I don't know about that. They examine two non-randomly chosen group of people and make extrapolations from that. (After all, there's going to be a reason the women denied the abortion were in fact denied it.)
The conclusions would be a very useful point if valid, but the shitty methodology means the conclusions can't be trusted.
This information was created/compiled by people who explicitly support these policies. You've stated you wouldn't accept data which could have been corrupted by motivated reasoning; well, this data exists in exactly the state, in the opposing ideology, as the data you said you wouldn't accept.
Sure. But if you wouldn't accept parallel data from the other side, why would you expect them to accept this data from ours?
What I'm getting at is that this data doesn't actually add anything substantive to the debate. It looks great from where you're sitting, but you're not motivated to seriously question it. The most obvious question to me - what the difference between the control group and the test group was, given that they weren't sorted into these groups randomly, but on some policy basis - remains unanswered.
This data is perfect in one sense - because it has a serious unresolved issue, it allows both sides to convince themselves that its existence proves them correct. Those whose position it supports can take it as-is, and those whose position it contradicts can take the obviously poor methodology (unspecified policy-based sorting) to be proof that the output of the obviously good methodology (truly random sorting) was discarded by its creators.