Topics in discrete choice experiments: Screening, framing, and sample size selection

Discrete Choice Experiments are now used in almost all fields to study choice behavior. The methods to design and analyze those experiments have been investigated and improved over the years, allowing researchers to yield more accurate and reliable information without putting a larger burden on the respondents. Despite these advancements, the literature on choice modeling recognizes that several areas are still open for improvement.  In this thesis, some of these areas were identified and solutions were proposed. Specifically, the thesis focuses on how to best model the impact of consideration screening on the estimated preference parameters. It is investigated whether it is worthwhile to gather information on the consideration sets and how to deal best with inconsistencies between the consideration and choice process. Next to that, the best model was identified for explaining the impact of framing in a discrete choice experiment on the preference for meat and meat substitutes. Furthermore, two new methods have been proposed to predict the sample size that is required in a discrete choice experiment to obtain a given power level for the statistical inference. These methods were compared with methods in the literature that also rely solely on the limited information that is typically available at the planning stage.