In Expelled, Richard Dawkins recounts how learning about science "killed off" his faith. And PZ Myers tells us that the more science literacy we have in society, the less religion we will have, and the more science, resulting in a nice feedback loop.
Their comments reflect conventional wisdom among atheists that the more you learn about science, the less religious you will become. In fact, it's the working assumption as to why in comparison to the American public, scientists are less likely to be religious.
But as I have mentioned in several comment threads, it turns out that the linear assumption that an advanced education in science burns away religious belief does not stand up to the data as published in the available peer-reviewed literature. I refer to this fallacy as the "atheists' delusion."
Scientists are human and products of society, just like everyone else. Rather than higher education impacting their belief, the data show that individuals raised in less religious families are more likely to self-select themselves into a PhD program and a scientific career. It's a complex socialization process that begins in the early years of family life rather than some simple knowledge-driven linear process that occurs during college or graduate school.
Here's what the most comprehensive study to date on the topic concludes, based on a survey of university scientists across a diversity of fields. From the University at Buffalo news release with the study here.
"Our study data do not strongly support the idea that scientists simply drop their religious identities upon professional training, due to an inherent conflict between science and faith, or to institutional pressure to conform," Ecklund says.
"It is important to understand this," she adds, "because we face religio-scientific controversies over stem-cell research and evolution, for instance, and increased debate about the role of religion in both national politics and in the public policies that influence science...
....For comparison with the general population, in the Social Problems article Ecklund and Scheitle employed data from the 1998 and 2004 rounds of the General Social Survey (GSS), a national survey by the National Opinion Research Center at the University of Chicago, which regularly collects data on demographic characteristics and attitudes of U.S. residents.
The RAAS survey asked questions on religious identity, belief and practice, which were replicated from the GSS, and other questions on spiritual practices, ethics and the intersection of religion and science in the respondent's discipline, some of which were replicated from other national surveys. In addition there was a series of inquiries about academic rank, publications and demographic information.
The authors then examined how natural and social scientists differ from the general public and how they differ from one another in terms of religiosity. They also considered some of the sources of these differences.
They concluded that academics in the natural and social sciences at elite research universities are significantly less religious than the general population. Almost 52 percent of scientists surveyed identified themselves as having no current religious affiliation compared with only 14 percent of the general population.
And while nearly 14 percent of the U.S. population who responded to the GSS describe themselves as "evangelical" or "fundamentalist," less than 2 percent of the RAAS population identifies with either label.
The only traditional religious identity category where the RAAS population has a much higher proportion of religious adherents than the general population is among those who identify as Jewish -- 15 percent compared to 2 percent of the general population.
Among scientists, as in the general population, being raised in a home in which religion and religious practice were valued is the most important predictor of present religiosity among the subjects.
Ecklund and Scheitle concluded that the assumption that becoming a scientist necessarily leads to loss of religion is untenable.
Ecklund says, "It appears that those from non-religious backgrounds disproportionately self-select into scientific professions. This may reflect the fact that there is tension between the religious tenets of some groups and the theories and methods of particular sciences and it contributes to the large number of non-religious scientists."...
...The oft-discussed distinction between natural and social scientists with regard to religious belief is inconsistent and weak, Ecklund says.
"This is interesting," she adds, "because most of the scholarly literature on faculty attitudes toward religiosity addresses the field-specific differences between natural and social scientists and many scholars hold that social scientists are significantly less religious than natural scientists."
Results from the study also show that the more children in a scientist's household, the more likely he or she is to adhere to a religion.
- Log in to post comments
Second try at this comment...
I don't take Matt particularly seriously any more; and I agree with Sean Eddy's point on the matter of "delusion". Matt's article deliberate echos Dawkins' style of rhetoric; and just as Dawkins alienates the vast majority of religious readers, so too Matt has apparently alienated the great majority of sciblings. However, Matt's own personal failings as a communicator and commentator aside, the article discussed here is an interesting one.
I discussed it myself last year, at a Christian forum (TheologyWeb) where I am a regular contributor, and where I try to maintain and foster a constructive and mutually respectful discourse between Christians and unbelievers. I write there as "Sylas".
The thread where I raised this article is New study: on Scientists and Religion. It provoked a bit of discussion; not much, and not particularly high level. But interesting.
My own perspective on the study may show more of my existing presumptions than any great analysis; but essentially I suggest that the lack of belief amongst scientists by comparison with the general population has a lot to do with who carries through to a career in science and who never gets started on that path. From my article:
And my concluding paragraph:
Based on the evidence I have seen, from this and other studies, it seems pretty definite to me that science education DOES tend to erode religious belief; but that the full story is more complex. This paper suggests that a substantially more important factor for the low rates of religiousity in scientists is the filtering effect, that means religious folks are less likely to carry through to this career.
If the hypothesis is that scientific education will tend towards reduced religiosity, what is the proportion of those scientists who were raised in households which valued religious practices who are non-religious contrasted with the general population's rate? It seems that non-religiously raised scientists self-selecting for non-religion is just noise among the signal here.
Another interesting idea would be how many scientists raised in non-religious households became religious versus the general population rate? If the atheist's hypothesis is correct, you should see a significantly lower rate of religion retention among scientists versus general population and also a lower rate of religious conversion among scientists versus general population.
We live in partially rational societies, make partially rational decisions and have partially rational motivations whether we are scientists or priests. Most day to day science work is perfectly compatible with an irrational mythological religious worldview because the worldview doesnt alter the specific narrow area of science and the science is kept away from the worldview. This is NOT some kind of difficult trick, so the Dawkins hypothesis (if that IS his hypothesis) is probably not true as a general rule. On the other hand, if religions insist on making specific statements about the physical world ("the earth goes around the sun", or "the world is 6000 years old" or "evolution is a fraud") those are invariably going to be modified IN TIME..it may take decades or centuries, but in the long run, truth will out. ...by that time, most adherents of the religion in question will have moved on to accepting the particular physical theory but WITHOUT ABANDONING their religion. Religion is also a moving target and the empirical evidence seems to be that its useful enough for all sorts of purposes , most of which have no problem with particular scientific discoveries..I do feel that the trend is towards less torture and fewer floggings. So I guess we should be thankful for small mercies.
Sorry, I don't understand your point.
