Joint use of attribute importance rankings and non-attendance data in choice experiments

The joint and alternative uses of attribute non-attendance and importance ranking data within discrete choice experiments are investigated using data from Lebanon examining consumers' preferences for safety certification in food. We find that both types of information such as attribute non-attendance and importance rankings improve estimates of respondent utility. We introduce a method of integrating both types of information simultaneously and find that this outperforms models where either importance ranking or non-attendance data are used alone. As in previous studies, stated non-attendance of attributes was not found to be consistent with respondents having zero marginal utility for those attributes.