How do respondents with uncertain willingness to pay answer contingent valuation questions?

Four elicitation methods are compared in a split-sample, contingent-valuation study valuing avoidance of episodes of ill health linked to air pollution: two discrete methods and two more-continuous methods. Respondents to a traditional payment card (PC) question gave willingness-to-pay values that were lower than those implied by dichotomous-choice (DC) responses. However, followup questions showed that DC respondents were less certain of their stated behavior than were PC respondents. When respondents were told to be "almost certain" of their responses, responses to the DC and the PC formats converged. (JEL Q21).