More about the Pew poll on evolution acceptance

January 7, 2014 • 1:30 am

NOTE BY JAC:  I still am baffled by the Pew’s finding that Republicans seem to have become more creationist between 2009 and 2013, for the Gallup Poll shows the 20% disparity already in 2008.  In that poll, the percentage of young-earth creationists was 60% among Republicans, 38% among Democrats, and 40% among Independents. The gap that Pew says is widening, then, appears in the Gallup data to have been that wide already five years ago.  Since the issue is the same, human evolution, I can only attribute it to different sampling techniques or, as Greg suggests below, to the order in which questions were asked.


by Greg Mayer

I’ve already posted twice on the Pew poll on evolution acceptance, first to bring it to WEIT readers’ attention while noting the disparity between the Pew poll and Gallup’s results on the same issue, and then to note an erroneous criticism of the poll by Dan Kahan. I’d like to note three further developments.

The most interesting is a further report from Pew written by Cary Funk (if you look at nothing else mentioned here, look at this report), I’ll mention two other items first.

First, Charles Blow at the New York Times, in a piece entitled  “Indoctrinating Religious Warriors“, considers what the poll says about the political and religious landscape of America. He’s saddened by the fact that more Republicans now accept creationism than evolution:

In fact, this isn’t only sad; it’s embarrassing.

I don’t personally have a problem with religious faith, even in the extreme, as long as it doesn’t supersede science and it’s not used to impose outdated mores on others.

But as Blow well knows, the only religious extremists that make the news are precisely the ones who want their faith to supercede science and to impose their mores on the rest of society. He attributes its recrudescence to the strategy of the Republican party:

But I believe that something else is also at play here, something more cynical. I believe this is a natural result of a long-running ploy by Republican party leaders to play on the most base convictions of conservative voters in order to solidify their support. Convince people that they’re fighting a religious war for religious freedom, a war in which passion and devotion are one’s weapons against doubt and confusion, and you make loyal soldiers.

There has been anti-science propagandizing running unchecked on the right for years, from anti-gay-equality misinformation to climate change denials.

Second, Andrew Sullivan, in “Converting to Belief in Evolution“, has looked at the poll again, and points to Karl Giberson (whom Jerry also commented on) asking whether evangelical Christianity’s antagonism to science will push young people away from evangelical Christianity. Giberson found this prospect “alarming”, but evidently Andrew doesn’t. (As a gay Catholic who accepts at least theistic evolution, Andrew has longstanding political and theological differences with evangelicalism.)

Finally, Dan Kahan has accepted that his chief argument against the Pew poll—that its reported numbers must be incorrect—is wrong. He did so in response to a commenter on his site, who provided a hypothetical numerical example refuting Kahan’s assertion. I showed that Kahan was in error with a general argument about the statistics of sums, but a concrete counterexample is also a satisfying form of refutation. But most importantly, Pew, without mentioning Kahan, has released a detailed answer to the question that Kahan thought indicated numerical hanky-panky: “If the views of the overall public have remained steady, and there has been little change among people of other political affiliations, how does one account for the Republican numbers? Shouldn’t the marked drop in Republican believers cause a decline in the 60% of all adults who say humans have evolved over time?” The answer is of course ‘not necessarily, and, in fact, not in this case’.

Kudos to Kahan for accepting the invalidity of his mathematical argument, but, oddly, he continues unchanged in his animus toward the Pew poll and one of its striking findings (see the updates and a further post here). As I said, his reactions to the poll seem to be “merely expressions of his own prejudices”, and not terribly dependent on the actual poll results, since he continues to hold them although though his conclusions on the poll have been shown to be in error. The whole sequence of what he writes about the poll is a wonderful example of the type of reasoning which, in another context, Sam Wang of Princeton has called “motivated reasoning“.

