Polling about abortion, like polling about a lot of issues, is beset by some pretty vague questions. Whether you’re “pro-life” or “pro-choice.” Whether abortion should be legal or illegal in “all cases” or “most cases,” or under “any circumstances” as opposed to “certain circumstances.” Certainly, those questions can serve a purpose.
But we also know that public opinion about abortion depends on what those cases or circumstances are – including when the abortion occurs and the reasons for the abortion. A new paper by Natalie Hernandez, a Ph.D. candidate in political science at Yale, shows just how important timing and reasons are for measuring opinion. A second, newly published paper by Hernandez and co-authors shows how this measure matters.
In the first paper, Hernandez does something I’ve always wanted to do: ask people about timing and reasons simultaneously – or what she calls the “reasons by weeks” measure.
For timing, people can choose whether abortion should be never legal or legal up to 6 weeks of gestation, 12 weeks, 15 weeks, 24 weeks (roughly, viability), or 40 weeks (no limit).
For reasons, people were given these:
- If the pregnant woman develops a life-threatening medical condition that can only be treated if the pregnancy is ended.
- The fetus is found to have a serious physical or mental disability, such as Down syndrome, that will have implications for the child’s life.
- The pregnancy is the result of rape or incest.
- The mother can’t afford to have the child.
- The mother believes having the child would interfere with her educational or career aspirations.
- The mother doesn’t want a child of that specific sex.
Essentially, what the reasons-by-weeks measure does is ask you, for each reason, up to which point in the pregnancy you would permit an abortion for that reason (which, again, could be never or always). This is a graph of the results (from the appendices of the second paper):
Unsurprisingly, the percentage who say “never” is rare when the mother’s life is threatened, the fetus has a disability, or the pregnancy is the result of rape or incest. Support for abortion because of the sex of the fetus is much more limited.
Notably, the timing of the abortion does appear to matter, whatever the reason. For example, although a majority would permit an abortion if the mother couldn’t afford to have the child, there is disagreement about the timing. Even in a case like rape or incest, timing matters. Only a minority of people choose “never” or 40 weeks (i.e., always). Most want some limit based on gestational age.
Hernandez goes on to show how the reasons-by-weeks measure improves on standard polling questions with vague categories. First, she shows that lots of people who say abortion should be “always legal” actually would place significant limits on abortion, depending on the reason and timing. And plenty of people who say “never legal” would allow abortion in some cases, especially in cases of threats to the mother’s life.
Moreover, Hernandez shows that Democrats and Republicans who give the same answer on the standard question give very different answers to the reasons-by-weeks question. A Democrat and a Republican who each say “always legal,” for example, apparently have pretty different views of what “always” means.
In the second paper, Hernandez and co-authors make three important points:
- Abortion attitudes measured with the “reasons-by-weeks” measure are coherent. Here’s an example. In this paper, respondents were not only asked the reasons-by-weeks question but also, in an interview 8 weeks later, were asked how much they favored different gestational limits on a 0-10 scale, given a randomly chosen reason for the abortion. What you don’t want is for people who said in the first interview that they favored a 24-week limit in cases of a fetal defect to then tell you later than, actually, they give a 24-week limit an 8 on the scale but give a 40-week limit a 9. This would suggest they should have said 40 weeks in the first interview. This sort of thing generally does not happen.
- Abortion attitudes are stable. Across the two interviews, respondents tended to give very similar (if not always identical) answers. The stability across these 8 weeks was comparable to the stability of presidential approval. There is also considerable stability across a much longer time span (4 years), using a different survey and one of the standard polling questions.
- Abortion attitudes are consequential. Hernandez and co-authors show that abortion attitudes were associated with shifts in presidential voting between 2020 and 2024 – especially for those who report confidence in their views of abortion and those who say that the issue of abortion is important to them.
To me, another important takeaway here is that abortion attitudes are simply complex. It’s easy to imagine a binary where people fall into clearly delineated pro-choice and pro-life camps, and thus abortion becomes yet another indicator of a highly polarized, red vs. blue America.
In reality, however, people’s thinking about abortion depends on any number of considerations. If you want yet another bolded term, I’d say that people’s attitudes about abortion are ambivalent. And coming to grips with that requires more than the usual poll questions.
