Target was able to figure out when a woman had likely become pregnant. So it's probably not impossible to figure out when a woman has had an abortion.
Target is not a social network, but I will play along. Target was able to discern pregnancy because of purchases made or not made over a span of months, many months. In the case of abortion there is a significant motivation to not wait; namely abortion laws but also social stigma. Expectant mothers purchased folic acid and other vitamins. There is not a recommended vitamin routine for an abortion. The same thing can be said for many other purchases made by expectant mothers. How are you going to separate abortions from miscarriages?
Using a social graph to figure out if a user is pro-choice is probably not that hard. I'd love to hear how this guy is going to use it to determine if someone has had an abortion. But I'm afraid to know what assumptions he makes about women who have abortions in order to come to this conclusion.