Dominant alternative design in health preference DCEs
A dominant alternative is one that is strictly better on every attribute. Including one by accident destroys the information value of every choice set it appears in.
What dominant alternatives are, how they arise, how to detect them, and how to remove them from a DCE design.
Knowledge Base -> Experiment Design -> Health
Ben White, 07.07.2026
What makes an alternative dominant?
An alternative dominates another if it is strictly better on every attribute simultaneously. Rational respondents always choose the dominant one regardless of preferences. The choice provides no information about trade-offs.
Dominant alternatives arise most commonly from poorly chosen level ranges. If the best level of every attribute is assigned to the same alternative, or attribute levels are correlated to produce consistent superiority, dominance results.
Why dominant alternatives are the best consistency check
Choice sets with dominant alternatives do not contribute to parameter estimation - they are wasted survey space. If respondents notice one alternative is always obviously better, they may disengage from the choice task entirely.
Dominance tests should be run on every experimental design before deployment. The test is simple and takes seconds. Missing a dominant alternative in a 300-respondent study wastes money in data collection.
TLDR Quick links
Constructing a dominant alternative in SurveyEngine
Step 1: For each choice set, compare each pair of alternatives attribute by attribute. Flag any set where one alternative is better than or equal to the other on all attributes and strictly better on at least one.
Step 2: Remove or replace dominated alternatives. If the design generator produces dominance, reduce the spread of levels or add a dominance constraint.
Step 3: Check for near-dominance. An alternative better on 4 of 5 attributes creates a choice most respondents make the same way regardless of preferences.
Step 4: A deliberately designed dominated set serves as a consistency check. Label it explicitly in the analysis plan.
Step 5: Test with cognitive interviews. Respondents often interpret attribute descriptions differently from the researcher, creating effective dominance that passes automated checks.
Worked example - dominant alternative in HEOR study
A health preference study generates a D-efficient design with 12 choice sets. Post-generation dominance checking reveals one set where Alternative A is better on all 4 attributes. The choice set is removed and replaced.
Cognitive interviews subsequently reveal respondents interpret 'monitoring frequency' differently - monthly monitoring is seen as more reassuring than burdensome. The attribute description is revised and the design regenerated.
References
Lancsar, E. and Louviere, J. (2008). Conducting discrete choice experiments. PharmacoEconomics, 26(8), 661–677.
Ready to embed a dominant alternative in your DCE? Log in to SurveyEngine and follow the consistency check tutorial.
Or Contact us at support@surveyengine.com — we're glad to help.