Atheism makes you a great scientist?

A reader asks, perhaps facetiously:

Funny, but given the homage being paid to science in so many of these first threads, I was just wondering if there’s any actual evidence that, in practice, atheists make better scientists than believers.

There is. In the United States a general survey population of PhD scientists showed a 40% rate of theism (belief in a personal god). But among a sample of National Academy of Science members, the creme of the scientific profession, the rate of theism is 7%. You can see the data here. Other data shows that research universities tend to have a higher proportion of atheists & agnostics than bachelor granting institutions, who have a higher proportion of atheists & agnostics than 2 year colleges.

Does this mean that atheism makes you a better scientist? Matthew Nisbett has communicated to me that the extant social science tends to lean toward the proposition that secular individuals choose particular vocations, specifically, scientists who are atheists & agnostics are already non-believers by the time they enter university. Perhaps within the sciences there is a positive feedback loop whereby the culture is congenial to non-believers; I once worked in a lab with a colleague who was an evangelical Christian who did research on evolutionary biology.  He told me once that probably every few days for the past year someone had asked him how he could reconcile his religious beliefs with his scientific work, both his colleagues and his fellow Christians. One can add many other speculative processes which might lead to the sorting you see above; e.g., I suspect that the demands of time and the relatively modest remuneration results in those with large family and community obligations (one requiring sufficient funds for modest comfort, and the other time for participation) to opt out of science. And there is data that the religious are more likely to be married and have children.

This entry was posted in data and tagged , , . Bookmark the permalink.

15 Responses to Atheism makes you a great scientist?

  1. Samuel Skinner says:

    There are factors that are linked to the truth value of atheism and factors that aren’t. I don’t know which has a bigger pull.

  2. Tim Kowal says:

    Those stats suggest that atheists make more frequent scientists than believers, not better ones.

  3. David Hume says:

    Those stats suggest that atheists make more frequent scientists than believers, not better ones.

    What exactly about the NAS don’t you understand?

  4. JM Hanes says:

    Yes, I’m happy to say, it was a genuine question, and I appreciate your taking it up. If there are material differences between the way believers and non-believers think, I wondered if it would show up in their relative scientific abilities. I’m not sure there’s any easy way to get at the answer. When non-believers preponderate in science generally, the fact that they preponderate among the ranks of great scientists too may not say much about potential disparities in the relative quality of scientific work performed by non-religious and religious practitioners, per se.

    Questions about why the non-religious are disproportionately attracted to or represented in the sciences, however, are interesting in their own right. I enjoyed exploring that angle through your links and the various sources I came across as a result. So too, are questions about whether the sciences might appear to be hostile environments from a believer’s perspective. The Peter Atkins comment is certainly not what I’d call encouraging:

    You clearly can be a scientist and have religious beliefs. But I don’t think you can be a real scientist in the deepest sense of the word because they are such alien categories of knowledge.


    I’m not surprised that mathematicians are marginally more religious than other scientists (per Nature). In my (anecdotal!) experience the hop between higher math and metaphysics can often be a short one. I’m also not surprised that psychologists and biologists are marginally less religious than their brethren, per Gross & Simmons, but I suspect a discussion about why would be all kinds of lively! Gross & Simmons’ paper is no longer hosted by Harvard, but there’s a PDF here: How Religious are America’s College and University Professors?. They note that:

    If there is a single sociological lesson to be learned from American religious pluralism, it is that how one believes in God matters as much as whether one does.

    That notion is reflected in a different set of questions from the either/or aspect of belief evidenced in the Nature piece.

    Elaine Howard Ecklund (SUNY Buffalo) notes what would seem to be a related — and somewhat counterintuitive? — finding in Religion and Sprituality among University Scientists:

    When asked “to what extent do you consider yourself a spiritual person?” about 66 percent of the natural scientists and about 69 percent of the social scientists describe themselves as spiritual.

    Ecklund’s comments from additional interviews on what the various respondents meant by sprituality are interesting, as is her observation that “Definitions of ‘religion’ and ‘spirituality’ are not benign constructs for this population.”

    Perhaps most relevant to my original question is the observation you passed along from Matthew Nisbett indicating that “scientists who are atheists & agnostics are already non-believers by the time they enter university.” This suggests that religious/non-religious constructs are formed before a professional, adult commitment to the sciences has been made. Ecklund confirms this:

    Childhood religious background, not exposure to scientific education, seems to be the most powerful predictor of future irreligion. Those scientists raised in almost any faith tradition are more likely to currently be religious than those raised without any tradition. In addition, scientists who describe religion as important in their families as children are much more likely to practice faith currently. When compared to the general population, a larger proportion of scientists are raised in non-religious homes.

    Something to chew on, I think. The material antecedents of atheism and religiosity, if I understand the concept you asserted earlier, seem as though they might be of some importance to discussions here.

  5. Daniel Dare says:

    I think the question reverses causality.
    Scientists make better atheists.
    It’s the training, the critical analysis. The existence of an alternative narrative to provide an explanation and context for existence.

    I think for most religious, it is the science that comes first, the atheism/agnositicism follows. This was my experience. The more passionate you are about science, the more you will internalize it. The pure light drives out the darkness.

  6. David Heddle says:

    Of course those NAS statistics do not mean that on the average atheist scientists are better than believing scientists.

    Suppose there are 200 baseball players, 100 black and 100 white. Each player has a quality ranking from 1 (bad) to 10 good).

    The all star team (the NAS) consists of the top 5% (ten players)

    Now suppose the white players have this distribution: 7 players ranked 10, and 93 ranked 1.

    Suppose the black players had 3 ranked 10, and 97 ranked 9.

    The allstar team would consist of 70% white players.
    The average white ranking is 1.63
    The average black ranking is 9.03

    Of course this example was contrived to make the case obvious. It doesn’t have to be so lopsided.

