Reducing Systematic Ridership Deviations between Farebox Counts and Sample Estimates
DESCRIPTION OF NEED
Transit ridership in unlinked passenger trips may be counted directly by fareboxes semi-automatically or by drivers manually. Counting passengers through fareboxes is semi-automatic because drivers still need to manually record boarding for many non-cash fare categories. As more transit operators install electronic registering fareboxes on their entire fleet, counting passengers directly is becoming the transit industry’s standard. Many agencies, however, continue to estimate unlinked passenger trips through a sampling procedure, particularly for annual reporting to the National Transit Database (NTD).
Besides reporting annual ridership to the NTD, many transit agencies also voluntarily report monthly ridership data to the American Public Transportation Association (APTA) on a quarterly basis. These monthly ridership data form the basis for APTA’s Ridership Reports, the basis for the annual ridership in APTA’s Public Transportation Fact Book, and are directly used in the Transportation Services Index by the Bureau of Transportation Statistics. Many agencies also report monthly ridership data to their governing bodies. These ridership data are widely used in the transit industry for policy debates and decision making both at the agency level and at the national level. In these national debates the transit industry loses credibility when it appears to the national media and highway interests that transit can’t count.
Recent research sponsored by the Florida Department of Transportation (FDOT) provides strong evidence that sample estimates of fixed-route bus ridership are greater than farebox counts for many transit agencies in the nation, and that these one-sided deviations are significant in magnitude, widespread both in the number of agencies and across agency size, and persistently present over many years. These one-sided deviations point to either serious undercounting in farebox counts or systematic non-sampling errors in sample estimates or both. The research did not provide evidence on what specific sources are responsible for the one-sided deviations or the particular nature of the systematic non-sampling errors if they are responsible.
Accurate ridership has many benefits for the transit industry. First, accurately measured ridership restores the incentive role of ridership as an allocation factor in grant allocations. Ridership in the form of passenger miles, for example, is one factor in the Federal Urbanized Area Formula Program. If the systematic errors are much greater than achievable ridership growth through improved services by transit agencies, much of the incentive role of ridership in the program is lost. Second, accurately measured ridership maintains the integrity of grant allocation formulas. Systematic errors also lead to unfair allocations that deviate from what the original formula intended. The allocation of transit formula grants is a zero-sum game in most cases. One agency’s gain is a loss to other agencies. Third, accurately measured ridership avoids unnecessary biases in policy debates and decision making. Systematic errors also have undesirable consequences in the use of the NTD ridership at the individual agency level for other purposes. The NTD ridership, for example, is commonly used by transit agencies to compare themselves to individual peer agencies in terms of ridership per unit of services provided, ridership per unit of operating cost, or just ridership per capita. Such performance comparisons are important considerations in local decisions for transit funding. These local decisions would be seriously questioned if the NTD ridership for some of the agencies involved in such performance comparisons contained systematic errors.
This research has two objectives designed to solve the problem stated above for fixed-route bus services. One objective is to identify the sources of the one-sided deviations between sample estimates and farebox counts. The other objective is to develop strategies for each of the identified sources to reduce or eliminate the errors from these sources.
ESTIMATE OF THE PROBLEM FUNDING AND RESEARCH PERIOD
Recommended Funding: $400,000
Research Period: 18 months