Jerry Coyne is trying to do math. A new survey out from Pew finds that, as in 2007, 61% of Americans say they’d be less likely to vote for someone who did not believe in God. Coyne thinks:
The unchanged level of disapprobation is a bit disconcerting, but at least gives the lie to accommodationist claims that vociferous atheism is turning people off. And we know that lack of religious belief is still increasing everywhere in America.
Several problems arise here, exacerbated by the generally handwavy attitude Coyne-as-blogger takes towards data and logical argument. We have to guess what hypothesis he thinks he’s falsified, and how he thinks the data do that.
First, I don’t think the reach of New Atheism has become so wide that it would necessarily have any dramatic (outside the margin of error) impact on broad surveys of public opinion. The only substantive evidence I’ve ever seen offered for the influence of New Atheism — number of book sales and number of Youtube videos viewed — are at best crude estimates. The Left Behind series sells well, but its influence on the broad culture is small, after all.
Second, the secularizing trend in American society goes back well before the New Atheists arrived on the scene, so they can’t take credit for it. The interesting question is whether they increased it or decreased it relative to the rate it would otherwise have taken (given underlying demographic trends, etc.). I don’t see anyone trying to build such a model for comparison.
Third, the 2007 survey would have been conducted after the rise of New Atheism had been in headlines for a while (Harris kicked off the movement in 2004, the phrase “New Atheism” was coined in 2006 to describe work by Dawkins, et al. from the previous few years), so whatever broad public impact it would have may well have kicked in already, rendering the comparison moot. For what it’s worth, the 1991 General Social Survey asked people’s opinions about whether atheists should hold public office; 30.6% agreed or strongly agreed they should not, and another 27.4% neither agreed nor disagreed. That amounts to 58% unwilling to endorse an atheist for public office, which aligns nicely with the results from the differently-worded Pew surveys, and from a 2007 poll from Gallup, which found 53% of Americans would not vote for an otherwise well-qualified atheist presidential candidate. That number has been essentially unchanged since a poll in 1978 (also 53%), but down noticeably from 1958–59, when it was 74–77%. Enough changes happened across the US social landscape in the 19 years between 1959 and 1978 that it’d be hard to guess which aspect of the post-’60s, post-Vietnam, post-civil rights, post-women’s lib, post-Watergate, post-sexual revolution, post-counterculture, post-Vatican II, Boomer-dominated, liberalized society made people more open to atheist candidates, or to guess why it hasn’t changed since. Two polls from after the New Atheism boomlet had peaked can hardly be expected to solve that dilemma.
Fourth, and most importantly, an increase in the number of nontheists in society ought, all else being equal, to increase support for atheist candidates. On balance, someone should be less iffy about voting for an atheist after becoming an atheist! You’d also expect an increase in willingness to vote for an atheist to follow from more people knowing they know an atheist, the effect atheist “out” campaigns and bus ads seek to promote. The lack of any such shift should be more than “disconcerting.” It verges on falsifying.
Now it’s possible that all of the folks becoming atheists were already willing to vote for atheists before their shift in religious views. But still, the constancy of attitudes toward atheist candidates implies that other theists are not becoming more liberal in their views on atheism, as the New Atheists’ “Overton Window” rhetoric would imply. (The Overton window is a data- and theory-deficient claim ginned up by a conservative thinktank looking for ways to kill public schools. It argues that the key to political change is to stake out crazypants positions to redefine the center of debate, thus expanding the “window” of acceptable views, and ultimately changing public opinion and public policy.)
There are three basic ways to explain this pair of trends: the rising secularism of society and the constant unwillingness to vote for atheists. Either the growing secular population is drawing entirely from folks already friendly to atheists, thus having no effect at all on anyone else (plausible, but not good news for folks who invested the last 6 years in a PR push for atheism), or atheist outreach has essentially turned off one person for every person it switched. Or neither of these has happened with magnitude outside the margin of error over the last 4 years. (See addendum below the fold for more.)
