Most wealthy countries in the world, including Japan and much of Western Europe, are not particularly religious, with fewer than 25% of their citizens professing belief in God. The United States is a notable exception. Although it has high per capita income, it also has high religiosity — around 60% of us believe in God. Sociologists have produced a lot of theories about why America is an outlier in this respect, but several recent studies are converging on an answer: insecurity.
I found the latest of these, by medical writer Tomas Rees (author of the blog Epiphenom), described the latest New Humanist. Unfortunately, Rees’s article is not one of the pieces posted online, but you can find his original article, published in The Journal of Religion and Society, here (click on the lick at the upper right of the page). Rees also has a blog post on the topic here.
What Rees did, to make a long story short, was to calculate (using the Gini statistic) the degree of income inequality among citizens in each of 67 countries and then correlate that with his index of religiosity, which Rees took as the frequency of daily prayer not involving prayers uttered in church. Here’s the correlation he got among fifty-odd of those countries:
Fig. 1 from Rees paper. Correlation between the mean frequency of prayer and income inequality (the mean of the log Gini index for the period 1971-1996).
What does this mean? Rees interprets this as showing that religiosity is higher in those countries whose inhabitants are less secure. He takes income inequality as the measure of security. Is he justified in doing this? Well, his assessment of other classic factors used by sociologists to show societal insecurity (life expectancy, infant morality, homicide, perception of corruption) shows that most of these correlate strongly with income inequality (three factors show a weaker correlation: “prevalence of curable STDs,” “child well being,” and “non-vehicle property crime.” He can thus use income inequality as a proxy for these other factors. He also ran multiple-regression analyses to try to eliminate spurious correlations.
Average well being, as indicated by per capita gross domestic product, was also negatively correlated with religiosity, but in multivariate models was not as strong as income inequality itself in explaining religiosity. Most notably, while the US was an outlier in the religiosity/average income correlation, it was not in the religiosity/income inequality correlation. (See Fig. 1 above).
Conclusion? Rees submits that there is a causal relationship between national insecurity (as indexed by income inequality) and religiosity. But what causes what? Rees chews on two explanations:
One possibility that cannot be excluded on the basis of available evidence is that religiosity, or some component part of it, directly or indirectly worsens these key aspects of personal insecurity. Such effects could be case specific and need not be a consequence of religion in its broadest sense. . .
You could think of reasons for this. For example, a certain religious political party (say, Republicans) could foster government policies that make people less secure (say, opposition to universal health care).
The other reason sounds a bit more plausible to me:
The alternative possibility, that religiosity is not a cause but a consequence of personal insecurity, appears plausible. Indeed, this may also help explain the inverse correlation between per capita GDP and religiosity observed in this study and in previous studies. The link between modernization and secularization, as measured by personal religiosity, is likely to be at least in part driven by the fact that people in wealthier nations lead more secure
lives.
And, of course, causality could run both ways, with the factors interacting to cause a growing spiral of increased religiosity and insecurity.
Here’s Rees’s conclusion:
In conclusion, the current analysis ties together and explains two apparent paradoxes. First, the observation that modernization, in terms of average material wealth, appears linked to secularization in some countries but not others. The key to this paradox is that it is not simply average wealth, but also the distribution of wealth and the degree to which wealth is used to improve average personal security, which in large part determines religiosity. Second, the observation that religion, although generally believed to have a pro-socializing effect on the individual level, is associated on the macro level with societal ill health. This is most likely because personal religiosity is in part a response to adverse social environments, but that aggregate religiosity does not significantly ameliorate them.
Rees’s results are not a one-off: he cites several earlier studies supporting, at least, the correlation between religiosity and indices of insecurity. There’s also an important study, conducted in 2009 by Gregory Paul, that apparently shows the same result.
Paul constructed what he called a “successful societies” scale, incorporating many of the same factors Rees used, as well as others (his 25 factors included prevalance of homicides and suicides, life expectancy, duration of marriage, measurements of life satisfaction, indices of corruption and so on). He showed that, among 17 developed Western countries, and Japan, there was a strong negative correlation between societal health and religiosity; in other words, less successful societies were more religious. Here’s Paul’s plot, with the countries labelled as initials (“U” = US, “J” = Japan, “H” = Holland, “T” = Italy, “N” = Norway, and so on).
