representativeness of week/weekend days of a dietary survey