I happen to think the last option is most likely, so I wouldn’t make a thing out of this, except that Coyne has made a habit of pointing to constancy in polls on evolution as proof positive that … well here’s how he puts it:
American’s attitudes to evolution have been relatively unchanged… for twenty-five years…
The dominant strategy of scientific organizations engaged in fighting creationism over the past twenty-five years has been accommodationism: coddling or refusing to criticize religious people for fear of alienating those of the faithful who support evolution. This has been combined with incessant claims that science and religion are perfectly compatible [actually no, if he’s referring to the same “accommodationists” ‑Josh]. This strategy has not worked.
Accommodationists like Forrest and the National Center for Science Education have been using the “let’s-make-nice-to-the-faithful” strategy for several decades. What is the result? … American acceptance of evolution has stayed exactly where it is for 25 years. The strategy is not changing minds.
And so forth (all emphasis from the original). At some point, it’d be nice if he applied the same standard of evidence to himself that he uses in criticizing others.
To be clear, and as I’ve said before, there’s no particular reason to think that the actions of groups like NCSE should show up in polls of the general public. We know there are well-funded groups working hard to undermine evolution acceptance, and NCSE’s efforts are not targeted at the general public. NCSE and other science groups don’t take out billboards or ads on buses. They work on state science standards (with narrow audiences of educational elites), and on science teacher training, and help individual teachers and individual parents push back against creationism, and provide resources for the narrow segments of the public who want more information on evolution. General surveys of the US public aren’t going to pick up most of the effects of that work (except in a complex and highly mediated way that would be hard to poll, especially given the creationist message being broadcast to the general public).
But New Atheism is oriented toward the general public, towards shifting public opinion (in some often-ill-defined way). We should expect to see a result from those actions when we look at general surveys. Not immediately, and it’s possible the effect would always be too small to measure, or that it could fall prey to confounding factors and other difficulties associated with any attempt at measuring change in public opinion without having a control group for comparison. But with the right poll questions and the right controls, static polls point either to no effect — i.e.,Â failure ‑Â or to a positive effect counterbalanced by a backlash ‑Â also failure.
If data make any difference, it’s worth noting that the General Social Survey has asked respondents in 1991, 1998, 2004, and 2008 to “Describe your beliefs about God,” selecting: don’t believe now, but used to; don’t believe now, used to; believe now, didn’t used to; believe now, always have.
This result could, potentially, tell us something about whether there’s any backlash effect, a change in the number of conversions one way or the other, for instance.
Alas, the data are fairly murky. The numbers bounce all over, with the born-atheist contingent jumping from 1.0% to 4.6% between 2004 and 2008 (it was 3.6% in 1991 and 5.5% in 1998). There’s no way the population of life-long atheists is really fluctuating so rapidly, which makes me skeptical of the result that converts to atheism went from 3.6% of the population in 1991, to 5.5% in 1998, down again to 1.4% in 2004, and back up to 5.3% in 2008. But taking those numbers as indicative of general openness to atheism, we find that folks in 2008 were no more sympathetic to atheism than they were in 1998, suggesting that these numbers bounce around for all sorts of reasons, and there’s no obvious result attributable only to the rise of New Atheism.
It’s also worth noting that the respondents to that question had earlier been asked a more straightforward question about their theistic views, and the result mesh only imperfectly between the two questions. In 2008 when 9.9% said that they don’t believe in God (either never did or used to), only 3.1% answered that they “don’t believe” when given the alternative options: “no way to find out” (4.9% in 2008), “some higher power” (10.1% in 2008), “believe sometimes” (3.4% in 2008), “believe but have doubts” (16.9% in 2008), and “know God exists” (61.6% in 2008). Cross-tabulating the results, we find that the two “don’t believe” categories from the first question discussed tend to draw heavily from people who said they don’t believe, have no way to find out, believe in some higher power, or (less frequently) believe sometimes. But some people who say they don’t believe and never did, turn out to have started the poll by saying they know God exists.
In other words, we might have been able to look for some sort of signal in people switching from theism to atheism, or vice versa, but there’s too much going on in the survey to parse out a simple answer.