Fig. 2. (Fig. 25 from Greg Paul’s paper). Correlation among nations between belief in God and ranking on “successful societies scale”
And again, we’re not sure how the causality runs here, but I suspect that insecurity does promote religiosity to some extent. Note that this doesn’t necessarily refute other explanations, such as Pacal Boyer’s, for the fact of religiosity. It merely explains some of the variance in religiosity.
What can we say about all this? Well, I’m not staking my life on these results, but I find them intriguing. (See some criticism by Susan Blackmore here.) Nevertheless, Rees and Paul may be onto something. And if they’re right, even in part, then we atheists have a bigger task than simply trying to dispel the influence of religion on people. For to do that, we may have to work for better and more just societies. But isn’t that, in the end, a nobler goal?
________
Rees, T. J. 2009. Is personal insecurity a cause of cross-national differences in the intensity of religious belief? J. Religion and Society 11:1-24
Paul, G. 2009. The chronic depedence of popular religiosity upon dysfunctional psychosociological conditions. Evol. Psychol. 7:398-441.
Earlier paper by Gregory Paul (online): Paul, G. 2005. Cross-national correlations of quantifiable societal health with popular religiosity and secularism in the prosperous democracies. J. Religion and Society 7 (link here).
I tend to agree with Susan Jacoby’s analysis concerning the historical roots of anti-intellectualism in the states. The after effects of such emotionally patriotic movements as the 1st and 2nd Great Awakening, the Red Scare and the Christian Campus Crusades have set the stage for “what it means to be an American.” Perhaps this has sent a reverberation of distrust and insecurity by proxy? As long as the idea of “America” is conflated with self-righteous piety, the godless stereotypes will persist.
Jacoby did a terrific analysis but I don’t agree with her 100%
Sarah clarifies:
Nicholas Wade discusses this in his Templeton book The Faith Instinct – but in a rather cursory and unhelpful way. (Especially unhelpful in that he calls the US a nation with a reasonably equitable distribution of wealth! Which doesn’t shore up trust in the rest of his exegesis.)
Why the Gods Are Not Winning, by Gregory Paul & Phil Zuckerman, is one of my favorite Edge.org articles:
Update–quick Google reveals online PDF of Gregory Paul’s latest article:
The Chronic Dependence of Popular Religiosity upon Dysfunctional
Psychosociological Conditions
Good to see Blackmore getting more airtime. Still remember her running up and down the auditorium before her Millennium Lecture in Bath asking everyone “Are you conscious?”.
This is likely my preconceptions and prejudices speaking, but I’m honestly surprised to learn that Japan is so non-religious. But perhaps “absolute belief in God” is a bad metric for Shintoism and Buddhism. And prayers may well be more often offered up in temples/shrines than privately – a houseshrine may be considered either.
The prominent religion in Japan is Shinto, and it is very closely linked with nationalism. I think it is good for Japan and the world as a whole that it has lost its position. I’m not looking forward to another “great” war in the Asia-Pacific.
This is actually not news. The original data that made the link between religiosity and existential security were layed out in a 2004 book, Sacred and Secular. It was one of the most heavily referenced books used by Greg Paul in his work.
Incidentally I know Paul personally and last time I saw him he complained that prominent atheists haven’t given his work as much attention as he thinks he deserves.
Sacred & Secular laid out the hypothesis – when my New Humanist article comes on line you’ll see that I explain that. But their book but didn’t test it except in a simple regression. What my analysis adds is a multivariate regression including other national-scale factors that are hypothesized to affect religiosity.
Can you explain what R^2=0.42 means? College stats was a long time ago.
It means that 42% of the variation in religion is ‘explained’ by variation in income inequality. Of course, this is a correlation, so the causality could go either way, or be indirect, or be a fluke. To judge which is the case here you need to look at the theories that might link the two.
