3 As in my original article, I restrict my sample to women over age 35 who had ever giving birth (have information on age at first birth) and who were born from 1935 to 1959. I exclude allocated values on age at first birth. There are two changes to my original sample: I include women over age 44 in the newly constructed sample (excluding them changes the results little). I also restrict my attention to women who had ever been married by the time of observation, because the June CPS only sampled never-married women in a subset of years. In practice, this last sample restriction eliminates 2531 observations (2.2 percent) of the 115,434 observations with information on the age at first birth. The final sample contains 112,903 observations. Stata code to generate table III’s summary statistics is posted on my website (http://www-personal. umich. edu/~ baileymj).