Table 1 Stated preference elicitation techniques
Method | Underlying theory | Measurement | Analysis | |
|---|---|---|---|---|
Contingent valuation | Rooted in welfare economics, namely in the neo-classical concept of economic value based on individual utility maximization. Contingent valuation surveys directly obtain a monetary (Hicksian) measure of welfare associated with providing a good/service. | • Direct query about willingness-to-pay or willingness-to-accept • For example: “Please state the largest amount you are willing to pay for the good/service.” | Various statistical methods depending on study aims (e.g., minimum, maximum, mean, and regression) | |
• Dichotomous question with reference to a given price • For example: “Would you be willing to pay X € for the good/service?” | Binary choice models (e.g., binary logit, binary probit) | |||
Bidding game [58] | • Dichotomous question in form of an auction • For example: Would you be willing to pay X € for the good/service? Would you be willing to pay X + Y € (X-Y €) for the good/service? | Various statistical methods depending on study aims (e.g., minimum, maximum, mean, parametric, and non-parametric tests) | ||
Self-explicated approaches [60] | No underlying economic theory | • Unacceptable attributes are removed • The level of each attribute is evaluated on a desirability scale (e.g., 0 the worst level of the attribute and 100 the best) • The respondent is asked to allocate, for example, 100 points across the attributes to reflect their relative importance • In stage 1 or 2, different combinations of comparative or non-comparative methods could be used | • Part-worth: multiplying the importance weights (stage 2) with the attribute and level of desirability ratings (stage 1), additive assumption • Overall utility: sum of the part-worth | |
No underlying economic theory | 1. The attributes that contribute to the problem must be identified and arranged in a hierarchy according to aims, attributes, and alternatives 2. Hierarchy levels are assessed by paired comparisons 3. A matrix is created using pairwise ratios and the relative weights are calculated 4. Relative weights of the levels in stage 3 are aggregated | Calculating the relative weights of hierarchy levels with the eigenvector method | ||
Conjoint Analysis | Not choice-based [62] | Depends on the method and approaches used | Variety of methods and approaches, such as rating or ranking of different alternatives | • Interval scaling (e.g., OLS* regressions) • Ordinal scaling (e.g., MONANOVA*, PREFMAP*, LINMAP*, ordered logit-/probit-regressions) |
Random utility theory | • Choice between two or more discrete alternatives (selection of most preferred alternative) • Alternatives are described by a set of attributes and each attribute takes one of several levels | • Two alternatives in the choice set: binary discrete choice models (e.g., binary logit, binary probit) • Three or more alternatives in the choice set: multiple discrete choice models (e.g., multinomial logit, nested logit, mixed logit, multinomial probit, heteroscedastic extreme value) | ||
Standard gamble [64] | Utility theory by von Neumann and Morgenstern | • Choice between a fixed health status and a lottery with the probability p to obtain the best possible outcome and the probability 1 - p to obtain the worst possible health status • For example: a chronic health state preferred over death: 1. Respondents are offered two alternatives: (A): two possible outcomes; aa) the subject lives in a good health with the probability p for a fixed time t, or ab) the subject dies immediately with the probability 1 - p (B): the subject lives in a fixed health status i for the rest of his/her life t 2. Respondents’ indifference point is located by varying the probability p | • For example: chronic health state preferred to death: At indifference point, the required preference score for health state i is h i = p | |
No underlying economic theory | • Trade-off between life years in a state of less than perfect health and a shorter life span in a state of perfect health • For example: a chronic health state preferred over death: 1. Respondents are offered two alternatives: (A) health state i for time t, followed by death; (B) full health for time x < t, followed by death 2. Respondents’ indifference point is located by varying the time x | • For example: chronic health state preferred to death: At indifference point the required preference score for health state i is given: h i = x/t | ||
No underlying economic theory | • Direct rating on a line with or without internal markings • For example: a chronic health state preferred to death: 1. Respondents receive information about a batch of chronic health states, age of onset, the age of death, and two reference states (“full health”, “death”) 2. Respondents are usually asked to select the best and the worst of those health states 3. The remaining health states are placed on the rating scale relative to each other | • For example: a chronic health state preferred to death. Preference value for health state is the scale value of its placement | ||