Health preferences and behavioural economics
A discrete choice experiment shows a person a series of made-up options and asks which they would take. Repeated across enough people and enough combinations, the pattern of answers identifies how much each feature of the option is worth in the currency of the others. It is the closest thing health economics has to a price for something that is never priced.
I have run these across organ transplantation, diabetes, prostate cancer, dementia, loneliness, telehealth, medical devices, health workforce training and vaccination policy. The clinical settings vary. The methodological questions recur.
What keeps coming back
Heterogeneity is not a nuisance. Averaging across people who want genuinely different things produces a number that describes nobody. The interesting work is in characterising the groups rather than smoothing over them.
Consistency is informative. How reliably a person answers is not just measurement error to be minimised. It varies systematically with the task, with how the task is presented, and with the person, and it carries information about how much confidence a coefficient deserves.
Equity does not sum. Adding up individual willingness to pay assumes a distributional judgement most people would reject if it were stated plainly. Work on equity and efficiency in value-based preference assessment tries to make that judgement explicit rather than implicit in the arithmetic.
Feedback changes choices. In work on heart-failure device decisions, telling people something about their earlier answers alters the answers that follow. Whether that is learning or anchoring is an empirical question, and it matters for how choice tasks should be built.