Introduction

Choice experiments consist of a sample of choice sets selected from the universal set of all possible choice sets that satisfy certain statistical properties. The key statistical properties relevant to the design of choice experiments are identification and precision, which must be considered together with non-statistical properties such as realism and complexity.

Chapter 3 introduced two types of choice alternatives: generic and alternative-specific. The former type have no specific name or label, but rather are members of a class of alternatives. The latter are alternatives to which a name or label naturally applies, such as brands of detergent, names of retail chains, names of holiday destinations, etc., and it is the label itself which is the object of choice. Thus, generic alternatives are members of a general class of options, whereas alternative-specific alternatives are specific members of a general class.

Not surprisingly, the preceding discussion implies that there are two general types of choice experiments: (1) labelled (alternative-specific), and (2) unlabelled (generic). There are two general ways to design choice experiments for both types: (a) sequentially design alternatives and then design the choice sets into which they are placed, and/or (b) simultaneously design alternatives and assign them to choice sets. The types of effects that can be estimated from the two main types of choice experiments differ by type.

As discussed in chapter 3, if parameters representing the effects of attributes and/or individual characteristics are constant across alternatives, we say that such effects are generic for those alternatives.