Demography and irreligion

by John Q on July 30, 2015

A few months ago, I was a bit surprised to read a report put out by the Pew Research Center predicting that the proportion of the world population without a religious affiliation would decline sharply by 2050. The basic argument sounds plausible: an increase in the unaffiliated proportion of the population within countries will be more than offset by faster population growth in countries with higher rates of affiliation. The main points are presented in a peer-reviewed article in the journal Demographic Research, which suggests the analysis should be solid.

Still, I thought I would dig a bit, and found a longer version of the report here, including the projection that Christians would decline from 78.3 per cent of the US population in 2010 to 66.4 per cent in 2050. That seemed like a very slow rate of change, so I did some amateur demography of my own. I found another Pew report, released almost at the same time, which focused on the beliefs of Millennials (those born from 1981 onwards). This report showed that less than 60 per cent of Millennials currently report a Christian religious affiliation, compared to around 70 per cent of X-ers (born 1965 onwards) and much higher levels for older cohorts.

That seems to me to cast some serious doubt on the projections of the first report. American adults likely to be alive in 2050 already show lower levels of affiliation than Pew is forecasting for that year. So, contrary to the claims of the report, the projections appear to assume that, on balance, Americans are more likely to change from unaffiliated to Christian than vice versa, or else that future cohorts will be more Christian than current young cohorts. A very brief look suggests that the same issue arises in the projections for other developed countries.

But as the Pew Millennial report says

It is possible that more Millennials who were raised unaffiliated will begin to identify with a religion as they get older, get married and have children, but previous Pew Research Center studies suggest that generational cohorts typically do not become more religiously affiliated as they get older. And the new survey finds that most generational cohorts actually are becoming less religiously affiliated as they age

This isn’t an area of professional expertise for me, or an issue in which I have a high enough stake to do more work than is needed for a blog post. Still, it seems to me to be a problem if the apparent contradiction I’ve pointed out is real. Apparently, neither Pew nor the peer reviewers at Demographic Research see such a problem. Am I missing something?



JakeB 07.30.15 at 7:21 am

Perhaps a side effect of the continuing overestimation of religiosity in the US owing to its association with virtue?

Given certain assumptions about longevity etc. it doesn’t seem impossible to assume a 66.4% level in the US given the current proportions.

But to be honest a projection this far into the future seems sort of stupid to me. If things go on much as they have I find it difficult to imagine that nonbelief won’t continue to infect both Anglo-American and western European culture as well as spreading elsewhere . . . and if they don’t, if things go all to hell as seems quite possible what with climate change and new technology-augmented totalitarian governments and so on, it would be surprising if the nonaffiliated don’t become an even smaller proportion of the total.


Commenter #? 07.30.15 at 7:26 am

There might also be some interested and knowledgeable commenters around Andrew Gelman’s blog for this. Statistical modelling, social science, surveys, etc…


dax 07.30.15 at 7:29 am

As someone who was born well prior to 1981, I object to your characterization of Millennials as those “adults likely to be alive in 2050.”


Z 07.30.15 at 7:44 am

People like to belong to stuff, so self-identification is problematic for measuring affiliation to political parties, unions or a gym clubs. More seriously, studies relying on self-identification for establishing trends have an obvious defect: the threshold to self-identify as affiliated to a religion lowers as the general religious sentiment wanes in the society, so the same level of religious belief leads to different answers diachronically. There will come a time when putting a nativity scene by the Christmas tree will be enough to self-identify as Christian. More accurate diachronic measures rely on objective indicators (number of people getting baptized, attendance at the weekly or exceptional mass, number of priests per person for Christians; praying five times a day or doing Ramadan for Muslims etc.). In the developed world, these indicators typically show a much steeper decline of the religious sentiment than the answer to the question “Do you belong to a religion?”.

I don’t know if the report you quote suffer from this effect though, in very briefly skimming it I couldn’t ascertain how they decide someone is affiliated to a religion. Independently of these questions, it is certainly true that one should expect a real surge of the religious sentiment in Africa with the rise of literacy, if African populations behave as those of the rest of the world (as I expect).


Brett 07.30.15 at 8:17 am

I wonder how reliable the muslim figures are, as well. The whole Middle East feels like a demographic bust waiting to happen in 10-20 years.


krippendorf 07.30.15 at 11:58 am

Not directly relevant to the demographic question, but Hout and Fischer have an interesting article in Sociological Science [open access, so no paywall] about the “rise of the nones” in the US over the last 25 years, from 7% to 20%.

Basic point is that the prevalence of the underlying religious beliefs (e.g., belief in the afterlife) has not changed, but that people who see themselves as liberal on social issues don’t particularly want to affiliate with organized religion. Or, as I read it, the political extremism of the Christian Right on social issues is chasing young people away from religious affiliation.

The lesson for the demographic question is really a caution: organized religion — the church, in lower case — is not a fixed entity, either, and can’t be modeled as such. Like organized political parties, churches can change their “platforms,” and in the process appeal to (or chase away) different subsets of the population.


