Why use a DCE for transport research?
Transport policy analysis requires monetary values for time savings, reliability, and comfort. DCEs are the standard method for estimating them.
In this article, you'll discover why DCEs are the standard method for transport preference research and what distinguishes transport applications from other DCE domains.
Knowledge Base -> Foundations -> Transport
Ben White, 07.07.2026
Why transport research needs DCEs
Transport cost-benefit analysis requires monetary values for attributes that are not traded in markets - travel time savings, reliability improvements, comfort, safety. These values must be estimated from preferences, and they must be comparable across studies and modes to be usable in national appraisal frameworks. They also need to entertain the reality that people have different incomes and typical journey times.
The UK WebTAG guidance, Dutch MKBA framework, and Nordic transport valuation studies all rely on SP-based estimates for core transport values. The value of travel time, the value of reliability, and the value of frequency are the three most important outputs of transport preference research.
DCEs in transport appraisal
DCEs produce WTP estimates directly comparable with market prices, expressed in monetary units per minute, per incident, or per service unit. This comparability is essential for CBA. Rating scales and importance rankings cannot produce this output.
The theoretical grounding in random utility theory also ensures that the estimates are consistent with the welfare economics foundations of CBA. Mode choice models estimated from DCE data can be embedded in wider transport demand models to forecast behaviour under policy scenarios.
Setting up a transport SP study in SurveyEngine
Step 1: Anchor the SP design to the respondent's actual journey. Pivot designs use the respondent's reported current journey as the reference point, varying travel time, cost, and other attributes as percentage changes or absolute offsets. This produces more realistic trade-offs than designs with fixed attribute levels.
Step 2: Use mixed logit for estimation. The MNL model assumes all respondents have the same preferences, which is implausible for mode choice. Mixed logit models heterogeneous preferences and is standard for transport SP analysis.
Step 3: Test for IIA violations. The independence of irrelevant alternatives assumption embedded in logit models may be violated in mode choice contexts where alternatives are correlated. Test with nested logit or cross-nested logit specifications.
Step 4: Report values with confidence intervals. Value of travel time estimates without confidence intervals are not acceptable for appraisal purposes. Mixed logit produces distributional estimates; report the mean, median, and standard deviation of the WTP distribution.
Worked example - intercity rail mode choice
The Norwegian Ministry of Transport commissions a national travel cost survey to update the national transport appraisal values. A SP experiment is conducted with 3,200 respondents covering car, rail, bus, and air modes, with pivot designs anchored to each respondent's most recent intercity journey.
Mixed logit estimation produces updated values of travel time, reliability, and comfort for each mode and trip purpose. The values are calibrated against RP data from the national travel survey and adopted in the national CBA framework for use in road, rail, and aviation investment appraisal.
References
Hensher, D.A., Rose, J.M. and Greene, W.H. (2015). Applied Choice Analysis, 2nd edition. Cambridge University Press.
DfT (2023). TAG Unit A1.3: User and Provider Impacts. Transport Analysis Guidance.
Planning a transport SP study? Contact SurveyEngine to discuss pivot design, RP/SP combination, and value of travel time estimation.
Or Contact us at support@surveyengine.com — we're glad to help.