Recruiting commuters for transport mode choice SP studies

Commuter samples need careful definition - commuting by what mode, how often, over what distance?
Correct sourcing and screening criteria determines the validity of your mode choice estimates.

This article explains how to define, screen, and recruit commuter samples for transport SP studies, including route-specific recruitment and representation of low-frequency modes.

Ben White, 07.07.2026

Defining your commuter population

Transport mode choice studies are only as good as the population they represent. A study of commuter mode choice that recruits a general population sample will include retirees, part-time workers, remote workers, and students whose travel behaviour is not relevant to the research question. The screening criteria for commuter recruitment need to be precisely defined before recruitment begins.

Common screening dimensions include: commuting frequency (daily vs weekly vs occasional), trip purpose (work vs education vs other), mode of travel, distance, and corridor (for route-specific studies). Getting these dimensions wrong - or leaving them undefined - produces a sample that cannot answer the research question.

Representation of all modes in the sample

A mode choice study that recruits from an online general population panel will over-represent car users (who are more likely to be at home to complete the survey) and under-represent public transport users (who complete their commute during the survey window). This sample composition bias directly affects the utility estimates.

For corridor-specific studies, on-site recruitment - at train stations, bus stops, or park-and-ride facilities - provides better mode representation but is logistically complex and expensive. Quota-based panel recruitment with specific mode quotas is the most practical approach for most transport studies.

SurveyEngine's sampling team manages mode quotas across panel providers, recruiting to specific targets for car, public transport, cycling, and walking commuters to match the corridor modal split.

TLDR Quick links

SurveyEngine transport resources

SurveyEngine respondent sourcing

Open a free account

Setting up commuter recruitment with SurveyEngine

Step 1: Define your commuter eligibility criteria. Specify: commuting frequency (minimum x times per week), trip purpose, geographic corridor or origin/destination area, current mode, and any relevant trip characteristics (minimum distance, etc.).

Step 2: Set mode quotas reflecting the corridor modal split. Research the existing modal split for the corridor being studied. Set panel recruitment quotas to match or slightly oversample underrepresented modes.

Step 3: Design the screener. The screener should confirm commuting frequency, corridor, and mode. Avoid asking about mode early in the screener - some respondents may claim to use a more research-relevant mode to qualify.

Step 4: Plan for revealed preference data collection. The screener can double as the RP data collection module, collecting journey time, fare, and other relevant attributes for each eligible respondent's most recent commute.

Step 5: Monitor sample composition during fieldwork. Check the mode composition and demographic composition of completions daily during fieldwork. Adjust quotas if the sample is drifting from the target composition.

Worked example - bus rapid transit corridor study

A transport authority studying demand for a proposed BRT corridor recruits 400 commuters in the corridor. Mode quotas are set to match the existing modal split: car 55% (220), bus 25% (100), train 10% (40), cycling 5% (20), walking 5% (20). Eligibility criteria: commutes in the defined corridor at least 3 days per week, works or studies within 500m of the BRT route.

Recruitment through panel achieves quotas for car and bus quickly but struggles to fill cycling and walking quotas. Supplementary on-site recruitment at cycling hubs achieves the cycling quota; the walking quota is partially relaxed to include respondents who walk to a bus stop within the corridor.