The new Pew report (which, as I said, is the thing really worth looking at here), clearly answers Kahan’s doubts. Here’s their table nicely illustrating, neither generally nor hypothetically, that there’s nothing wrong with their numbers (note that the last column shows, as stated in my first post, that the overall result is a weighted sum that includes all political response classes):

Pew 2nd evolution 2013-1

But what was the cause of the shift in Republican opinion? It’s not obviously due to changes in the demographic, religious, or ideological profiles of the Republican party, as they changed little between the two surveys:

Pew 2nd evolution 2013-2

Pew 2nd evolution 2013-3 To my mind, the most interesting new nugget in this report is that the biggest shift of Republicans toward creationism has occurred among the least religious Republicans. From the report:

In fact, however, the surveys suggest that the change in views on evolution occurred especially among the less religious segments of the GOP. Among Republicans who attend worship services monthly or less often, the share who say humans have evolved over time is down 14 percentage points, from 71% in 2009 to 57% today. Among Republicans who attend services at least weekly the share who believe in evolution has gone from 36% in 2009 to 31% today, a difference that is not statistically significant.

This may support the suggestion of, among others, Zack Beauchamp and Paul Krugman that accepting creationism has become part of Republicans’ “team” or “tribal” identity: very religious Republicans were already mostly creationist for religious reasons, and now less religious Republicans are following for reasons of party solidarity. (Oddly, Kahan, who called Krugman’s response to the poll “absurd” and “devoid of reflection”, seems to agree with this as well.)

The new Pew report also considers the possibility of wording issues affecting the response. In this case, it was not the wording of the questions on evolution (which were unchanged), but the words of the preceding questions. The 2009 survey was full of questions on science, which may have “primed” respondents to give more ‘scientific’ answers, while in the 2013 survey the evolution questions were preceded by religious questions. I would not be surprised if such differences have an effect; such wording effects may account for some of the disparities between Pew and Gallup results on the same issues.

18 thoughts on “More about the Pew poll on evolution acceptance

  1. I love this series of posts digging into the numbers of change in public opinion. Thanks to you and Jerry for keeping this coming.

    One small typo: ‘Kagan’ was written twice when ‘Kahan’ was presumably intended.

  2. Is the number of people that actually changed their political affiliation between the two polls so inconsequential as to not even be worthy of mention? I’ve thought about that and have seen many others commenting along those lines. Couldn’t “reasonable people having switched ‘sides'” account for some of this as well? They can change their beliefs on the origin of our species, but changing their political affiliation is unthinkable?

  3. The « Aggregate Stability » table shows that the overall ratio of Republicans vs. Democrats among those who accept that humans have evolved is a trifle less than 1:2, roughly constant between 2009 and 2014.

    Good of Pew to point it out.

  4. When I first detected the United States fifty years ago I realised that there is something in American society not then known in Europe. It is that the legacy of inward immigration, and the principle of Social Self Selection (SSS) had worked to form states and communities of people with strikingly similar political and social outlooks. Inward immigration brought so many people who were anti-government; rugged individuals, who carried with them the mistaken notion that individual beliefs have just as much validity as nationally agreed knowledge to be found in schools and universities. And they carried those individual beliefs with arrogance and aggression, even to the point of believing that they should defend themselves from government by buying guns. As to SSS, it is a remarkable thing that America is so much on the move that drunks go to Frisco, gays to Key West, Fire Island or P-Town, those without consciences to N.Y; and those who were born upon another planet to L.A.
    A part of the aggressive individuality of Americans is the fact that many people have dual personalities; one political and the other small-town parental. I spend many a perplexed hour as Americans told me of their individual take on things, all of which flew into the face of reason. I particularly remember the Californian gal who was a Holocaust survivor, at the age of 31. When I queried the dates, she exclaimed that I am ‘denying her reality!!’
    The point being that polls in America are different. Those questioned adopt their political and tribal face (read ‘Republican’) and answer as they think a Republican should. Heck, even Darwin would deny evolution if confronted outside the Diner in small-town America, and asked by Ned Myerson (Groundhog Day) of his beliefs!
    There is much movement of people in Europe these days. There are people like me in this Burgundian backwater, but there were none in London. Social Self Selection is a powerful concept describing the decreasing possibility of getting information from polls.
    On another subject, we have all heard of ‘Peak-Oil’. Is it possible that today is Peak-Jerry Coyne’, in that this site may now fade as all the loquacious coiners of fine phrases will drift away to sites on retirement hobbies and how to amuse your grandkids. After all, the boasting rights have been won fair and square. Congratulations, and well-played to Diana McP of Canada. If you haven’t been paying attention, she admitted to having read Kant in German and Plato in Greek. No topping that. I crawl away in humiliation. And don’t you dare say that you didn’t know it was a competition! (Smiley face)

  5. Republicans are scared. They know demographic trends are against them, so they turn ever more inward for a sense of security. The walls of their echo chamber keep getting higher. The train has jumped the tracks and they keep the pedal to the metal. Republicans have been pimping Christianity (and using it as a weapon against their opponents) for so long it’s getting hard to tell them apart.

