Repeated follow-up as a method for reducing non-trading behaviour in discrete choice experiments