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