"Demographers estimate that approximately two-thirds of all legal abortions replaced illegal ones in the first year after legalization. Estimates are based on the change in births between 1970 and 1971 compared to the number of reported abortions in 1971 (Sklar and Berkov 1974; Tietze 1973). As noted above, Donohue and Levitt have no data on abortion for cohorts born before 1974 and thus assume a zero abortion ratio for more than half their observations. A facile argument is to assume that any error is likely random and estimates are biased downward. But this assumption is decisively contradicted by the data. As a simple example, Kansas had an abortion ratio of 414 per 1,000 live births in 1973. Donohue and Levitt assume the abortion ratio in Kansas is zero in 1972. However, data collected by the Centers for Disease Control (CDC) (Centers for Disease Control 1974) indicate that Kansas had an observed abortion ratio of 369 per 1,000 live births in 1972! Going further, I estimated the resident abortion rate in 1972 using published CDC data and the algorithm used by AGI for assigning abortions by state of residence in 1973. The correlation between resident abortion rates or ratios in 1972 and 1973 is 0.95. In other words, states with the greatest abortion ratios in 1973 had the greatest abortion ratios in 1972. By assuming the abortion ratio was zero in the 45 nonrepeal states and Washington, D.C., Donohue and Levitt build in an error that is negatively correlated with the true abortion rate. As a result, the direction of the bias is unknown."