    1. This is true. They are the Party of Fear, driven by fear and doing their best to govern by fear. And it is working to an extent. They have become very scary.

  6. Kahan is motivated, because his work is about political reasoning, and how many political disputes don’t respond to “reason” because they stem from differences in underlying values, and everyone is prone to logical errors or bias regardless of their values starting point.

    Much of his work is interesting Having one party clearly more immune to simple facts undermines a significant conclusion of his work, so I do think he reflexively carries water for Republicans on issues like science acceptance.

    On a related but not strictly on topic subject:

    I find Jonathan Haidt’s work to be similarly flawed. For example, he argues that while everybody (supposedly) values fairness, conservatives also highly value other things like purity. But he ignores the fact that not all values are created equal. Placing too high and too irrational a value on purity leads to things like being freaked out by and stigmatizing a natural process like menstruation. Purity “regulations” in many religious traditions burden and oppress women, or minorities like the Dalits, undermining fairness.

    On the other hand, the purity value’s contribution to public health is overrated. Our modern accomplishments of hygiene and food and drug purity came from doctors, scientists, and social reformers, acting from a Caring motive, not religious leaders or traditions acting from a purity motive.

    By his studied obliviousness to the tensions between values and the idea that a value like purity should be subordinated to a value like fairness, Haidt ends up implicitly defending conservatives in a similar way to Kahan, and for similar reasons, IMHO.

  7. This carries a huge caution for liberals like myself.

    Beliefs (and that is what they are for many people) do become tribal; that is why the science minded liberals must aggressively speak out against woo woo (e. g. “alternative medicine”, anti-GMO hysteria) lest they become “known liberal beliefs”

  8. I never said the Pew numbers “must be incorrect.”

    I said they didn’t add up. They didn’t.

    & that Pew should explain. It did.

    Only it was able to do so, b/c only it had the data.

    Your math was — your guess. Their were *multiple* possible explanations, most of which were not the true one.

    My only point was that w/o the full data — including the partisan breakdown on the “naturalistic” & “theistic” evolution 1/2 of the question, it was not possible to draw the sorts of inferences that commentators were drawing.

    You agree.

    Yet you insist on misrepresenting me (as anyone who reads my posts can confirm for themselves).


    1. You said the Pew numbers “don’t add up”, do “not compute”, and suffer from “logical inconsistency”, because you believed (incorrectly) that insignificant changes in Democrats and independents combined with significant changes in Republicans necessitated a significant change in the overall number. If you want to parse that now as not claiming the Pew numbers must be in error, go right ahead, but any fair reader of your comments will conclude otherwise.

      I thought you had admitted your error about this, but you now astonishingly say again that the Pew numbers didn’t add up. Your commenter MW and I, by different routes, both showed that your inference of error by Pew was unwarranted. Neither of us, of course, could supply all the actual terms of the weighted sum (I didn’t try– MW hazarded plausible hypothetical ones), and I in fact agreed with you about lamenting the unavailability of the full data set. Pew has now provided all the terms of the sum (highlighted in my post above), but the Pew numbers have always added up.

      Most commentators drew one salient inference from the original (and correct) Pew numbers: that their had been a significant increase in support for creationism among Republicans from 2009 to 2013. This is a statistically valid inference, and no amount of pleading can make it go away. It was not necessary to have the additional details to draw that inference.

      Despite being a statistically valid inference, it is still possible that the Pew result is wrong, in that the apparent change might result from sampling error, or peculiarities of sampling design, or wording effects. Indeed, the contrast with Gallup results is notable in various ways. Most commentators have not remarked on this, although Jerry, I, and now you have all highlighted this disparity.