This study appears to be a survey showing that fewer scientists self-identify as having religious faith, compared to the general population. How does this address the question of whether scientists lost religious faith (as a result of studying science), versus never had it (and self-selected into science careers)? (Perhaps it's in the study, but the original is behind a subscription-only wall.)
Regardless, doesn't either explanation tend to support the idea that many scientists find faith and science to be incompatible?
And for what it's worth - referring to one possible hypothesis as a "delusion" is not really the hallmark of a rational analysis.
My father said the same thing about his faith (he was raised a Catholic and became a scientist and a non-believer). I had no reason to doubt him, and I have no reason to doubt Dawkins about his reasons for rejecting religion.
If you want to argue that scientific knowledge has no theological implications you need to answer a couple of questions:
1. What are creationists so upset about regarding the study of the natural world?
2. Why did the gradual lessening of religiosity coincide with the scientific revolution? It could be just a coincidence, and no doubt it is a complex, multi-faceted phenomenon, but I think that the advance of scientific knowledge had something to do with it.
'This may reflect the fact that there is tension between the religious tenets of some groups and the theories and methods of particular sciences and it contributes to the large number of non-religious scientists."...'
I fail to see how this differs in a significant way from the "conventional wisdom."
I think they mean "... it contributes to the large number of non-religiously raised people who become scientists". But my original point stands. What remains to be seen in the data is how many irreligious-raised scientists convert to religion and how many religiously-raised scientists deconvert to irreligion.
I think what's missing in these "conversion stories" is the nature of the Church that one leaves (or that one stays in).
Consider that many of the stories of atheists who saw the "light" are really those that saw the darkness inherent in the actions of men of "faith", the hypocrisy of money-grabbing evangelicals preaching fear of anything different or the historically power-mad Catholic Church and its non-nonsensical dogmas and doctrines on matters of sex.
By contrast, how many scientists that stayed "faithful" were actually members of more moderate and perhaps liberal churches, like most American Jews, Episcopalians, Unitarians, and the like, or even Pagans? These are churches that haven't built their current existence on a doctrine of lies that if revealed for fact would undermine someone's power structure.
The facts of ex-religious atheist scientists are certainly there, and the numbers are significant, but my impression is that everyone who leaves a church leaves first because of dissatisfaction with that church and the people in it. If one still feels community, they'll find reasons to justify staying in. They will still find reasons to "believe".
Matt,
Could you post some of the critical statistics from the paper you cite, and the argument that leads to the conclusion?
I can well believe there's a significant self-selection factor, with irreligiosity making people more likely to be scientists.
On the other hand, it's hard for me to believe that there's no significant effect the other way, with increasing scientific understanding eroding religious belief.
If there's not, something is seriously wrong somewhere.
It's interesting that among outstanding scientists, atheism is the dominant view.
If that's just a matter of atheists being more likely to become outstanding scientists, the implications are staggering---the best way to promote great science is to increase the incidence of atheism.
Somehow, I doubt it's that simple, but the truth's gotta be very interesting, one way or another.
I don't think this study shows what you want it to show.
From this description, the most that it looks like you can conclude is that, within an atheistic group, a religious upbringing is the best predictor of religious sentiment later in life. Also very important is that they're just studying the differences within a highly atheistic group, and not differences between this group and the general population.
At a stretch, you could argue that fundamentalist families are hostile to science, but I don't see anything which would let you conclude that the extreme concentration of atheism in science is due to self-selection. Even the abstract bears this out:
"We find that field-specific and interdisciplinary differences are not as significant in predicting religiosity as other research suggests. Instead, demographic factors such as age, marital status, and presence of children in the household are the strongest predictors of religious difference among scientists."
You are jumping way beyond what the data appears to show.
Richard:
I have no idea. Moreover, I have no idea why they'd have a problem with this but not, say, the lack of water dividing the firmament from the sky.
But I take your point. It's not that science has no theological implications. It does, however, rule out a bunch of schools of theological thought.
I think it's at least as plausible to argue the other way around: The reformation made it possible to acquire knowledge by experience and reason (as opposed to just whatever the Pope said), and this in turn made the scientific revolution possible. In addition, the scientific revolution was partly a product of technology, such as the development of telescopes that were good enough to be pointed at the sky. The reformation was also a product of technology, such as the industrial paper mill and the printing press.
I do agree that the correlation is not a coincidence, but correlation does not imply causation.
Sean Eddy: "How does this address the question of whether scientists lost religious faith (as a result of studying science), versus never had it (and self-selected into science careers)?"
Take a look at the portion that Nisbet bolded: "Among scientists, as in the general population, being raised in a home in which religion and religious practice were valued is the most important predictor of present religiosity among the subjects."
If science education really was the factor that made scientists more irreligious, one might expect education to be the most important predictor. I skimmed the actual study, though, and one catch is that I didn't see whether anyone tracked how many who were religious when starting their scientific education were still religious at the end of it. I'm not sure if I didn't see it because it wasn't there or because I was scanning the paper relatively quickly.
The abstract itself isn't a problem (the conclusion that religious scientists have most likely been raised in a religious setting is hardly going to be a shock to anyone). What is a problem is the inferences you are drawing from the paywalled data. Does anyone seriously think that science "necessarily" leads to atheism? I seriously doubt it. I'm sure we all know many scientists that are religious.
If we are dealing with the hypothesis that non-religious people choose science then should we not expect that the original family environment of scientists should reflect their beliefs? In other words if Matthews hypothesis is correct wouldn't we expect that the proportion of religious parents should be equal to the number of religious scientists. Speaking for myself its not the case, but thats just anecdotal. Do 60% or more of scientists actually come from atheistic families?
Matthew, is that information in the paper?
"If that's just a matter of atheists being more likely to become outstanding scientists, the implications are staggering---the best way to promote great science is to increase the incidence of atheism."
Jews are overrepresented in the sciences by a factor of 7.5 and amongst Nobel Prize winners by a factor of 80. They're also tremendously overrepresented amongst the winners of a variety of other awards, including the Lasker Award in Basic Medical Research and the Wolf Prize in Medicine. Maybe we should be promoting Judaism. Too bad proselytizing a violation of Jewish law, huh?
