combined. The average number of trips per zone is 237.5.

A similar analysis of the external-external (EE) trips produced a regression correlation of 0.9954 and the regression lines had an average intercept of +1.5 trips with a line slope of 0.996.

In addition to studying the distribution of the trips by zone, the O-D data was stratified into four purposes: 1) work 2) personal, business and shopping 3) social recreation, and 4) other. Percent trips by purpose and by direction were calculated and for all stations combined there was less that a 2% discrepancy for any single purpose. The best match was work trips (0.4% difference) and the greatest difference was in the social-recreation trips, with a 1.9% difference.

A comparison of the distribution of trips by trip length was the final area investigated. For I.E. trips inside the cordon line the inbound direction showed an average trip length of 8,86 minutes with a standard deviation of 3.44 minutes and the outbound direction produced and average trip length of 8.89 minutes with a standard deviation of 3.43 minutes.

The total analysis performed shows a very high degree of association between directions of interview for the studied characteristics of trip ends, trip purposes, and trip length distributions for the Longview-Kelso, Washington area.

In analyzing the data it was found that the inbound direction was statistically closer to the expected distribution than was outbound, however; the difference is small (about 2 trips difference in the standard error). The two trip difference should not be construed to indicate that inbound interviewing is "better" than outbound, but rather that the difference is so small that, in fact, the two directions "mirror" one another.

The second part of the research study dealt with determining the optimum time period for interviewing. Sample time periods ranging from three to eleven hours in each direction and both directions were selected and compared to the total 0-D file. Standard deviations were calculated by purpose for each sample indicating how close each one approximated the distribution found in the universe. The best time period tested was a four hour sample in both directions from 2 p.m. to 6 p.m. which resulted in an overall average standard deviation of 14 trips, which indicates that 68% of the zones were numerically less than 14 trips different than the universe. Since there are about 60,000 trips in total, this difference is acceptable especially when one considers the numerous other phases of a comprehensive transportation planning study where uncalculated probabilistic decisions are necessary.

Publication Date: 
Friday, October 31, 1975
Publication Number: 
WA-RD 028.1
Last modified: 
10/12/2016 - 15:40
Bernard C. Koontz, Kenneth C. Kirkland.
Washington State Highway Commission; Washington (State). Dept. of Highways. Division of Planning, Research & State Aid.
Number of Pages: 
Business trips, Evaluation, Origin and destination, Regression analysis, Shopping trips, Standard error, Standard deviation, Traffic data, Traffic volume, Travel, Travel surveys, Trip length, Work trips.