Multifactorial Experiments in Surveys : Conjoint Analysis, Choice Experiments, and Factorial Surveys

Abstract

In multifactorial survey experiments, respondents are asked to evaluate short descriptions of objects, situations, or persons composed of different attributes (factors) that are experimentally varied in their levels. Based on these evaluations, the investigators are able to draw conclusions about the respondents’ preference structures and behavioral decisions. The high internal validity of experimental designs provides high confidence that associations between experimental stimuli and outcome measurements are causal. It is comparatively simple to study large samples of the general population or special samples such as professionals. In addition, it is possible to implement multifactorial designs with simultaneous variation of several experimental factors, which allows the estimation of (monetary) trade-offs such as willingness to pay measures. This article provides a short introduction to the general idea, advantages, and shortcomings of multifactorial survey designs. The main purpose is to contrast three different sub-techniques: conjoint analyses, factorial surveys, and choice experiments. These methods were developed within different research traditions and academic fields; thus, there are important variations in their typical design and the quality of data obtained. Readers will be provided with information on classical applications, designs, research goals, required resources, and other relevant details to facilitate selecting from among the different methods.