My point: I think it's a bit more complicated than that. There's no direct correlation between being an atheist and being a great scientist. (52% of scientists said that had "no religious affiliation" not that they were atheists. Big difference.) Especially since "prominence" in the sciences isn't a one-to-one correlation with ability or the importance/accuracy of one's ideas. Quite a few once "prominent" scientists are now remembered as quacks.
Hi everyone,
I am glad that pointing out this study has generated some healthy discussion and debate. I paste key portions of the findings and conclusion below.
METHOD
The data we examine were part of a broader study of religion, spirituality, and ethics among academics in seven different natural and social science disciplines at twenty-one elite U.S. research universities.
Faculty members included in the study were randomly selected
from seven natural and social science disciplines at universities that appeared on the University of Florida�s annual report of the �Top American Research Universities� (Lombardi et al.2006). The University of Florida ranked elite institutions according to nine different measures, which included: total research funding, federal research funding, endowment assets, annual giving, number of national academy members, faculty awards, doctorates granted, postdoctoral appointees, and median SAT scores for undergraduates. These measures were similar to those used in other studies that examined elite universities (Bowen and Bok 1998; Massey et al. 2002). Universities were ranked and selected according to the number of times they appeared in the top twenty-five for each of these nine indicators.
During a seven-week period from May through June 2005, the study�s PI randomly selected 2,198 faculty members in the disciplines of physics, chemistry, biology, sociology, economics, political science, and psychology from the universities in the sample. Although faculty were randomly selected, oversampling occurred in the smaller fields and undersampling in the larger fields. For example, a little more than 62 percent of all sociologists in the sampling frame were selected, while only 29 percent of physicists and biologists were selected, reflecting the greater numerical presence of physicists and biologists at these universities when compared to sociologists. In analyses where discipline was not controlled for, we used data weights to correct for the over/under sampling.
Table 1 describes the sample and weighting in greater detail. An initial contact letter was written by the PI and contained a fifteen-dollar cash preincentive (i.e., the PI and research team sent fifteen dollars in cash to each of the potential respondents regardless of whether they decided to participate in the survey). Each respondent received a unique ID with which to log into a Web site
and complete the survey. After five reminder e-mails the research firm commissioned to field the survey, Schulman, Ronca, and Bucuvalas, Inc (SRBI), called respondents up to a total of 20 times, requesting participation over the phone or Web. Six and a half percent of the respondents completed the survey on the phone and 93.5 percent completed the Web-based survey. Overall, this combination of methods resulted in a relatively high response rate of 75 percent or 1,646 respondents, ranging from a 68 percent rate for psychologists to a 78 percent rate for biologists. This is a high response rate for a survey of faculty. For example, even the highly successful Carnegie Commission study of faculty resulted in only a 59.8 percent rate (Ladd and Lipset 1972).
The survey asked some questions on religious identity, belief, and practice, which were replicated from the General Social Survey (GSS) (Davis, Smith, and Marsden 2005), and other questions on spiritual practices, ethics, and the intersection of religion and science in the respondent�s discipline, some of which were replicated from other national surveys. There was also a series of inquiries about academic rank, publications, and demographic information.
Table 2 shows the demographics of the sample.
For this article, data about religious identity, belief, and practice were analyzed from the RAAS survey. These included questions about the respondent�s view of religion, God, and church attendance. Basic frequencies of these outcomes are shown in Table 3. The questions that were replicated from the General Social Survey (GSS) to allow for comparisons between the general population and academic scientists. Specifically, we compared the responses of scientists to the general population on measures of current and childhood religious affiliation using GSS data from 1998 and 2004 (Davis, Smith, and Marsden 1998, 2004).
We also present logistic regression models predicting scientists� religious behaviors and attitudes.
FINDINGS
-->On the absence of significant differences between so called hard scientists and social scientists:
We see inconsistent and unconvincing evidence that the much discussed field-specific differences between natural and social scientists still exist. If there is something endemic to the natural (or social) science fields that causes scientists in those fields to be more or less religious we would expect the four social science disciplines of sociology, economics, political science, and psychology, to consistently differ from the reference category of physicists, while the two other natural science disciplines should not differ significantly from physicists. These patterns, however, do not appear in the data. While political scientists do appear somewhat more likely to have few doubts about God�s existence and more likely to believe there is truth in religion when compared to physicists, chemists also score higher on all three religiosity outcomes compared to physicists. Furthermore, differences in the religiosity measures for economists and psychologists are not significantly different from physicists. These results show that any interdisciplinary diversity in religious belief or approach to religion cannot be described according to clean field-specific differences between natural and social scientists, especially after considering other demographic factors. These results confirm information gleaned from the frequencies presented earlier in Table 3. Looking at the �religious truth� outcome, more natural scientists answer that there is very little truth in religion. This difference is deceptive, though. When looking at each individual discipline, it is clear that this natural social difference is mainly due to the low number of political scientists expressing this view and the high number of physicists, not a clear difference between all natural and social science disciplines. In short, there is not a consistent trend between the natural-social categories. Furthermore, when differences between natural and social scientists are discovered they are very small.
--> On demographic and childhood religiosity effects on the religiosity of academic scientists:
Since the particular discipline a scientist is trained in has little to do with individual religious beliefs or practices, to build theory about differences in religiosity within the academic scientific community we move next to examining variables that explain religiosity in the general population. In Table 5 we expand analyses beyond disciplinary differences to incorporating other variables that could explain some of the variation in the religious beliefs of academic scientists...
...The factors considered so far explain some of the intra-scientist differences. They do not, however, provide much insight towards understanding differences in religiosity between scientists and the general population. The demographic variables provide potential insight,
although not conclusive evidence. For example, if both scientists and the general population become more religious when they have children, then maybe the low religiosity of scientists is due to their low fertility rates. However, the data in Table 2 reveal that scientists� fertility rates are similar to those among the general population, with scientists on average having about two children.
More importantly, these types of differences would not explain the extent of the distinctions seen in Table 4, which compared religious affiliation of the general
population with the scientists.
----->Measures of childhood religiosity, the final set of predictors seen in Table 4, provide greater potential for understanding differences between academic scientists and the general population. We added two measures of childhood religiosity or upbringing. The first is a set of predictors representing the tradition in which the respondent was raised and the second is the scientists� reported �importance of religion in your family while you were growing up.�
The latter is measured by four responses ranging from �very important� to �not at all important.� We find that scientists raised as Protestants are more likely to retain religious beliefs and practices than those raised without a religious affiliation. Similarly, those who say that
religion was important in their family when growing up are less likely to say that they currently do not see truth in religion, do not believe in God, and do not attend religious services.
