Illusory Wage Differentials: Reply
回复了关于工资差异的讨论，指出1966年两组在职培训估计差异较小（0.55美元，标准误0.67），因此早期测量的工资差异低估了真实差异，但低估程度存疑。
We are quick to add, however, that the difference in estimated OJT across the two groups is not large in 1966. In fact, H(1966)ID=1 H(1966)|D=O = $0.55 (with a standard error of $0.67) so that the difference is relatively small and the standard error is quite large. The conclusion is then that for this early period, the measured wage differential underestimates the true differential, but the extent of that understatement is questionable. [p. 559]