Discrete Choice Experiment

They share the same statistical core, random utility theory and logit-family estimation from designed choice tasks, but differ in tradition and emphasis. 'Choice-based conjoint' is the marketing label, focused on product features, pricing, and market-share simulation, while 'discrete choice experiment' is the term in transport, health, and environmental economics, where the goal is often to value non-market goods and estimate willingness to pay for policy analysis. DCEs more frequently include a status-quo or opt-out alternative and emphasize marginal rates of substitution, whereas conjoint emphasizes market simulators. The underlying design and estimation, as set out by Louviere, Train, and Hensher, are essentially the same.

The basic conditional logit model implies that the ratio of choice probabilities between two alternatives does not depend on what other alternatives are available, a property called independence of irrelevant alternatives (IIA). This can produce unrealistic substitution, the classic 'red bus / blue bus' problem, where adding a similar alternative steals share proportionally from all others rather than mostly from its close substitute. As Train and Hensher explain, mixed logit and nested logit relax IIA by allowing random coefficients or correlated nests, giving more realistic substitution patterns, which is why credible modern DCE analyses rarely stop at plain conditional logit.

Because every attribute coefficient is a marginal utility measured on the same latent scale, dividing an attribute's coefficient by the price or cost coefficient converts utility into money: it gives the amount of money that delivers the same utility change as a one-unit change in the attribute. This marginal rate of substitution is the willingness to pay for that attribute. Hensher, Rose, and Greene treat these ratios as the central output of applied choice analysis. The figures are only as good as the design and the price range shown, however, and because choices are hypothetical they should be sanity-checked against holdouts or revealed-preference evidence.