Efficient Online Choice Experiments: incorporating screening rules in the selection of sequential choice sets.

When people choose between alternatives, they make trade-offs between the different characteristics of the alternatives. Discrete choice experiments are conducted to get information about these trade-offs: the respondents are confronted with choice sets consisting of a number alternatives and have to indicate their most preferred alternative.

Such experiments are often conducted online which offers the opportunity to use individually tailored choice sets. Optimal and fast generation of such choice sets has been the topic of previous research projects.

The standard model assumes that people make trade-offs between all the attributes and that they do this for all alternatives. But people often use screening rules to simplify the task. Mainly two types of screening rules have been modeled. Common consideration-set screening rules assume that respondents first evaluate whether an alternative is acceptable and otherwise delete it from the choice set. In attribute non-attendance screening, respondents are assumed to completely ignore part of the attributes. Neglecting such screening behavior when analyzing choice data will lead to unreliable part-worths.

The main goal of this project is to take screening rules into account while generating individually tailored choice sets in online choice experiments. By sequentially updating the information on the screening process, the choice sets can be tailored even further which improves the information about the preferences.