Another way to examine the impact of religious upbringing is through predicted probabilities. For instance, consider two sociologists who are male, in the 18 to 35 range, born in the United States, have no children, and are currently married. One was raised some form of Protestant and religion was �very important� while growing up. The other was raised as a religious �none� and religion was �not at all important� while growing up. The former has a predicted probability of 14 percent for saying that he does not believe in God. This compares to a 54 percent chance of the latter saying he does not believe, a striking difference. Such differences do not offer conclusive evidence about the causes of disproportionate self-selection of scientists from certain religious backgrounds into the scientific disciplines. They do offer potential theoretical pathways for explaining the differences in religiosity between scientists and the general population.
CONCLUSIONS
It is an assumption of much scholarly work that the religious beliefs of scientists are a function of their commitment to science. The findings presented here show that indeed academics in the natural and social sciences at elite research universities are less religious than many of those in the general public, at least according to traditional indicators of religiosity.
Assuming, however, that becoming a scientist necessarily leads to loss of religious commitments is untenable when we take into account the differential selection of scientists from certain religious backgrounds. Our results indicate that people from certain backgrounds (the non-religious, for example) disproportionately self-select into scientific professions. In contrast, being raised a Protestant and in a home where religion was very important, for example, leads to a greater likelihood that a scientist will remain relatively religious...
...Finding that the strongest predictor of religious adherence among this group was childhood religiosity recasts previous theories about lack of religiosity among academic scientists in a new light.
The idea that scientists simply drop their religious identities upon professional training, whether due to an inherent conflict between science and faith or institutional pressure, is not strongly supported by these data.
If this was the case, then religious upbringing would have little effect on religion among scientists, with even those scientists who were raised in religious homes losing religion once they entered the academy or received scientific training.
Instead, as shown by our results in Table 5, religious socialization and heritage remains the strongest predictor of present religiosity among this population of scientists.
When we compare this group of academic scientists to the general population we find that scientists were disproportionately selected from homes where there was no religion or where religion was not important.
The relationship between these backgrounds and low scores on measures of belief and practice exist in the general population, not just among scientists. For example, 15 percent of respondents to the 1998 General Social Survey who were raised in homes with no religion said they did not believe in God compared to 1.6 percent of those raised Protestant.
Academic science has a disproportionately large number of people raised with no religion, potentially producing many more people who do not believe in God. Since these data do not allow comparison of religious switches during the life course we cannot determine conclusively what kind of predictors function to keep scientists in one type of religion compared to another (Sherkat and Wilson 1995). A panel data set that follows the same group of faculty over time would be better able to address these kinds of issues.
Of course, this obviously raises the question of why scientists are self-selected from non-religious households or households where religion does not play a major role. This is a question that will need further exploration beyond the data presented here.
We will, however, provide some possible routes of inquiry. In some cases this selection effect may indeed be due to the tension between the religious tenets of some groups (e.g., those that advocate young earth creationism) and the theories and methods of particular sciences.
On the other hand, some of the selection effect may simply be due to differential emphasis on education and\or differential resources. These could be mediating factors between religious background and likelihood of becoming a scientist or independent from religious background. The possibility of such mediating factors reveals that the story is unlikely to be the simple one of �religion is contradictory to science and hence religious individuals do not go into science.� Finding that scientists who are raised religious often stay relatively religious casts doubt on this simple cause and effect scenario.
Another possibility is that religious individuals might select into science graduate programs equally but that the graduate programs and scientific environments themselves have strong anti-religious messages and reward structures, either passive or active, such that some abandon their faith in the process and others leave programs.
To study the previous we would need data not just on faculty at elite institutions but a data collection including a broader set of individuals in the academic sciences (graduate students, postdoctoral fellows, researchers, in addition to faculty) as well as the ability to follow these same individuals over time.
Matt,
Thank you for this very enlightening post. I find the "resources" possibility intriguing. Many studies have shown that religiosity decreases with wealth. It may be that those who were raised in wealthy families and thus were more likely to be raised in nonreligious families simply have greater resources to pursue the advanced education necessary to becoming a scientist rather than, say, a lab technician, science teacher, etc. People raised in less wealthy/more religious backgrounds may have an interest in science but not the resources to pursue it as a career.
Matthew, the tables mentioned in the text are not visible so we cannot actually see the data, just their conclusions. They seem to find a big difference between those who were raised in protestant households where religion was very important and those raised in other more moderate or non religious households. I would like to see the numbers involved here as its my impression that 'very religious households' should be a minority figure. I doubt that you would find many atheists who would disagree that these individuals are the ones most likely to end up as religious adults or that loss of religion would be in some way inevitable for such a group.
I was referring to outstanding scientists, like the National Academy of Science members in the Larson-Witham study, not to average scientists.
http://www.stephenjaygould.org/ctrl/news/file002.html
Larson and Witham found that only 7 percent of NAS scientists actively believe in a personal god. Ten times as many---72 percent---actively disbelieve. About 21 percent expressed "doubt or agnosticism".
So it seems that a clear majority (72 percent) are atheists in a fairly strong sense (disbelieving) and the overwhelming majority (93 percent) are atheists in the general sense (not believing).
I'm pretty sure that of the 7 percent theists, a big fraction are not the kind of orthodox theists most Americans are. (Deists or something.)
So it seems not only that average scientists are several times more likely to be atheists than other folks, but outstanding scientists are several times less likely to be theists than average scientists.
It is also noteworthy that scientists in fields with substantial theological implications are more likely to be atheists than other scientists. Evolutionary biologists and theoretical physicists are overwhelmingly atheists. (Professional philosophers, too, BTW.)
Thank you for the elaboration. I still have some big problems and I wish that I could see the figures.
Going only by your summary (I'm not able to look at the data), the conclusions still do not support the case that Dawkins or Meyers are incorrect. Religious belief is definitely still present in elite scientists, but the argument which Dawkins & Meyers are making is that scientific education reduces the extremes and generally leads to increased atheism and decreased fundamentalism. What you've cited here seems to treat religious belief as a binary value, you have it or you don't, which makes it impossible to test the important changes in belief which are thought to happen.
