Ideally, representative samples should be: Taken at random so that every member of the population of data has an equal chance of selection; Large enough to give sufficient precision; Unbiased by the sampling procedure or equipment.