DCE vs other preference elicitation methods
DCEs, BWS, ranking tasks, and rating scales all measure preferences but produce different outputs. The choice of method depends on what the estimates will be used for.
Comparison of DCEs with best-worst scaling, conjoint analysis, ranking, and rating methods - when each is appropriate and what each produces.
Knowledge Base -> Foundations -> Methods & Academic
Ben White, 07.07.2026
Too many methods, unclear choices
Preference elicitation methods differ in what they ask respondents to do, what statistical model is applied to the responses, and what the outputs are. The right method depends on the decision the estimates will inform, the target population, and the resources available.
Using the wrong method produces preference estimates that cannot answer the research question. WTP estimates require a DCE or contingent valuation. Attribute importance rankings require BWS or a rating task. The methods are not generally interchangeable.
How the methods compare
DCEs produce WTP estimates and are grounded in random utility theory. This makes them the preferred method for regulatory and HTA submissions where defensibility matters and where estimates must be comparable across studies. The cost is complexity - DCEs require experimental design, higher sample sizes, and more sophisticated analysis.
BWS (best-worst scaling) are a quick and easy way to estimate relative importance scores without WTP estimates. It is simpler to implement and requires smaller samples, making it appropriate for early-stage attribute prioritisation or studies where WTP is not the required output.
Rating and ranking tasks are the simplest to implement and understand but produce unwieldy ordinal data that cannot be aggregated to show relative trade-offs - key to decision modelling. As Arrow's Impossibility Theorum shows, no valid procedure exists for group decision-making under ordinal utilities. ... they are also painful for respondents who tend to defect to straight lining. As such they cannot be used in welfare analysis, nor for regulatory or HTA submissions. They may however be useful in exploratory research and concept testing.
TLDR Quick links
Choosing the right method for your study
Step 1: Define the required outputs. If WTP estimates are required - for regulatory benefit-risk assessment, HTA cost-effectiveness analysis, or transport appraisal - use a DCE. If relative importance scores are sufficient, BWS is simpler and more efficient.
Step 2: Consider the target population. Cognitively demanding populations - elderly patients, seriously ill respondents, children - may perform better with simpler tasks. A rating task with cognitive interviewing may outperform a DCE for populations with limited ability to process complex trade-offs.
Step 3: Consider the regulatory context. Check what methods are accepted by the relevant regulatory or HTA body before designing the study. Method switching late in development is expensive.
Step 4: Consider the timeline and budget. DCEs require qualitative precursor work, experimental design, higher sample sizes, and specialist analysis. If resources are constrained, a simpler method executed well outperforms a DCE executed poorly.
Worked example - method selection for a policy study
An early-stage drug development team needs to prioritise which attributes to include in a patient preference study for an FDA submission, but does not yet have sufficient clinical data to define the levels required for a DCE.
A BWS study is conducted with 80 patients to rank attribute importance. The results identify the four attributes with the strongest influence on patient choice, which become the basis for the DCE design once clinical data is available. The two studies are conducted 18 months apart, with the BWS informing the DCE rather than replacing it.
References
Louviere, J., Hensher, D. and Swait, J. (2000). Stated Choice Methods. Cambridge University Press.
Flynn, T.N. et al. (2007). Best-worst scaling: What it can do for health care research and how to do it. Journal of Health Economics, 26(1), 171–189.
Not sure which method is right for your study? Contact SurveyEngine's expert team for a no-obligation consultation.
Or Contact us at support@surveyengine.com — we're glad to help.