New study shows that 64% of Americans favor religion over science

February 1, 2015 • 10:20 am

A new paper by Timothy O’Brien and Shiri Noy in American Sociological Review (free download; reference at bottom) has gotten a lot of attention from the media. It shows that Americans are, by and large, friendly to science—but that the friendship stops when the science conflicts with people’s faith.

I’m not sure why this paper got so much press, unless people like the new trichotomy it proposes about American attitudes, as well as its trendy neologism of a “post-secular perspective”. For the results pretty much confirm what we all know: most Americans are religious, many of those religious people do accept science, but those who reject science often do so only when the science conflicts with their religion. To the authors of this survey, that means the Big Bang and evolution.

The authors analyzed already-existing data from the General Social Survey, a poll of Americans’ views taken every two years. Besides polling Americans on their demographic information (age, location, education, and so on), and on their political and religious views (denomination, strength of belief, views on abortion, fuel economy standards for cars, GMOs, etc.), it also got information on the subjects’ knowledge of science. They asked 12 “innocuous” questions (e.g., “Does the Sun go around the Earth or the Earth around the sun?”, “Are electrons smaller than atoms?”) and 2 hot-button questions (“Did the Universe begin with a huge explosion?” and “Did human beings develop from earlier species of animals?”). Finally, they got information on people’s attitudes toward science in general (e.g., agree or disagree: “Science makes our way of life change too fast,” or “Scientific research that advances the frontiers of knowledge is necessary and should be supported by the federal government”).

To analyze the data, the authors used a method with which I’m unfamiliar: “latent class analysis,” or LCA. It apparently extracts the number of classes of individuals that are necessary to best explain patterns in the data about science and science attitudes, assuming that attitudes among different variables are correlated, as they are. Before they did the analysis, they identified four possible views on science and religion, to wit:

Screen Shot 2015-02-01 at 8.45.59 AM

But what they found were three classes: all but the “postmodern” view, which the authors say would include those people who view both science and religion unfavorably, perhaps because they disdain all claims of truth. Instead, they found the data divided fairly neatly into the traditional, modern, and “post-secular” views.

Traditionals view religion favorably and science unfavorably, know less about science than members of the other two groups, but report stronger levels of religious affiliation (for example, 46% of them say that the Bible is the literal word of God compared to only 31% in the overall sample—a figure that is still astounding to me). Traditionals constituted the largest class: 43%.

Moderns, who have the reverse attitude (yes to science, no to faith) constituted 36% of the sample.  They know more about science and have more favorable attitudes about science than do members of the two other classes.  They are of course less religious than the other groups: for instance, only 3% felt that the Bible was the literal word of God.

And the post-secularists, who are friendly to both science and faith, composed 21% of the population. They are just as religious as the traditionals: 48% see the Bible as the literal word of God. But they are indistinguishable in scientific knowledge from the moderns—EXCEPT on three issues, issues on which post-secularists diverge strongly from moderns but aren’t quite as denialist as are traditionals. Those three questions, and the proportion of each of the three classes agreeing with them, are shown in the summary chart from PuffHo:


(Surprisingly, moderns have lower agreement with the Big Bang than with evolution and continental drift.)  What do those questions have in common? They are the only ones that conflict with Abrahamic religious faith. Traditionals in general disagree with or are ignorant of what science says, but post-secularists reject the science when it implies that the universe started with a “huge explosion” (it didn’t really) or that humans evolved or that continental drift occurred over millions of years (it’s the “millions of years” that bothers them). I’m a bit surprised that post-secularists reject the notion of a Big Bang, because that’s largely compatible with theism, unless you believe that the universe was created exactly the way God said in Genesis, and even then you could metaphorize a tad and still make the story compatible with the Big Bang.

If you reject science when it comes into conflict with your faith, then you don’t have a truly scientific attitude, i.e., one that follows the facts wherever they lead. In toto, then, 43% + 21%, or 64% of Americans, deny science when it conflicts with their faith. 36% are with science down the line, with 88% of these accepting human evolution—one litmus test for evolution since it rejects human exceptionalism.

I’m actually pleasantly surprised at that 36%. It appears higher than the Gallup Poll’s recent finding that only 19% of Americans accept purely naturalistic evolution, but there is a big difference between the two results.  Gallup found that an additional 31% of Americans accept human evolution with the proviso that God guided it. O’Brien and Noy’s data don’t say anything about how many who accept evolution see God’s hand in it. Accepting theistic evolution is not being down with science, since you think God tweaked the process; and we have no idea what proportion of the 88% of moderns who accept human evolution—or, indeed of the 33% of post-secularists and 3% of traditionals who also do—see that evolution as guided by God.

In the end, I’m not clear why so much is being made of this study. Anyone who’s studied science acceptance in America knows that most people are science friendly, but that this friendship disappears when it involves issues that conflict with religion. After all, the U.S. is the world’s most religious First World country.

The inability to find “postmoderns” who reject both science and religion doesn’t surprise me. By and large, Americans are not morons, and if you’re not religious you’d have to be either an idiot or a postmodern humanities professor to reject science.