"Assuming, however, that becoming a scientist necessarily leads to loss of religious commitments is untenable when we take into account the differential selection of scientists from certain religious backgrounds."
And who is saying that becoming a scientist necessarily leads to a loss of religion? No one that I'm aware of.
"Instead, as shown by our results in Table 5, religious socialization and heritage remains the strongest predictor of present religiosity among this population of scientists."
Again, this does not undermine anything that Dawkins or Meyers are arguing, only that atheism is prevalent in social sciences and other disciplines. This shouldn't come as a big surprise at all. Some of their assumptions about "hard" science leading to atheism seems weird considering the concentration of Engineers and physicists who are Creationists.
It's an interesting study and the self-selection raises very disturbing questions about the future of science in an extremely religious culture, but I don't think the conclusions you draw are supported by the data, at least not what has been presented. Perhaps you could explain it to me.
Matt, thanks for the copy-paste. I don't want to ask too much, but if we could get the data as well, that'd be great.
The Study:
This actually seems to be direct evidence for my argument, though. The percentage of the general population who are atheists (not "agnostic" or "non-religious", but people who affirmatively believe in the non-existence of gods) is quite small, taken to be ~4% (~12 million). And this is general population, not people raised in a specifically conservative (as I'm taking the point) Protestant household. I would argue the point that a 14% conservative Protestant upbringing to atheist "conversion" rate is markedly higher than the general population CPU->A rate.
Further one step, this is sociology, a field where I don't know of any specific confrontation between religious truth claims and scientific truth claims. Perhaps a comparison with paleontologists, biologists or cosmologists might be more interesting.
However, to be fair, I am slightly perturbed by the 46% (implied) belief in God by non-religiously-raised sociologists. The data should bear some interesting answers.
Adrian,
Actually, as far as I can tell, the only argument I have heard from Dawkins, PZ, and among commenters here at Scienceblogs is that advanced training in the sciences burns away your religious belief. Nothing to the effect it makes you less extreme, more moderately religious etc.
As PZ says: More science literacy, less religion, more science. Apparently it's that simple.
It seems to me you're going towards the same reductionism you see in Myers and Dawkins. 1) Even if they have said that there are some kind of "more science = less religion" equation, you can not use this study to "demonstrate" they are wrong, since this study does not go in that direction. 2) Moreover, even if they were right, even they would argue that the disappearance of religion would be a very, very, long process, since the scientific way of thinking as historians would define it, is emerging since the Middle Age.
So if your objective is to see advancements in science communication, you can not evaluate the Dawkins-Myers duo as if they were the symbols of some kind of jamming.
Matthew you are throwing up a strawman that only you seem to take seriously. Do you seriously think that PZ and Dawkins think that more science education will cause every single person to turn less religious? What they are suggesting - in my opinion - is that within a population of religious moderates if you increase science education then a proportion of this group will tend to lose their religious beliefs. There is nothing inevitable about it for the individual.
Indeed the "more science literacy, less religion, more science" idea seems to be pretty much what Francis Collins and Ken Miller believe, Collins for instance advocating moderately religious scientists to be involved in establishing a more moderate 'new theology' that takes into account scientific facts that contradict biblical 'facts'.
Back to the data question.
Is there a figure for the percentage of moderately religious households and the percentage of moderately religious scientists who came from these households? Presumably, if your hypothesis of self selection is correct, then these numbers should be roughly equal. If, however, there are many non religious scientists who came from moderately religious households the hypothesis will be falsified.
Its a pretty straightforward question that can be answered in a second by showing us the data.
Nisbet: "Actually, as far as I can tell, the only argument I have heard from Dawkins, PZ, and among commenters here at Scienceblogs is that advanced training in the sciences burns away your religious belief. Nothing to the effect it makes you less extreme, more moderately religious etc."
I have not seen articles or posts of the "atheist noisemen" where they state that the loss of religious beliefs due to scientific training is absolute and/or instanteneous. On the other hand, it seems that even Nisbet acknowledges that scientific training has a moderating effect on religious beliefs. So where is the real disagreement?
If scientific training has a moderating effect on religious views, why does it not have the potential to "moderate" them away completely? It might be that this total loss of faith happens only to a minority of science students, but it is plausible that with increasing knowledge in science, religion plays a lesser part in explaining world events. If one knows more about scientific causalities, there is less incentive to imagine divine interventions, I think.
This possibility of "dilution" of faith makes some religious people suspicious of science, of course. Is this a good enough reason not to speak out about the moderating effect of science education on religion? I don't think so.
"I was referring to outstanding scientists, like the National Academy of Science members in the Larson-Witham study, not to average scientists."
The problem comes in defining "outstanding scientists." NAS members are impressive, no doubt, but we'd need a lot more data to determine whether membership is wholly a product of one's scientific ability and knowledge or is predicated upon other factors. Advancement in science, like many other fields, involves a variety of factors: personal psychology (i.e. commitment to career advancement, social participation in the scientific community, etc.), group psychology (in group and out group dynamics, social mores, etc.), personal priorities and their ordering (career, family, religious participation, hobbies, charitable work, politics, etc.) and other socioeconomic factors (wealth, gender, sexual orientation, race, etc.). Other than levels of atheism, what other demographic differences are there between NAS members and most scientists or the general public? I think we'd have to answer that question to get any real elucidation on the question of science and its relationship to atheism.
Considering the nomination/election process and the disproportionate underrepresentation of many groups (like women), I would hypothesize that the disproportionate representation of atheists in the NAS may involve factors other than a link between scientific ability/knowledge and atheism. I would also hypothesize that NAS membership involves factors other than being an "outstanding" scientist, whatever that means.
Now, if only I were a scientist, I could do the study myself. Since I'm not, any takers?
I'm sure it's an imperfect measure of anything we'd want to measure, but most things are, and I don't know of a better measure of scientific "outstandingness" than being elected to the NAS.
(I personally find it plausible that the imperfection of that measure would understate the correlation between (relevant, advanced) scientific knowledge and atheism than overstate it. Either way, the correlation is clearly strong and the real question is which direction(s) the causality goes---or whether they're both caused by other factors, and what those factors are.)
The statistics used in the paper that Matt cites have a grave weakness, in that they classify people in a binary way, as scientists vs. not scientists.
