The Gallup Pro-Choice Number

by John Sides on May 23, 2012 · 7 comments

in Public opinion

A new Gallup poll shows that the percent of Americans calling themselves pro-choice has fallen to 41%.  In 2008, when that number hit 42%, there was a predictable flurry of news attention.  So I want to call attention to what I wrote then. In short, this “pro-life” vs. “pro-choice” question obscures the true nature of American attitudes toward abortion.  Support for the right to abortion depends strongly on the circumstances of the pregnancy.  They cannot be summarized with the labels “pro-choice” and “pro-life.”

Moreover, and most importantly, more nuanced measures show little of the fluctuation that Gallup’s pro-choice vs. pro-life measure shows.  Indeed Gallup’s new poll confirms this finding:

However, it is notable that while Americans’ labeling of their position has changed, their fundamental views on the issue have not.

{ 7 comments }

Logan May 23, 2012 at 10:36 am

I don’t quite like their question of whether or not abortion should be legal in “some” cases.

Americans could have shifted wildly on which cases those are precisely, and Gallup wouldn’t know. If Americans used to favor abortions in the first and second trimester but not the third, but now favor abortion only in the case of rape or incest, those would both manifest themselves on this Gallup survey as a respondent indicating that they favor abortion in “some” cases, despite the fact that the two positions are wildly divergent.

John Sides May 23, 2012 at 10:47 am

I agree, Logan. The survey questions discussed in my first post are much more helpful in this regard.

Clyde Wilcox May 23, 2012 at 3:55 pm

But John, the GSS data suggests that those who are “between two absolutes” now support fewer specific circumstances than in the past, and this is especially true for the youngest cohort.

John Sides May 23, 2012 at 6:36 pm

Clyde: Those shifts in the GSS data aren’t large, though, correct?

obelmann May 23, 2012 at 5:40 pm

Gallup claims a +/- 4% total sampling error on the survey.

When Gallup concisely stated that “41% of Americans who now identify themselves as “pro-choice” is down from 47% last July” — the actual Gallup numbers were an inconclusive, overlapping range (at best):

37% – 45% ‘pro-choice’ this year

43% – 51% ‘pro-choice’ last year

So it’s possible that 45% (for example) is the true population measure of ‘pro-choice’… and indeed has not changed at all since last year, according to Gallup’s own results.

Of course, there are many other likely sources of error in any survey… increasing the margin of error VERY significantly.
Certainly Gallup’s subjective choice of position labels and question wording are a major bias source, as discussed above.

Gallup claims a “random sample” as the core basis of this telephone poll. It was not a random/probability sample, but a convenience sample, as executed; factual generalization to the American population indicated by Gallup– is not possible with the Gallup methodology employed.

Best to ignore Gallup and their products entirely.

Andrew Gelman May 23, 2012 at 6:08 pm

Yeah, we should definitely ignore Gallup. After all, Rush Limbaugh says they’re biased, something about “upping the sample to black Americans.”

Matt Jarvis May 25, 2012 at 4:14 pm

Ah, but ignoring the elephant in the room on margins of error and differences of 1-5%…..

Is it meaningful when people adopt a different label for their opinions? Take lib/con. Liberalism has been declining pretty sharply in surveys for a long time (and conservative rising), but there actually isn’t anywhere near as much movement on specific issues.

Comments on this entry are closed.

Previous post:

Next post: