Contingent Valuation: Indiscretion in the Adoption of Discrete Choice Question Formats?

2017-06-07T05:21:03Z (GMT) by Smith, Richard D
Contingent valuation (CV) refers to a hypothetical survey method of valuing the benefits of an intervention in monetary terms by estimation of the individuals maximum willingness-to-pay (WTP). Currently, an issue in the application of CV methods is the technique used to elicit this monetary valuation. Historically, the favoured technique has been the `open-ended' questionnaire, where the respondent is asked directly for their maximum WTP for the commodity being valued. However, in recent years there has been a move away from the use of this open-ended technique, towards the use of discrete choice questionnaires (also referred to variously as closed-ended, binary or dichotomous choice questionnaires; as well as referendum surveys if the median, rather than mean, WTP is the desired measure of value). In contrast to open-ended questionnaires, discrete choice questions offer the respondent a single value (bid), which they either accept or reject. By varying this single bid across various sub-samples a demand curve for the commodity is estimated, and from this the maximum WTP calculated. The basis for the use of discrete surveys in preference to open-ended seems to rest on some combination of: (i) a belief that this technique offers a more `realistic' market, and will therefore lead to more valid responses (a truer estimate of actual WTP by respondents); and (ii) the supposed tendency of discrete surveys to yield higher response rates, through reduced mental demands (especially for mailed surveys). This paper presents a review of the use of discrete versus open-ended survey techniques, and addresses these issues as well as others of importance. It is concluded that, although there are issues yet to be resolved concerning the degree of bias within the open-ended technique, there appear to be substantial additional issues with the use of discrete survey techniques. While the discrete choice questionnaire seems to have been favoured, major problems still remain in its implementation. These problems arise from the fact that, simply, discrete choice survey design and analysis is very complex. Survey design elements include total sample size, bid range, specific bid levels, allocation of the total sample among the bid levels, and form of statistical model used to analyse such data. None of these issues has by any means been clearly resolved. This author would therefore suggest that the choice of discrete versus open-ended techniques is by no means settled, and is potentially a red-herring in the search for `valid' WTP values derived from CV surveys. There appears no reason why the open-ended survey should be summarily dismissed, and research may better be targeted to refinement of the open-ended approach to reduce bias. In the meantime, it is recommended that caution be exercised in the seemingly indiscreet adoption of the discrete choice question approach.