Most scientists I know are no great shakes as scientists go. A few are clearly way better scientists than most. Some of them are NAS members, and a few have Nobels and the like.
The NAS numbers seem to me interesting and relevant. They seem to show that whatever else is going on, and whichever way(s) the causality goes, the simple binary distinction between scientist and nonscientists is a pretty blunt instrument; it understates the very strong negative correlation between scientific knowledge and religiosity.
My own impression from knowing a lot of scientists, and knowing the work and views of a lot more, is the better a scientist is, the more likely it is that he or she is an atheist.
It may be rhetorically handy to say that many scientists are religious, but I think it's pretty clear and deeply interesting that (0) most scientists are not religious, (2) most religious scientists are less religious and/or less orthodox in their religion than other people, (3) relatively few outstanding scientists are religious at all, and (4) very few outstanding scientists are religious in the way that the general run of Americans is.
The statistics may not be very precise, but it certainly appears that high-achieving religious scientists like Collins are the exception, not the rule.
It appears that for whatever reason, Christians aren't pulling their weight, science-wise. Per capita, they do less science, and far less great science, than average. The slack is mostly taken up by atheists, who do far more than their share of science and especially great science.
Atheists are overrepresented in science even more than Jews are, and I would guess that overrepresentation of Jews is largely due to the fact that many Jews in science are cultural Jews who are actually atheists.
(I'm not making a claim about the causality there... maybe the Jewish culture they come from allows for both atheism and scienciness, rather than one causing the other. I'd just guess that irreligion is a better statistical predictor of scientific achievement than Jewishness, whatever the underlying mechanisms.)
If this all comes down to how religiously people are raised, that suggests that Christianity is, in practice, quite anti-science; they just don't produce a lot of science-oriented and science-abled people.
The 80+ percent of American households that are religious are outperformed by the ~4 percent of households that are atheistic.
If that's true, it's very impressive. It means that atheists out-science Christians in science by more than an order of magnitude.
The tiny atheist minority is doing an enormous service to the Christian majority, doing at least a plurality of of the science, and apparently the majority of great science.
Matti K.:
I don't think that anyone is arguing that the potential isn't there. The question is: Does it actually happen, and if so, to what extent?
I do think that whether or not science challenges your faith depends in a large part on how much your faith was in opposition to science in the first place.
Without having any data to prove or disprove this, compare someone like Michael Shermer, who was brought up in a creationist church, with someone like Ken Miller, who was brought up as a modern post-Vatican II Roman Catholic. Shermer's childhood faith is, I think, more antagonistic to science, and as such, he seems like a more likely candidate to lose his faith when he studies science.
This theory doesn't explain Richard Dawkins, who was brought up as an Anglican. Certainly, from what I've heard, I don't think that Dawkins really struggled with it, he just realised that he didn't believe. But that doesn't explain his strong anti-theism, which might actually be more a product of having real problems with the established church.
"It may be rhetorically handy to say that many scientists are religious, but I think it's pretty clear and deeply interesting that (0) most scientists are not religious, (2) most religious scientists are less religious and/or less orthodox in their religion than other people, (3) relatively few outstanding scientists are religious at all, and (4) very few outstanding scientists are religious in the way that the general run of Americans is."
This is clear BUT the "why" is not. Unless a study controls for the variety of demographic factors involved, we cannot say that the numbers support a substantial link between science and atheism any more than we can say that the dearth of women and racial/ethnic minorities in the NAS demonstrates a substantial link between maleness/whiteness and being an outstanding scientist.
"Either way, the correlation is clearly strong and the real question is which direction(s) the causality goes---or whether they're both caused by other factors, and what those factors are."
Correlation does not imply causation. To argue otherwise is logically fallacious. We would have to control for ALL other relevant factors before demonstrating even a weak causative link.
For a similar demographic "problem" in the sciences and amongst NAS members, I'll refer you to the National Academy's report on women in academic science and engineering: http://www.nap.edu/catalog.php?record_id=11741#toc
A brief quote from the summary:
"Women are a small portion of the science and engineering faculty members at research universities, and they typically receive fewer resources and less support than their male colleagues. The representation of women in leadership positions in our academic institutions, scientific and professional societies, and honorary organizations is low relative to the numbers of women qualified to hold these positions. It is not lack of talent, but unintentional biases and outmoded institutional structures that are hindering the access and advancement of women. Neither our academic institutions nor our nation can afford such underuse of precious human capital in science and engineering. The time to take action is now."
Back to the relevant topic, I believe from demographic data that we'll see that there are institutional, social, and cultural factors that hinder access and advancement for the most "religious" segments: women, ethnic minorities, and the less wealthy. As a byproduct, the levels of religiosity are out of proportion.
Correlation does generally suggest causation of some kind.
That fact that A is correlated with B doesn't necessarily imply a that A causes B. Maybe B causes A. Or maybe C causes both, with neither A nor B causing the other, as I said in the passage you quoted. But something causal is evidently going on somewhere; atheism and science are not causally unrelated. This is obviously not just a coincidence, and the correlation is so strong that something very interesting must be happening.
Everybody here knows that, including you, as is evident from what you say later in the very same comment:
Maybe I should have made it clearer that my repeated hedging about the direction(s) of causality does (of course) apply to the hypothesis that Christianity inhibits science interest and/or ability. That hypothesis could be flatly false, or fail to account for most of the effect observed.
(Not the italicized If in my posting, however. I was saying that if it's simply a matter of religious environment of origin, then atheism seems to out-science Christianity by an order of magnitude or so. I don't actually think the truth is going to be that simple.)
I think you are likely right that there are institutional, social, and cultural factors at work.
I personally don't find it plausible that (for example) bias by atheist scientists against theist scientists accounts for the-order-of-magnitude difference in science productivity between the underlying Christian and the atheist populations. I don't think that it's most of it.
I also don't think it's entirely or even mostly a matter of discrimination against poor or female people, and thus (indirectly) against religious ones. My understanding is that the correlation between science and atheism is stronger than the correlations between science and wealth, science and gender, etc.
I could of course be wrong; this is a hairy subject.
Pseudonym: "Certainly, from what I've heard, I don't think that Dawkins really struggled with it, he just realised that he didn't believe. But that doesn't explain his strong anti-theism, which might actually be more a product of having real problems with the established church."