      Most commentators, including yourself, noted that the change could be due to shifts in the views of Republicans, or shifts in party identification. Some went on to speculate about the causes of the Republicans’ response. One suggestion, made by both liberals such as Paul Krugman and conservatives such as Allahpundit, is that it is a partisan response, a rallying together of the team or tribe or party against adverse political trends. While decrying this interpretation, you actually seem to share it, arguing that being a creationist has become a part of Republican “self-identification”. If you want to distinguish Republican “self-identification” from the Republican “team” or the Republican “tribe”, go right ahead, but, again, I think a fair reader would conclude you are largely adopting a view you are at the same time excoriating.

      As I had already stated in my original post, while it is nice to have the partisan breakdown on Pew’s follow up naturalistic vs. theistic question, the results on that question could not possibly undermine the valid inference of increasing and strong support for creationism among Republicans, and thus the chief result emphasized by most commentators still holds.

      Pew’s original release stated that demographic and religious differences could not account for the partisan differences, and it is nice to see this demonstrated in the additional data release, which also shows little shift in the profiles of these characteristics among Republicans (suggesting that it is shifts of view within the Republican party, rather than changes in party identification that have occurred). But I never suspected Pew was wrong about this, or deliberately withholding data so as to subsidize a response. Pew is a respectable outfit, which releases its full data sets after a few months delay.

      I urge readers to read your two posts and the several updates. I am confident as to the outcome of their consideration.


  9. Putting religion to the side, there’s a very good reason for the Republicans, with their tight links to (and funding from) big business to foster an attitude of distrust towards science and government.

    Folks like the Koch brothers want to be able to do whatever they want in order to maximize (short term) profits from their businesses. The only entities large enough to oppose their will are governments – primarily national governments, and, to a lesser extent, a few state governments. Especially in the area of energy policy, governments are most likely to create roadblocks for big business based on scientific findings such as the catastrophic predictions from climate scientists.

    So, the best way to prevent that is to convince people to oppose governmental restrictions on business and to erode trust in any scientific basis for those actions.

    (As an added benefit, if you convince people that government actions are almost always bad, it’s easier to push a reduction in the size of government and, as a result, decrease taxes.)

  10. It’s worth noting that designing good polls and executing them with good technique is damned hard, and that it’s almost trivial to put your thumb on the scale and skew the results any way you like.

    Here we see a perfect example, where a simple reordering of the sections is almost unquestionably a non-trivial factor in what appears to be an huge demographic shift. This is not at all a surprising finding, as it’s an introductory textbook example.

    Is the shift real? If so, is it significant? If significant, is the magnitude as reported correct?

    I think we can very safely answer “no” to the last, and “quite possibly” to the first. But because Pew fucked up their methodology so badly, that’s all I’m personally comfortable gleaning from this one.

    Damned shame. Pew should — nay, does — know better. This type of incompetence and / or bias is inexcusable and counterproductive.



  11. An 11 point change in belief about evolution in just 4 years seems implausibly large. But modest changes in belief and party composition, priming with science questions in the 2009 survey, and random error could certainly add up to 11%.

  12. The proportion of Republicans, who attend services less than once a week, and accept evolution has tanked (71% to 57%).

    Does that mean that:

    (i) Large numbers of Republicans who attend services less than once a week have stopped accepting evolution?

    (ii) Or that large numbers of (former) Republicans, who attend services less than once a week, and accept evolution have abandoned the Republican party?

    The second hypothesis would seem to better explain the consistent aggregate acceptance levels (though there are explanations consistent with (i)).

    Also, the group in question (those who attend services less than once a week, and accept evolution) would appear likely to be (i) a minority in the Republican party & (ii) those with least to lose by abandoning it.

    What we really need is information on any changes in the proportion of Republicans by religiosity over the time period.

    1. The data in the last table above show that average frequency of attendance has gone up a bit from 2009 to 2013 (weekly attendees up to 51% from 47%). This would seem to lean toward your explanation (i). A difficulty with nearly all polls is that they do not follow respondents longitudinally, and thus we don’t actually directly know who (if anybody) changed their views.


Leave a Reply