Accurate transit forecasting starts with good transit surveys and reasonable expansions of those surveys. This presentation describes how an Advanced Expansion Process (AEP) can be utilized to correct survey biases at a systemwide level. Practitioners have been aware of response bias in transit surveys for some time, particularly with respect to differences in response rates by individual routes and times of the day. Transit planning practitioners have been focused on mitigating sampled flow bias; striving to ensure that the sampled or expanded sample flows reflect actual conditions. The route, time period, and direction (RTD) expansion method of dividing ridership by number of completed surveys is recognized for its straightforward simplicity, however, recent research and practice have noted biases related to passenger flows are maintained using RTD.
The AEP utilizes additional information and techniques during Origin/Destination expansion in addition to the RTD information from each record.
• It uses route segments created by aggregating stop locations from portions of a route that were in the same geographic vicinity,
• Iterative proportion fitting (IPF) techniques,
• And auxiliary flow data (Boarding to Alighting counts and data from Automatic Passenger Counters (APC)).
Expanding the Origin/Destination survey records using the AEP successfully mitigates short trip biases seen when using RTD techniques. The AEP also allows the records to be expanded in such a way that missing flow movements can be incorporated into the expanded dataset, which removes biases that would have resulted from overweighting and underweighting of the various flow movements. Finally, the use of this methodology, when compared to RTD expansion, provides aggregated segment boardings and alightings that more closely replicate the collected Automatic Passenger Counter (APC) data. Issues and challenges associated with this intensive effort will be presented and process improvements for future endeavors will be provided.