I think his "problems" are common with those who speak out loudly against homeopathy. Why search hidden reasons behind their agenda, when they spell it out very clearly?
It seems that you consider anti-theism necessarily to be some kind of mental disturbance, which needs to be explained by some past psychological trauma. If so, what other anti-stances need similar analysis?
"Correlation does generally suggest causation of some kind."
Again, that would be a violation of the rules of logic. Imagine, if you will, the supposed vaccine/autism link. With increases in vaccination, there was an increase in autism. Scientific testing has shown that these two are completely unrelated and that much of the "increase" in autism is related to a change in the diagnostic criteria.
As for their being a "causative" link between science and atheism. I believe that there may be some weak link but not in the way that Dawkins, Myers, et al like to think, that the theories of science lead to atheism. There is the issue of self-selection, where an atheist may be more likely to pursue a career in the sciences or where religious people (due to religion or a whole host of other factors) may be less likely to choose science as a career. People from certain economic backgrounds (which overlaps with ethnic/racial minorities), for instance, may go for fields where they can have immediate "guarantee" of a decent-paying job after a 4-year degree. Indirectly, this could decrease the number of religious people entering the sciences.
There are also all of the issues I mentioned, where there is tremendous evidence that those most religious segments are less likely to advance in the sciences despite their abilities/qualifications. I don't think they're failing to advance due to religion, but that the decrease in religiosity as you go further up the science ladder is an indirect result of other biases and institutional structures. Anti-religious bias amongst atheist scientists is conceivably a factor but I doubt that it's a determinative one considering how few atheist scientists are genuinely anti-theistic.
"I also don't think it's entirely or even mostly a matter of discrimination against poor or female people, and thus (indirectly) against religious ones. My understanding is that the correlation between science and atheism is stronger than the correlations between science and wealth, science and gender, etc."
I think the order of magnitude is based on a combination of factors, with each contributing. Discrimination, self-selection, wealth, gender, personal obligations, life priorities, etc. all contribute.
One of the main reasons I doubt the atheism/science link is that changes in worldview rarely (if ever) come about through reason alone. Mostly, a whole host of factors are involved but the "convert" will couch the conversion memory in terms amenable to the current worldview. Christian converts will focus on the great epiphany, for instance, because this is an important part of Christian mythos and has been since Saul of Tarsus became St. Paul on the road to Damascus. Jewish converts will focus on years of study, consideration and meditation. Atheist "converts" may focus on reason or science. In all of these cases, there are probably a whole host of psychological and/or emotional factors involved that get "forgotten" in the process. (Please note I say this being a Jewish convert myself.)
Michael Shermer has great insights on this in his book Borderlands Science. He doesn't go much into the implications of the research for religious conversion and in one instance, completely fails to see how it applies, BUT it's an extraordinarily interesting and enlightening book. I'd highly recommend it.
Melinda, I've already explained this twice at least.
The dictum that "correlation doesn't imply causation" does not literally mean that an observed correlation doesn't suggest that there's some causal explanation.
It means only that a correlation between A and B does not imply that A causes B. (Or, if we swap labels in the first statement, which is symmetric, that B causes A.)
You obviously know this on some level---your own hypotheses assume that there are causal links from common causes (gender bias, affluence effects, etc.) to A and B.
I really do understand the basics of correlation and causation. (I'm a fairly accomplished, tenured scientist with graduate work in philosophy as well as science.) I am not making the kind of gross, elementary error you obviously think I am.
It is an utterly basic part of the scientific method to look for causal explanations for correlations. Often those causal explanations do involve non-obvious variables.
It would be very surprising if there were not causal links connecting science and atheism. That does not deductively imply that the atheism causes science, or that science causes atheism. I agree with that.
On the other hand, I don't think your alternative explanations can account for the magnitude of the effect we're talking about.
Suppose that we assume that women, nonwhites, and poor people are discriminated against... would that make the correlation go away? I don't think so; it wouldn't come close.
If we just look at relatively affluent, relatively educated white men, there's still a large correlation between science and atheism. Most well-off educated white guys in the general population are theists, though admittedly not by as much as the population as a whole. Most well-off educated white guys in science, on the other hand, are atheists. So there's a correlation remaining to be explained. And in the upper levels of science, such as the NAS, the correlation seems to be very strong.
Whether atheism causes science, or science causes atheism, or neither causes the other but something else causes both, there's clearly something to be explained causally.
It could be that advanced scientific knowledge erodes religious belief. It could be that religious belief interferes with scientific thinking. Or it could be that there's an Evil Atheist Conspiracy that has taken over science (or academia more generally) and discriminates against theists in a completely unfair and unjustified way. Or it could be something else entirely that we haven't thought of.
I don't think it's just a matter of gender bias and/or effects of racism and/or finances. Those may all be very real, and may all be fairly important, but they are not strong enough to explain the enormous difference in scientific achievement; they can't account for the very large differences between demographically similar atheists and theists. (Especially but not exclusively orthodox theists.)
Melinda,
About the vaccination/autism example... I don't think this case is like that.
In general, correlation suggest causation of one thing by another, or the other way around, or causation of both by some other thing(s).
In the case of the vaccination/autism example, one problem there was that there wasn't really a correlation between vaccination and autism at all, just a correlation between vaccination and reports of autism.
Also, the two things being correlated were temporal trends. In that case, there's a third variable, T, i.e., time. There was a causal link from T to A, and another from T to B. (Vaccination and diagnoses both increased with time.)
I don't think this is that kind of problematic case. (Not that it's not hairy; it's just not the same kind of "correlation implying causation.")
I do think---and I think we agree---that there are many other important variables and causal links yet to be figured out.
One thing that makes me interested in this is that the effect is so large that several interestingly different theories could be mostly right, each with a fairly large effect.
Melinda, apologies in advance (modulo posting delay) for my second-to-last comment's tone and failure to engage properly with your comment. I shouldn't have waited until I had time to do better. (And probably still should.)
Paul, your apology is accepted. Please accept mine: I've been arguing not just against you but against the many people who pull this study out as "proof" that one has to be an atheist to be a good scientist or that a full understanding of science necessarily leads to atheism. These are arguments made in many instances by Dawkins, Harris, Myers et al and repeated ad nauseum by the less educated sycophants in the blogosphere.
