By Scott Lemieux
People interested in subjects like the effects of question wording and selection on public opinion surveys should definitely read John Sides and Nate Silver on recent abortion surveys. Their analysis brings up a couple important puzzles:
- When is there enough evidence to determine that there has, in fact, been a substantial change in public opinion, in a way which might affect the political calculus of public officials, the outcomes of initiative votes, etc.? When public opinion has been as stable as it's been on the abortion issue in the United States, it's prudent to be skeptical about the results of individual polls representing a shift, but of course it's also important to be open-minded. The next round of surveys will be interesting, and I'll try to follow up.
- Even on some fairly straightforward issues, measuring public opinion can be difficult, with subtle changes in question wording producing different results. Abortion is even trickier, particularly given what appears to be a rough public consensus in favor of legal-but-regulated abortion. Should questions ask general or specific? Should they ask about the morality of abortion or about specific legal regulations of abortion? Because of this, it's probably especially useful to have multiple polling groups asking their own particular set of questions, even if it makes long-range comparisons more difficult.
The bottom line is that any conclusions in any direction have to be tentative. It will be interesting to follow this issue in the coming months.