It’s also worth pointing out that income inequality is linked to GDP, which is in turn linked to religion. So to get a better grasp of the connection between religion and inequality, you need to adjust for GDP and other relevant factors (which I do, to some extent, in my paper)
I once heard an interview with a conservative pastor associated with a church that was deeply involved in charity work and was also vocally against government social programs. His point was that secular programs that provided a safety net were the devil’s handiwork because then when bad things happened in peoples lives they would turn to the government for help and not turn to God. To paraphrase, he felt the government was poaching his clients (to be fair, he did believe that his services were just as good as the governments and he also provided access to eternal life as a lagniappe).
The problem as I understand it is that these models don’t explain much of the variance. Apparently there are other factors that need to be identified as well. But it seems like a straightforward factor (never mind which way causality runs, or in other words the underlying process(es)), so I assume the result won’t go away.
And of course, correlation runs both ways. To dispel the influence of religion on people is to work for better and more just societies.
Because lying about facts is never a good thing.
[And of course because of the rotten morality in religion, it’s terrible influence on public health, et cetera. That list never stops growing, does it?]
Durn it! Well, mostly never a good thing. [And this is not a comfort blanket for medicine and other service providers who frame reality a tad much for it to be socially healthy.]
These data are awfully reminiscent of the “religion is the opiate of the masses” idea. The greater the economic disparity, the more those toward the bottom embrace religions that promise “reparations” in an afterlife for injustices in this life; and the more inclined they are to wait for those reparations instead of raising hell about injustices in this life! And that serves to perpetuate those injustices.
“opiate of the people” came to my mind immediately as well…
The return of the opiate. Maybe it never left. The reparation idea is a bit cynical. They do need social support religion offers more than future fluffy comfort.
…like your handle-Phiwilly
I’ve always thought of religion as being an affliction of the poorly educated. Poor education = no hope in hell of having a competitive economy. I don’t understand this obsession with tying everything to religion when simple lack of education should be obvious. The challenge is how to educate those people and, in many places, dismantle the authoritarian government.
@Ophelia: But all of us ‘murcans have been raised with “We’re #1!” and “We’re all equal!” and “You too can be a multi-gazillionaire (thanks to Lotto).” Social equality and equitable distribution of income are well perpetrated myths (those poor people are just lazy, so they don’t count as poor, of course Mr. X deserves his $X million per year salary …).
Yes, but Wade’s book has pretensions to doing considerably more than recycling folk wisdom of the Horatio Alger variety. And to be fair it does do more than that – but the bit about equal distribution isn’t one of the better parts!
Religion is, indeed, the self-consciousness and self-esteem of man who has either not yet won through to himself, or has already lost himself again. But man is no abstract being squatting outside the world. Man is the world of man—state, society. This state and this society produce religion, which is an inverted consciousness of the world, because they are an inverted world. Religion is the general theory of this world, its encyclopedic compendium, its logic in popular form, its spiritual point d’honneur, its enthusiasm, its moral sanction, its solemn complement, and its universal basis of consolation and justification. It is the fantastic realization of the human essence since the human essence has not acquired any true reality. The struggle against religion is, therefore, indirectly the struggle against that world whose spiritual aroma is religion. Religious suffering is, at one and the same time, the expression of real suffering and a protest against real suffering. Religion is the sigh of the oppressed creature, the heart of a heartless world, and the soul of soulless conditions. It is the opium of the people. The abolition of religion as the illusory happiness of the people is the demand for their real happiness. To call on them to give up their illusions about their condition is to call on them to give up a condition that requires illusions. The criticism of religion is, therefore, in embryo, the criticism of that vale of tears of which religion is the halo.
– Karl Marx
Starting from “X and Y are correlated” you considered three possibilities* –
X causes Y
Y causes X
“causality runs both ways”
However, there are additional possibilities. Here’s one that always seems to get left out:
There is an additional variable (or more than one), N, say, that causes both X and Y – the correlation is caused by them both being influenced by another variable.
(Like the classic example of “number of teenage pregnancies in a city is correlated with the number of churches” – with population driving much of that relationship).