Justin 07.30.15 at 12:45 pm

My guesses would be higher birth rates in religious populations is just as good a guess inside a country as it is between countries and immigrant populations are expected to be more religious than native populations.


bob mcmanus 07.30.15 at 1:02 pm

As the consequences of Global Warming and civilizational collapse and species die-off become obvious in the next few decades or sooner, I absolutely expect an increase in religious affiliation. I actually expect, among many other things, some new radical Millenarianisms to become even dominant.

I am somewhat casually observing current social behavior for signs of incipient pessimism and sub- and pre-conscious horror and guilt. Outcry over Cecil the lion, for example. The comfort of identity politics.


reason 07.30.15 at 1:11 pm

dax @3
are you disputing that you are on balance unlikely to be alive in 2050? If so, I think you have a problem with reality (or maybe you just believe that the holocaust is coming).


reason 07.30.15 at 1:17 pm

John Quiggin,
I just saw much increased population projections for Africa, that I doubt will happen. You don’t just need to look at birthrates, you also have to stop deathrates from eventually rising. Malthus hasn’t been repealed, he is just suspended. (I suspect what has mostly happened is that the AIDS epidemic predictions have been revised downwards, although they don’t say that.)


oldster 07.30.15 at 1:21 pm

Bob, surely you remember your theses on Feuerbach?

You can remain content with observing the global drift towards a new revealed religion with a charismatic leader.

Or you can step up and become that leader. Bobism. Blessed be his name.

Do you plan to be prophet, à la Mohammed? Or go for the full whack, à la Allah?


bianca steele 07.30.15 at 1:30 pm


I think it’s possible to read the OP as implying that people likely to be alive in 2050 are the same as Millenials, and that’s how Dax is reading it, though on another reading it appears that he might not be saying that.


Chris Grant 07.30.15 at 3:04 pm

According to Social Security Actuarial tables, the expected death date of (currently living) persons born in 1970 is after 2050.


kidneystones 07.30.15 at 4:30 pm

I was on a metro line today. There were two women sitting opposite. One, an older white women, was reading Dawkins on the end of faith. The other, a dark-skinned young, attractive women in headscarf and kit, glanced at the book and reader with a bemused expression on her face. Free country, and all that. Generalizing from the particular is always a mistake, but I know LOTS of highly superstitious people who claim to ‘have no faith.’

Meanwhile, President Trump – ahem. How could anyone possibly vote for such a clown. The Dems could run a corpse against Trump and win. That’s precisely the kind of thing I used to say to myself when Reagan made his play. An actor turned governor running for President? Twenty-mule team Borax. You’ve got to be kidding.


JimV 07.30.15 at 4:40 pm

My guess has already been stated above by Justin: that one factor is that religious Americans tend to have more children than non-religious Americans (and to indoctrinate them with religious summer camps, “Young Life” youth-groups, playing religious tapes during car travel, sending them to “bible” colleges, religious slogans in every room, e.g. “As for me and my house, we will serve the Lord”, etc.) . Anecdotally, in my family of five children, the most religious siblings have had four and four children, the less religious two and two, and the least religious zero. I think one reason for this is the notion that “God will provide” (child-care costs including college education) (and if God doesn’t provide, He had a different plan which didn’t include college education – perhaps the Army).


Witt 07.30.15 at 4:49 pm

Justin is on the right track.

That seems to me to cast some serious doubt on the projections of the first report. American adults likely to be alive in 2050 already show lower levels of affiliation than Pew is forecasting for that year. So, contrary to the claims of the report, the projections appear to assume that, on balance, Americans are more likely to change from unaffiliated to Christian than vice versa, or else that future cohorts will be more Christian than current young cohorts.

The key to the discrepancy is in the second sentence above. Any prediction of the US population in 2050 is really a prediction about immigration. Immigrants and their children account for a significant and increasing share of the population growth in the US, and they are more likely to be religiously affiliated (at least in the first generation).

So you can’t just assume that “Americans in 2050” will be today’s Millennials already living in the US and their descendents, you have to account for how much of 2050’s population will be related to immigration. And that is a very dicey proposition. Most projections assume a steady level of immigration* (that is, not dramatically changed from today’s flows), but that is a HUGE assumption and arguably unlikely.

You can pretty much stack the deck in terms of what conclusion you want to come to for the US’s demographic future based on your immigration assumptions. Up until the past few years we have been welcoming ~1.1 million legally authorized permanent immigrants and ~300-400,000 undocumented immigrants each year. If economic or demographic conditions in sending countries changed (as has already happened in Mexico, where we are now at net-zero unauthorized migration), or if the US rejiggers its immigration system, we could see profound swings in a) the number of new immigrants, or b) the demographic profile of new immigrants, either of which could (would) cause changes in the religiosity projections.

*Which given the increasing educational attainment->later marriage->lower birthrates among the US native-born population would inevitably lead to a more immigrant-heavy population stock.


guthrie 07.30.15 at 4:51 pm

Are there similar reports from say 20 or 30 years ago with projections? In which case can their accuracy be compared with how things have actually turned out?


