Advanced design methodology for conjoint studies.

In recent years, discrete choice experiments have received a lot of attention by market researchers to investigate consumers' preferences for a certain product or service. It has been shown that an efficient design can significantly improve the accuracy of parameter estimation and predictions which assist companies in their decision making. In this project, I propose several design methodologies which are very promising for tackling some difficult, but important design problems such as the construction of efficient designs for estimating standard latent class models which assume that respondents in each class are homogeneous and generalized latent class models which assume that respondents are heterogeneous within-class, for solving the self selection problem where the sample used in the survey is not representative for the target population, for exploring whether including the Extent-of-Preference Information in the design and data analysis can provide extra information which can improve the parameter estimates. In addition, several recently developed sampling approaches will be compared for constructing efficient designs for mixed logit models.

Keywords:Discrete choice experiments, Efficient design, Latent class model, Respondent heterogeneity, Mixed logit model

Disciplines:Applied mathematics in specific fields, Statistics and numerical methods, Marketing, Applied economics, Economic history, Macroeconomics and monetary economics, Microeconomics, Tourism