*(well, okay, income inequality is a proxy for insecurity, so it’s X and Y’ I guess, but let’s take it – that income inequality measures insecurity – as given for the purpose of this part of the discussion)
Now I find the insecurity as a cause of level of religiosity entirely plausible, but we should at least be clear what the major alternatives are.
Don’t forget that x and y may have no direct causal link but may be caused by a third unspecified cause.
The example used in my intro statistics book was the observed positive correlation between baskets made by a basketball player and the number of fouls committed by that player. There is no direct causality between sinking a basket and committing a foul; however, both may be impacted by aggressiveness. Hence the mantra: correlation does not imply causality.
Yes, yes, this has been brought up before, apparently by people who didn’t read the paper, for Rees does worry about this and discusses it there.
R^2 = 0.42 ?
Cringe!
On logs, apparently?
Cringe again!
Nowehere near good enough.
After perusing Rees’ paper, I can see a number of potential issues with the statistics.
The association may not be linear (no reason why it should be), the scaling may not be adequate, distribution effects seem not to have been dealt with, and there are certain basic issues with the Gini coeff per se.
Not sure about the particular model, not sure about this linear regression approach to what seems essentially a multi-dimensional problem with possibly nested dependencies either.
Sorry to spoil the party (again), but I think this matter deserves a little more delving into, and a little more number-crunching.
The Gregory Paul paper was way more convincing.
The Goddess of Statistics, blessed be Her Logarithm, abhors a vacuous correlation.
Shhh … not so loud! And don’t mention the small samples and how when you combine them properly you come to the conclusion that there’s really nothing of interest. You’ll upset some data miner’s feelings.
Yikes…makes them feel insecure…they may lose faith…
Don’t get hung up on R^2: it is a measure of relative explanatory strength not of validity. A simple way to get an appreciation of this is to open up a spreadsheet (or other suitable program) and create a (x,y) data set with a linear relationship, tack on a random number and then perform a simple regression analysis. Now play with changing the size of the slope and the size of the random variable. Inherent in the data is valid relationship but the R^2 will vary dramatically depending on relative size of the coefficient and the size of the random variable.
In the physical sciences, that permit bringing a phenomenon into a lab, you can control for and minimize those random variables. Thus a high R^2 indicates that there was good experimental controls in place. Laboratory control is not possible for these types societal studies, so lower R^2 is expected. And just because the data is noisy does not mean that the perceived relationship is invalid.
I disagree.
1. Finding 58% of the variation not explained by my linear model would be very worrisome, no matter what. I’d look for alternatives.
2. I did not discuss validity, I questioned the linear measure of association as presented. What is your assessment of the validity, based on the data as shown?
3. The determination coefficient does not depend on the domain of the data (e.g. physical, biological, or social sciences); it is just a measure of linear association. If you expect lower R^2 in social studies (I don’t), this implies a lower expectation of linearity. This often points to the complexity of the problem, in which instance linear measures of association, such as LR, may be ill suited and even undermine your case.
4. “Noisy” data: one salient difference between experimental data and social sampling is precisely that large residuals in social sampling are not “noise”: they are due to multiple, complex, often interdependent factors. They are also often due to poorly or inadequately defined variables. A model which neither isolates nor accounts for them is a poor model.
1. If it is indeed the case that 58% of the variation IS due to a completely random process, then finding the 42% that is not random is the best that is theoretically possible. This is a glass half empty or half full situation.
2. I don’t have a problem with the statistical analysis per se but I do have, shall we say, qualms about using the prayer indicator as a proxy for religiosity. I think that since different religions treat personal prayer in very different ways that I wonder if those differences might skew the measurement of religiosity a great deal.
3. I don’t think that I implied that coefficients in a model were dependent on the field of study. But different field of study do allow for different levels of control over your observations. In physics you can put your detector at the bottom of lead mine, in chemistry you can use ultra pure reagents, in biology you can use genetically identical clones but these types of controls are not possible when looking at the messy world of the social sciences. A large part of the observations are going to capture, per force, random noise. It is this unavoidable randomness that causes the R^2 to be consistently low in these types of studies.
