Diagnosing Mobility and Accessibility · Lesson 1

Define the mobility system boundary and baseline

7 min read

Lesson objectives

  • Construct a reproducible functional-area baseline and calculate a person-trip modal split without mixing people, trips, vehicles and trip legs.
Reading tools

Before choosing a bus route or cycle corridor, define the system that the intervention will change. A useful planning unit is a person's complete journey from origin to destination for a stated purpose during a stated period. A journey is not a person: one person may make several journeys in a day. Two bus boardings are not necessarily two journeys if they are legs of one journey to work. Put four boundaries at the top of the baseline: the functional area, including origins and destinations that cross an administrative line; the measured days and hours; the people and trip purposes included; and the counting unit. Without these boundaries, a later change may reflect a changed definition rather than changed behaviour. The European SUMP Guidelines use a functional urban area, participation and evaluation as planning concepts. They are used here as a method, not as a claim that one city's legal rules apply everywhere.

Build a compact data dictionary before calculating a percentage. Person journeys count complete purpose-based movements, with a stated rule for classifying a journey that has several legs. Boardings are useful for operations, but transfers can inflate the denominator of a mode-share calculation. Vehicle trips do not reveal the number of passengers without occupancy data. Vehicle kilometres are relevant to traffic-related emissions activity, yet cannot silently replace passenger kilometres. Record trip purpose, departure time, origin and destination, mode, transfers, distance or duration, and collection method. Do not merge a fare-system file and a household survey as if they observed two independent groups of travellers: one may record boardings and the other complete journeys. Retain the source, valid period and missing-value rate for each field.

Mode share is the number of eligible person journeys assigned to a mode divided by all eligible person journeys within the same boundary. Display the denominator next to the result. A survey recruited only at a station cannot yield a citywide mode share, since its sampling frame excludes those who do not reach that station. A weekday survey cannot be extended to weekends without evidence. Handle unknown mode values explicitly: either exclude them from the analytical denominator and disclose the number excluded, or retain an 'unknown' category when describing all observations. Check that categories add to the announced denominator and their rounded shares add to approximately 100 percent. A second check compares ticket boardings with survey journeys. A difference can be expected because of transfers, or it can flag a data defect. Inspect definitions before offering a causal explanation.

A citywide average is not enough. Prepare comparable rows by neighbourhood, by an ethically collected income or access group where suitable, by peak and off-peak period, and by journey purpose. Public-transport mode share can rise while a peripheral district lacks useful service; car journeys can fall because economic activity contracted rather than because access improved. For appraisal, pair mode shares with door-to-door time, traveller cost, arrival reliability and opportunities reachable. Record the baseline year, network version and definitions so another analyst can reproduce the figures. A review table should have a row for each indicator and columns for numerator, denominator, source, measurement date, owner, missing values and potential bias. This register permits correction if a source problem is found after a decision has been proposed.

When the survey is a sample rather than a census, journey weights are consequential, not decorative. Households near a bus station may be more likely to respond, while short off-peak journeys may be under-recorded. Inspect the sampling design: is the published weight attached to a household, a person or a journey? Do not multiply a person's journey by a household weight without understanding how journey weights were derived; household members make different numbers of journeys. Show the raw record count and the weighted estimate separately, and state the population year behind the weights. Compare distributions across neighbourhood, age and time of day with a population frame that may lawfully be used. If a group is markedly under-represented, examine nonresponse before reweighting. A weight cannot invent journeys never observed and does not automatically correct the absence of shift workers. When reporting a weighted 40% bus share, label it a survey estimate and report its sampling uncertainty or design limitations when available. Avoid decimal precision that the source cannot justify. Retain the raw distribution for technical review, but never combine a raw denominator with a weighted numerator in the decision indicator.

Before fixing the baseline, make three reconciliations. The internal check gives each journey one identifier and links its legs to that identifier; mode categories sum to eligible journeys. The temporal check asks whether ticketing, survey and service-network data describe comparable periods; if they do not, disclose the difference instead of calculating a pooled trend. The causal check writes at least two competing explanations for an apparent change, such as a payment-system revision, a road closure or more working from home. More boardings after new service therefore warrant investigation but do not alone prove fewer car journeys or lower emissions. An emissions claim later needs distance, displaced mode, vehicle occupancy and a factor that matches the declared accounting boundary. An equity claim needs the groups' access to opportunities, not merely a stop near their homes. Put these open questions in the baseline register rather than filling gaps with undisclosed estimates. A defensible register separates what is observed, what is unknown and what the pilot will measure.

Lesson summary

Before choosing a bus route or cycle corridor, define the system that the intervention will change. A useful planning unit is a person's complete journey from origin to destination for a stated purpose during