On-time performance measures used by transportation operating agencies typically use definitions, procedures, and report formats that represent an operational rather than passenger-oriented perspective. Although providing a useful barometer of operational effectiveness, such systems only indirectly measure the passengers' experience of service. Using a random sampling methodology to construct a computerized data base of about 50, 000 morning rush hour subway trains, the subway performance model developed by new york state's metropolitan transportation authority inspector general's office is designed tomeasure service as subway passengers experience it. The system focuses on actual, not scheduled, service; it measures aspects of service most meaningful to riders, in terms they can relate to, and on a scale experienced by passengers. Measuring performance according to this principle affects every aspect of research design and analysis, including the selection of measurement points, the definition of a trip and a route, the time periods used, the scale of analysis (system, route, or more detailed), and the statistics to be reported. The basic concept also entails a reconsideration of the way train cancellations, bypasses, service adjustments, extra service, and headway irregularities are treated in measuring on-time performance. Featuresof the methodology resolve many of these analytical issues, while presenting numerous avenues for further research and development. This paper appears in transportation research record no. 1266, Urban public transportation research 1990 .
Abstract