This inherent randomness is not part of the form of the underlying relationship (be they linear, multi-linear, or non-linear) – a researcher tries very hard to pick the model that can extract the maximum amount of information from the data but even if the research happens to pick a mathematical model that completely reflects the underlying reality there will always be a substantial residual error.
4. This statement seems to assume that social phenomenon have a precise deterministic (albeit complex)form. I have serious doubts that that is the case. To my mind, the fact that we can find any type of regularity in the behavior of these ‘damn hairless apes’ is astounding to the nth degree.
ps.
Paul, G. 2009. The chronic depedence of popular religiosity upon dysfunctional psychosociological conditions. Evol. Psychol. 7:398-441.
uh, YEAH.
“we atheists have a bigger task than simply trying to dispel the influence of religion on people. For to do that, we may have to work for better and more just societies. But isn’t that, in the end, a nobler goal?”
The only way it could be a nobler goal is if you don’t think that dispelling the influence of religion IS itself working towards a better and more just society. I happen to think it is.
In the first graph, is it statistically inappropriate to have one data point for high population contries; (India (1.1bn pop’n), USA(300m)), and also have one data point for low population contries (Romania (21m), Czech (10m)) ?
Should the countries be weighted by population?
Should the countries be weighted by population?
Of course. That’s what I meant by distribution effects. But that’s just one of the simpler issues.
You’d have to take social, economic, ethnic, cultural, religious diversity into account.
Urban vs. rural, industrialised vs. agricultural, homogeneous vs. polyethnic, etc. etc.
Take just two extremes on the population scale: USA and Switzerland, one roughly 50 times larger than the other. Both are in reality ethnic, cultural, and religious mosaics; averaging them is meaningless.
The idea behind the study is neat, the assumption seems plausible, but given the data available and the analysis as performed, it’s about as meaningful as the correlation of witchcraft and Firefox downloads:
http://xkcd.com/111/
Not to sound like a not-it-all, but it took them all that effort to figure something that obvious out. Insecurity promotes faith. Thanks for pointing out what we already know.
I presume you mean “know-it-all. ” And if I just made the assertion that “obviously, insecurity promotes faith,” readers of this website would demand EVIDENCE.
Interesting research and results, but I don’t see a very strong causation link. Sociology is such a complex subject, maybe rivaling biology. I don’t think that the religious phenomenon can be traced back to a single simple cause, like social inequality. I don’t want to sound like a right wing bigot, but this sounds a little Marxist to me. You see, experiments in the social sciences are rarely unambiguous.
Quite true. That’s what my study (and other similar ones) shows – insecurity seems to be an important part of the explanation, but not by any means the total explanation. Hopefully New Humanist will put my article online where I explain a little of the background. Until it does, you can read it via the Humanists4Science blog: http://humanists4science.blogspot.com/2010/01/why-is-religion-on-rise-in-so-many.html
Where does the 25% non-religious people in “high-income” countries figure come from? Great, so now market economy explains evolved minds-without religion?…gimme a break..
Could someone help me out here? I don’t understand Paul’s interpretation of his data, at least as Blackmore tells it. If religion is a “crutch to which people turn when they are under extreme stress,” doesn’t that support the religious impulse’s deep-seatedness?
If it has taken so long for us devise and achieve social arrangements that allow us to alleviate enough stress to the point that religious belief is only now subsiding, but still very far from a minority position, then doesn’t that indicate the challenge of attaining the irreligious world view for humans? In other words, what Blackmore describes as “extreme stress” appears to me to be the norm.
60% of the people in the US believe in God?!?
But we know already that 75 to 80 percent of us are Christian. Does that mean a large number of people who self-identify as “Christian” are actually atheists?
I’m pretty sure that we atheists make up only about 1o to 15 percent of the US, not anywhere close to 40%.
Agreed. I think the high teens percentage is for those who are unaffiliated with a religion. Atheists strong enough so Jerry won’t cry faitheist are surely under 10%, maybe even under 5%. BTW, I’m one.
I also think this is too strong an implication: “several recent studies are converging on an answer: insecurity.”
It is interesting to wonder how differing definitions of “successful societies” will change this analysis, or what difference using all countries, or changing to groupings other than “country” would induce.