The perils of drawing conclusions from a sample of only two