Discrete Choice

Comparison with Other Stated Preference Methods

Stated preference methods, in which respondents value hypothetical descriptions of products or choices, are useful in valuing nonmarket goods, such as health. They are useful in situations in which the market for a good, or for the full range of attributes of a good, does not exist and “revealed preference” studies cannot be conducted. In a revealed preference study, preferences are estimated by observing the actual choices that have been made in a real-world setting. For example, the relative value of individual attributes of automobiles, such as size, color, make, and model, could be measured by analyzing retrospective data on automobile sales prices along with the specific characteristics of the automobiles sold. For a new model, a revealed preference approach would not be possible since data are not yet available for the new model; instead, a stated preference approach could be used. Other stated preference methods typically used to value health outcomes are health state utility assessment or contingent valuation. Elicitation techniques for these stated preference methods include standard gamble, time trade-off, or willingness to pay. Compared with these methods, DCEs can be used to value health, nonhealth, and process attributes and provide information about the tradeoffs between these attributes. DCEs can also be used to value willingness to pay for an attribute, whereas traditional methods provide a single numerical rating for the whole service.

All stated preference methods have the limitation that the valuation task asks about hypothetical choices and, therefore, may not fully predict future choices. Using stated preference methods can often provide a valuable starting point for further research given the difficulty of obtaining preference data on nonmarket goods. All stated preference methods allow data to be collected on programs and interventions while they are still under development, similar to how studies might be conducted to develop new consumer products. Once a program or intervention has been introduced, additional research could combine revealed and stated preference data to provide even more detailed information about user preferences.

Understanding preferences for different aspects of health and health interventions and incorporating these values into clinical and policy decisions can result in clinical and policy decisions that better reflect individuals' preferences and potentially improve adherence to clinical treatments or public health programs.

Terminology

The terms discrete choice experiments or conjoint analysis are typically used to describe a type of stated preference method in which preferences are inferred according to responses to hypothetical scenarios. These terms are often used interchangeably. Conjoint analysis comes from marketing applications and DCEs from transportation and engineering applications. The common element of DCEs and conjoint analysis is that they both allow the researcher to examine the trade-offs that people make for each attribute, attribute level, and combinations of attributes. They differ in that the term conjoint analysis is more generally used to refer to a method whereby the respondent rates or ranks a scenario and DCEs involve a discrete choice between alternative scenarios. DCEs, and the related approach of conjoint analysis, have been successfully applied to measuring preferences for a diverse range of health applications, and the use of these approaches is growing rapidly.

...