Practitioner survey and measurement error in manual bicycle and pedestrian count programs

ABSTRACT

Manual bicycle and pedestrian count programs that involve community volunteers can be a low-cost means to collect nonmotorized traffic data while providing opportunity for public involvement. The purposes of this study were to investigate why communities are performing manual counts and estimate the degree of measurement error associated with conducting manual counts. An online questionnaire was sent to transportation specialists across the United States. There were 92 responses received from 25 states. Eleven communities were contacted for phone interviews. Information about community volunteers, scheduling and logistics, data collection techniques, and reasons for conducting manual counts were summarized. Some of the reasons cited by the survey respondents for conducting manual counts seem unrealistic and possibly flawed. Measurement error was assessed through a controlled field experiment with 25 counters at five intersections. Lower measurement error rates were observed when using a four-movement data collection technique compared to a more complicated 12-movement technique; however, the differences were not statistically significant. The overall median absolute percent error for the 12-movement technique was 27% and 7% for bicyclists and pedestrians, respectively. The field experiment showed no consistent increases in measurement error when counters were assigned to collect additional information, such as sex of traveler or whether bicyclists wore helmets.

KEYWORDS:

• Assessment
• bicyclist
• manual counts
• pedestrian