There are other interesting results of this study—the association between these three classes and political views, for instance—but since the study is free online I will leave you to read those bits for yourself.

O’Brien, T. L. and S. Noy. 2015. Traditional, modern, and post-secular perspectives on science and religion in the United States. American Sociological Review, 80:92-115.

51 thoughts on “New study shows that 64% of Americans favor religion over science

    1. It is a bad question anyway. In many modern cosmological theories the universe existed before the Big Bang. After all, why should we equate the beginning of the expansion with the beginning of existence?

      1. It is common for cosmologists to make a distinction between the two with terms like “multiverse” and “our universe.” It is not necessarily inaccurate to refer to our particular bit of expanding spacetime as “the universe.”

        A bigger mistake, I think, is referring to the Big Bang as an explosion. I mean sure, even NDGT has called it an explosion on popular science shows, but I bet the reason that question scored so low among Moderns compared to the others is that many had a problem with that analogy.

    2. In addition, a “huge explosion” is a very bad way of thinking about the Big Bang. No cosmologist would phrase the question that way. It wasn’t an explosion and it started off rather small. E.g. see here.

      1. I was going to say the same thing. I don’t know how many people know this, but I would think a fair percentage of the science literate would have some idea that “explosion” is not the best word for it.

      2. Yeah — as Jerry says, “(it didn’t really)”.

        I wonder if the percentage is “artificially” low because the really smart Modern people know that it didn’t really!


      3. I am not sure of cosmologists specifically, (is NDGT a cosmologist(?), don’t think so), but plenty of scientists who study very closely related fields like astrophysics and astronomy do / have used that term in the context of talking to the public. Sometimes enthusiastically.

        Since much of the public does not understand that “explosion” is an analogy, probably because the thing being analogized is too far removed from ordinary experience to make any sense to them, it really would have been better if scientists had never used the term.

        1. Of course, “Big Bang” was coined by Fred Hoyle to send up the then hypothesis … Once that term was established, an explosion is the image that most readily comes to mind.


  1. Hardcare hippie homeopaths typically fit a mold of: no religion, no science. It’s all bad under the New Age banner.

  2. In addition to rejecting evolution and the Big Bang when it conflicts with religious views, I think I had read somewhere that a common view from the religious right is to also reject global warming on religious grounds. The erroneous thinking (very wrong on multiple levels) was that since God promised to never flood the earth again, global warming could not be real or be a real problem.

    1. Put test subjects in a room and tell them they’d been exposed to a fatal virus. Then give them a choice of two doors: one leads to a “prayer room” with a minister of the subjects’ faith, the other to a doctor with a syringe containing the antidote.

      Christian Science, Jehovah’s Witnesses and others might – maybe – go to the prayer room, but I think the vast majority of believers would head for the doctor.

      Convenient – and reliable! – accoutrement.

  3. There is also another Pew study that provides more stats and percentages on how the public accepts science but the odd item reported in the article I read quoted Alan Leshner, CEO of the American Assoc. for Advancement of Science. Said: “researchers should be working to find common ground where science generates discomfort between science and the rest of society.”

    What is a scientist to do if his findings cause discomfort in people. Change them? And what is discomfort but another name for incompatibility.

      1. Still trying to think what an accommodation with a geocentrist would be (the Sun revolves around the Earth for half the year?)Or what about a flat-earther? (the Earth is really half a sphere – one part is flat!).

        1. Accomodationism in those situations consists of “scientists have one opinion and you have another — that’s okay. Just leave the textbooks alone and it’s hunky-dory.”

      2. I’ve read that one. I was curious about the questions around whether America was seen as the best in the world at science. I thought they were fairly pointless. The take away seems to be that if you don’t think America is the best you’re not a patriot. This attitude is exactly what’s wrong – thinking you’re automatically the best without recognising the need to invest in science and science education. The religious takeover of education in many states is a problem politicians that get elected can’t see. The survey shows people recognized at least that high school science education wasn’t up to par.

        US PISA scores are amongst the worst in the OECD.

        1. “The survey shows people recognized at least that high school science education wasn’t up to par.

          US PISA scores are amongst the worst in the OECD.”

          For sure, U.S. science education can be improved, and U.S. science teacher pedagogy stands or falls on its own (de-)merits.

          However, from my own substantial experience substitute teaching, student classroom misbehavior and anti-intellectualism are factors contributing to this mediocre performance. (A result of their having given themselves over to this crass American mass pop culture. Re: Richard Hofstadter’s “Anti-Intellectualism in American Life” and Susan Jacoby’s “The Age of American Unreason”) If those realities/facts are ever addressed in studies reported in the media, the media hardly if ever mention them.

          (I have been disabused of the notion that students in a high school HONORS chemistry class – by virtue of being HONORS students – surely have the wherewithal to resist constantly fondling their electronic digital devices.)

          Most Lawrence Krauss enthusiasts here have surely heard him reflect on the U.S. National Science Foundation adult science literacy surveys showing roughly 50% of American adults miss the following question:

          “T or F: the Earth goes around the sun and takes a year to do it.”

          (Notwithstanding the fact that the sun itself revolves around a point located well within itself – if I have accurately-enough articulated that technicality.)

          Few science facts are more basic than that. I myself clearly-enough heard and learned it in school. (And Ah’m frum Ten-uh-say.) How hard is it to remember that? Can anyone here work hard enough to forget it?

          I read that 96% of Amuricuns surveyed recognize the name “Lady Gaga.”

  4. One common argument that used by Traditionals for denying evolution and the Big Bang is that “no one was there to see it”. I wonder how many Traditionals know continental drift is an extant, directly observable process. It is routinely measured and used for many purposes including navigation of spacecraft to Mars.

    1. I wonder how many know that the Big Bang is an extant, directly observable process. It is routinely measured in the microwave regions and visibly apparent in that the sky is fucking dark.

    1. The design is horrible. It took me forever to realize that they simply aren’t supposed to add up to 100% because I initially didn’t focus on the numbers. It also means that different graphs are not comparable because each is normalized to a random number between 0 and 300.

    2. The evolution pie chart is very strange. It seems to have the Traditional and the Post Secular swapped. Why would they make such an obvious error?

  5. I think perhaps the most telling snippet from this study is that those who reject science the most are the ones who understand it the least. Contrast that with the studies that have shown that atheists generally know more about religion than believers and we see more evidence that mainstream religion revels in ignorance despite what Haught, Hart, Craig, Armstrong, et al. would have us believe.

    1. Believing that God is some mystical guru than a genocidal dictator requires that you have not read much of the Old Testament (and even if you do than you must think “the Old Testament doesn’t matter since Jesus” even though Jesus specifically said he came to uphold the laws of Moses). And you basically ignore all the contradictions and pretend there isn’t any whenever anyone points them out.

  6. So basically people like science when it gives technology which makes their lives easier or extends them. But when science starts interfering with their plagiarised Bronze age fairy tales they start to get defensive. Yay for cognitive dissonance!

  7. I thought the high interest in the results was because the 36% figure for yes to science, no to religion, was higher than most people thought it would be. It seemed like progress to me.

  8. As usual, I have problems with the survey questions. It always seems to me that there’s too much ambiguity and they won’t necessarily get at what they want to get at.

    Finally, they got information on people’s attitudes toward science in general (e.g., agree or disagree: “Science makes our way of life change too fast,” or “Scientific research that advances the frontiers of knowledge is necessary and should be supported by the federal government”).

    This, for example. The first one confuses science and technology. Someone who thinks we spend too much time with electronics and not enough time with family could/would tick that. I don’t think that says anything about the respondent’s belief in science as a way of knowing.

    Couple this sort of problem with stuff like the good probability that the people who understand the most about science won’t answer “Did the Universe begin with a huge explosion?” the way they’re supposed to and I imagine the margin of error looking like a highway.

    1. Maybe this is what happens when sociologists or pollsters write the questions instead of experts. The questions should be vetted but then where you find an expert in religious truth?

    2. I agree completely, and it gets even worse. These are three of the four criteria for assigning respondents to the religious categories:

      Bible is the actual word of God?
      Bible is inspired by the word of God?
      Bible is a book of myths and fables?

      This might do to divide the world into Christians (of various stripes) and everybody else, but it fails miserably to capture even a bit of the variety of religious conviction.

      The fourth criterion is ‘Strength of religious affiliation (1 = none, 4 = very strong)’, which is presumably self-reported. That the mean is only 2.597 is actually kind of encouraging.

      What I’m afraid has happened here is that social scientists have drawn up categories based on their own flawed notions of the key issues in matters of science and religion (the political questions are pretty superficial, as well).

      With the caveat that I have skimmed rather than read the entire document, the statistical methods are, to the extent I’m qualified to judge, beyond reproach, but, as we software developers have been saying since I entered the field in the 60s, GIGO (Garbage In, Garbage Out).

  9. Post-secular is a euphemistic way of labeling a willfully ignorant population who is skilled at compartmentalization to avoid cognitive dissonance.

    I suggest that we all them neo-ignorant or post-rational.

    1. Or at least post-traditional, as their own Table 1 suggests.

      I’ve seen the press release. It is accommodationist slanted…

  10. LCA is a convenient technique for explaining the correlations among observed variables in terms of a “latent class” (a hidden discrete variable) such that for each of the values of the hidden variable, it infers a table of distributions for each of the observed variables. If the observed variables are binary, the table just lists the probability of “true” for each of the observed variables. LCA was originally developed to model the correlations between different tests for a medical condition given that the patient has the condition vs given that the patient does not have it. The same idea was concurrently developed in bayesian networks, or as finite mixture models. This is the best simple description I know of: You can get an R package for LCA here: I and former students and collaborators have used extensions of LCA successfully to combine multiple sources of genomic evidence in gene prediction:

Leave a Reply