    I am not proposing this as any sort of explanation–it might indeed be as simple as smart people are less likely to be believers (a notion for which their is even some biblical support.) I’m just pointing out that the oft-quoted survey does not permit simple conclusions on the distribution of all scientists–and the more elite the NAS is, the more likely it will have a skewed distribution. (Which is especially obvious if you imagine the NAS is so selective it accepts only one member, in which case it will be either 100% atheist or 100% theist and say absolutely nothing.)

  7. Daniel Dare says:

    JM Hanes

    Those scientists raised in almost any faith tradition are more likely to currently be religious than those raised without any tradition.

    That would be true for any population not just scientists. People raised in a tradition are more likely to be religious.

    When compared to the general population, a larger proportion of scientists are raised in non-religious homes.

    I’d suspect IQ?

    scientists who are atheists & agnostics are already non-believers by the time they enter university

    I was passionate about science by 15 y o. I gave up my religion at about the same time although I flirted with other ideas in late teens. Always coming back to science in the end.

  8. JM Hanes says:

    Daniel Dare:

    “I’d suspect IQ?”

    I think it takes some circularity to get there, but that could change. I don’t know what evidence exists as to comparative intelligence across disciplines, but not only do the non-religious stats vary considerably between scientific and non-scientific disciplines, they also vary within scientifc disciplines. Thus, psychologists and biologists, for example, would presumably be more intelligent than their scientific peers. One would also have to assume that the “smartest” people end up in university careers — as opposed to, say, business or the performing arts, etc. — and that the smartest folks on campus gravitate to science vs. the liberal arts. It’s one of those places where the margins of error may, as yet, exceed the variables (terms I use as a layman).

    My primary take away so far is that one’s religious attitudes are pretty well established in one’s youth. Although environment appears to be king of the (non)religious castle, that would also correspond to my own anecdotal experience, in which the exploration of extra-familial (i.e. personal & universal) religious conviction and/or doubt commences in the early teens. Among my peers, the attendant questions were largely resolved or set by the late teens — which happens to coincide with one of the most widely accepted measurements of achievement for predictive purposes represented by the SATs.

    Assuming at least some correlation between IQ and achievement, I think it’s noteworthy that the SATs seem to recognize at least two (fundamentally?) different kinds of skills when they test for verbal and mathematical achievement separately (or used to, anyway). The idea that there is more than one kind of intelligence doesn’t seem like a terribly big leap to me, and I believe that concept might be worth keeping in mind generally. I personally would be hard put to assert any direct connection between academic acuity and common sense, for example.

    That both the measurements and purposes of prospective testing are, in fact, academic in nature seems potentially related to the religious disparities between academic and non-academic communities, and between academics and the population at large. Per Ecklund, the professoriate is less religious than the general public across almost all disciplines. Per Gross & Simmons, “professors at elite doctoral universities are much less religious than professors teaching in other kinds of institutions.” The strikingly anomalous character of academia on so many other fronts as well (e.g. liberal vs. conservative, social structuring, incentives & disincentives, authoritarian character) adds commensurate layers of complexity and may prove significant in unexpected and, it seems to me, largely unexamined, conceivably self-perpetuating, ways.

    All of would seem to speak to the pitfalls of extrapolating from aggregates, not just assumptions!

  9. Daniel Dare says:

    “My primary take away so far is that one’s religious attitudes are pretty well established in one’s youth.”

    I have personally, checked for curiosity’s sake, this is not something I am an expert in, the religious-belief questions in the official Australian Census. (my country). And there I have found:

    If you follow a single age-cohort, through consecutive censuses, there is a sharp drop of religious-belief in mid to late teens, (the rebellious years), and then a partial recovery around 30. (persumably as they start raising families).

    The recovery is only partial, because Australia shares with Europe a strong secular rise in the “no-religion” response since around 1960.

  10. JM Hanes says:


    Sounds like that’s the most common progression across the board (and equator), doesn’t it? Ecklund also reported that:

    Those not raised in religious homes, the case for the majority of scientists without religious affiliation, also emphasize their lack of experience with religion.

    Considered all together, perhaps that’s why so many of those taking up arms against religion, so to speak, often sound as if they’re trying to translate a foreign language as they go along, with nothing but a dictionary to work from.

  11. Staash says:

    Wasn’t Newton very religious? Even for his time?

  12. Daniel Dare says:

    Yes he was Stash. Wikipedia

  13. Samuel Skinner says:

    Many early scientists had… unusual religious views. Generally a mix of deistic and more unusual.

    I think it was a trait of the process- they got people who were well-educated main stream (diests) and crack pots. For example, with the Greeks, you had deistic scientists… and people like Pythagoras.

    Of course, as our knowledge of the universe has become more secure, we have had lower and lower numbers of… unusual religious views.

  14. Caledonian says:

    The question is not whether you can hold religious beliefs and be a scientist.

    The question is whether you can do so consistently.

    Humans are known for being inconsistent and holding mutually incompatible views through a process of compartmentalization. I’m sure you can be a scientist and still convince yourself that the Earth is flat. But you can’t be consistent in your views that way.

    So: can you be consistent while being a scientist and religious? Another way of putting it: is religious belief rationally defensible?

  15. F F Robb says:

    It is not very scientific to take it for granted that a correlation entails a causal relation. However, to go along with this for a moment – surely the missing data are about what kind of religion/faith? E.g. were one to believe that everything that goes on in the world is predestined, or an expression of an ever-present deity, then the search for causal relations other than those either predestined or created for the moment by the deity would be the task. Science would then be directed to discovering the limits to the deity’s powers or something like that, and not to attempting to make generalizations from experimental data or forecasts of future events.

Comments are closed.