Consultancy Research - Delivery of Discrete Choice Experiments

In brief, a range of COPD treatments will be characterised by their component features, known as attributes. Each attribute will then be represented by a range of options, known as levels. These attributes and levels can be formulated into a range of different scenarios, each of which represents a different combination of levels for each attribute. Respondents are asked to make a series of choices within scenario sets and, through their choices, inferences can be made about their relative preferences across attributes/levels. While respondents are only subjected to a small number of possible combinations of these levels, Bayesian statistical methods can be applied to determine the impact on treatment preference for any combination of attribute levels.

A first step of the DCE is to define the most relevant attributes for relevant COPD treatment/ management strategies and, for each attribute, to define the levels to be tested. It is recommended to use a maximum of eight attributes with two to three levels. With more than 8 attributes, there is a risk that respondents start to use simplifying heuristics to answer questions and they will typically only focus on 2-3 attributes important to themselves when making choices between scenarios, thereby potentially reducing the robustness of the results.