fresh random samples from the same target population, asked the same questions at each time point — are preferable to panel designs, for several reasons: