Too many theories, too few facts? What the data tell us about the link between span of control, compensation and career dynamics