My problem is that we don't have detailed demographic comparisons based on subgroups. What we have is a very general study that shows that the NAS differs from the mainstream scientific community and the general population in this one way. I don't doubt that there could be some causative links, but I believe that controlling for all of the other demographic factors and examining carefully why these links exist is better science and better logic than simply pointing to a correlation as proof of causation (which many prominent atheist scientists have done). I've searched and searched and found no studies that do any kind of detailed analysis of the issue. Only this study and previous ones conducted by a similar methodology. So, I have only hypotheses based on how demographic differences between the NAS, the scientific community and the general population have worked out for other groups.
I think that doing the kind of comparisons you describe, for instance, between upperclass white males in the NAS, the scientific community and the general population would give us greater insight. It doesn't seem, however, that that research has been done by those who wish to propose a causative link between science and atheism.
I have searched for demographic studies of religiosity amongst upperclass white males and have failed. The GSS cuts off at the "over $25,000" income group and would offer little insight into the upperclass. However, it does show white males leading all other groups by substantial margins in never attending church, believing the Bible is a book of fables, and having no religious preference.
Why cite Richard Dawkins in this context ?
If you have read Dawkins you will know he lost his faith when he was in his mid-teens. In otherwords BEFORE he started learning about ADVANCED science. He would have been doing O or A level science at the time, not degree level stuff.
Melinda,
You might want to look at pewforum.org for some information relevant to what we've been talking about.
For example, there's some income/religion data buried in this religion-and-politics report:
http://pewforum.org/docs/?DocID=61
(See pages 12 & 13 of the .pdf.)
Seculars are more likely to have incomes over $75K than committed white evangelicals or committed black evangelicals, but by considerably less than a factor of two.
If we just look at whites, that's a difference of a few tens of percent, nowhere near the kind of orders-of-magnitude effect we'd need to explain why atheists do so much more top science than Christians, per capita.
You can also play around with generating GSS "quick tables" to get a feel for the magnitude of race and gender vs. religion effects:
http://sda.berkeley.edu:8080/quicktables/quicksetoptions.do?reportKey=g…
It seems that you find effects that are usually less than a factor of two.
So even if we have a factor of two for income and a factor of two for white males vs (say) black females, that gives us at most a factor of four (if they're multiplicative) and more likely two or three.
That leaves a whole order of magnitude unexplained. Each of those effect may be big---a factor of two is a lot---but there's something a whole lot bigger going on in the inverse correlation between religiosity and advanced scientific achievement.
A caveat may be in order about whether those data use the right measures, for example with income topping out at "greater than $75,000". I suspect the income measure is pretty good; my impression is that most scientists come from middle- and upper middle-class families, so that's near the sweet spot for scientists in general. It might not be as accurate for top scientists.
(I don't know if there's a correlation between great wealth and great scientific achievement; I suspect there's not a major one. If you've looked at NAS demographics in detail, you might know better.)
Melinda has pointed out that only 6 percent of the NAS members are women.
I think it's interesting to try to take that into account in thinking about the high incidence of non-theism among NAS members...
Suppose that women at that level of scientific achievement were typical of (U.S.) women in general. About 5 out of those 6 percent would be theists.
But only 7 percent of the NAS members overall are theists. If the women were typical, that would make the men even more atypical. Of the 94 percent men, 92 percent would be non-theists, so only a bit over two percent of the men would be theists.
That's far, far lower than in the population at large, or the male population, or the population of U.S. males with graduate degrees.
Using the GSS quick tables feature I linked to above, we can see that the large majority of white men are theists---including white men with graduate degrees.
(About 2/3 are Protestant or Catholic. Of the remaining categories, I'd guess that only about half are actually nontheists; the other half are probably unaffiliated christians, generic nonchristian theists, religious jews, etc.)
So realistically, roughly 15 percent white men with graduate degrees are nontheists. But about 93 percent of NAS members are. (Even more if the men are less theistic than the women.)
So among men, the NAS is about 6 times as atheistic as you'd expect if it was simply dominated by white men with graduate degrees.
That's roughly comparable to the factor-of-5 effect of sex. (Melinda says on her blog that about 30 percent of the science Ph.D's go to women, but only 6 percent of NAS members are women.)
Women are several times less likely to get into the NAS than men, and theists are several times less likely to get into the NAS than atheists.
If I've got the reasoning right, however, neither effect accounts for the other. Whatever the selection mechanism is that filters out most women, the factor-of-about-6 selection for nontheism is either over and above that or mostly independent from it. (As well as being over and above any selection on the way to getting a random graduate degree.)
Some data on religiosity and income, suggesting that high family income doesn't explain much of the observed atheism among scientists:
First a couple crude measures, showing the effect of income on high religiosity, i.e., whether people (in the U.S.) say that they pray every day, and whether they say religion is "very important" to them.
http://www.umich.edu/news/index.html?Releases/2004/Sep04/pray_daily
You can eyeball the regression line plot and see that by either measure, about 57 percent of people are very religious by these measures. (Look at the middles of the regression lines to get the average.) At the high end of the income scale, they're down to about about 48 percent. Conversely, the people who are not very religious are about 100-57=43 percent on average, or 100-48=52 percent on the high end. So the rich are about 52/43 = 1.2 times as likely to not be very religious, i.e., 20 percent more likely.
I think roughly the same is true for lower degrees of religiosity---high income may make you tens of percent more likely to be irreligious, but nowhere near a factor of two.
Unfortunately, I don't have quite the right stats at my fingertips.
Here's some supporting data from the Pew Forum. If you go to this page
http://religions.pewforum.org/reports/detailed_tables
and look under "income" and click "by religious tradition" you'll get a PDF showing what percentage of various groups make incomes in several ranges. In the upper two ranges, ($75-100K and over $100K) you'll see that atheists, agnostics, and unaffiliated seculars have larger percentages than the population as a whole, but the percentages are only about 50 percent higher. (E.g., 18 percent of population is in the top bracket, vs. 25 percent of agnostics.) That is, nontheists are overrepresented among the well-off by tens of percent, but by not nearly a factor of two.
While race, sex, income, and education are all correlated with science achievement, even most white, male, highly educated, well-off people are theists.
There must be a larger factor unaccounted for to explain why most scientists aren't theists, and a huge one to account for their overwhelming majority in the NAS.