Witt 07.30.15 at 4:54 pm

Excellent, detailed Census projections for the US population from 2014-60 in this pdf.

As you can see, they are projecting that immigrants will rise from 13% of the population to 19% of the population.


sillybill 07.30.15 at 4:59 pm

Oldster @11,

Sorry, that one has been already appropriated. Google ‘church of the Subgenius’, they already have a ‘deity’ named Bob. Bob Dobbs – High Epopt & Slackmaster.

You either have to come up with another name or get McManus to proclaim a schism and try to cast out the false prophet. Good luck with that tho – THEY have been trying to “Kill Bob!” for years.


oldster 07.30.15 at 5:21 pm

Okay, so Bobism is out.

Why not “McManicheanism”?

It has a nice ring to it, and you can also order it with fries.


reason 07.30.15 at 5:47 pm

Chris Grant @13
And what is the MEDIAN death date.


reason 07.30.15 at 5:53 pm

Chris Grant @13
But I will give you that part of the previous cohort will still have about a 50% chance of being alive.
I’m much older than that so it didn’t occur to me.


Z 07.30.15 at 6:16 pm

It has to be McManicheanism!


The Temporary Name 07.30.15 at 6:26 pm

Pew Research hasn’t even entertained the possibility that we are in the Last Days. It’ll be 100% once the sea turns to blood.


bob mcmanus 07.30.15 at 6:50 pm

Sorry guys, I have definite plans to check out in at most 5 years, if nature doesn’t do it for me. You are free to worship my corpse, if you send a check for the taxidermy in advance.

I expect left intellectuals to express their denial in black humor, which is at least more fun than the Dark Mountain sappiness.

The resource wars will redirect the mourning, and the liberals can do their usual blameshifting to Putin and Jeb.

We coulda woulda have saved the whole world easy if it wasn’t for those damn conservatives. Damn their souls. Not our fault”


Dick Muliken 07.30.15 at 7:45 pm

All polling on this question is enormously suspect. I recall reading studies at the Hartford Institute for Religions website indicating that something like 50% of respondents fudged their answers – and that the more religious you claimed to be the more likely you would fudge. I did a -very rough cut – study of my local rural county, comparing town populations and number of seats available in local active churches and found that only some 10% of the population could be seated. My guess is there even less seating capacity in urban settings


PlutoniumKun 07.30.15 at 8:32 pm

I think one problem with these studies is that peoples concept of religion is much more open than pollsters tend to assume. Unfortunately I can’t find links, but I remember a study on belief in Ireland which found that well in excess of 10% of self-identified Roman Catholics neither believed in God or heaven. This isn’t as strange as you might think when you consider that in many countries religion is as much a cultural identity as a belief system. I know several atheists who stopped self identifying as atheists in response to what they saw as extremist views associated with the Dawkins and Hitchens variety of atheism (they usually opt for ‘humanist’ or ‘agnostic’ instead). Purely anecdotally, it would seem that in American there has become less social pressure to identify with a church, which may be more apparent in polling than actual changes in belief.

Its also I think true to say that changes in religious belief does tend to be subject to very strong and unpredictable surges one way or another. I don’t think anyone in Ireland in the 1980’s would have predicted the sheer speed at which catholic church attendance collapsed in the next decade. Likewise, the manner in which European based secular muslims switched to more fundamentalist strains was not, I believed, predicted by anyone. I’ve several times heard secular Turkish, Syrian and Jordanians express astonishment at the devoutness of relatives raised in European countries. In parts of South America, there has been a surge in quite fundamentalist evangelicalism in traditionally quite easygoing Catholic communities. In China, there has been quite a surprising surge in Christianity among the educated middle classes (this does seem to be a feature of many Asian countries when they hit a certain stage of development).


Trader Joe 07.30.15 at 8:38 pm

Your 10% estimate is reasonably normal for a ‘seats to parishioners’ ratio. Can’t speak for other religions, but for US Catholics, when a new church is being considered the approximate sizing is based on 10% of the estimated size of the parish population – so if a parrish believes it has 2500 families averaging 4 per or 10,000 they’d look to build a church at +/- 1,000 seats. They’d then figure on either 2 or 3 masses to accomodate normal weekly attendance and double that for high holidays like Christmas and Easter….naturally any existing church might be oversized or undersized relative to a the current parrish population as its fairly rare that a church is fully decommissioned although they are sometimes repurposed.


John Quiggin 07.30.15 at 9:36 pm

Dax @3 (and others): I had in mind a weighted average of X-ers and Millennials, which would come out in the low 60s. There won’t be many Boomers around in 2050.


John Quiggin 07.31.15 at 12:00 am

I don’t think higher immigration works as an explanation. Suppose that the extra 6 per cent of the population who are foreign born is 100 per cent Christian, and the rest of the population is 60 per cent Christian (about the weighted average for current Millennials and longer-lived Xers, both native-born and immigrants). The extra immigrants only raise the average by 2.4 percentage points.

So, you still need a shift from unaffiliated to Christian *within* the Millennial/late X-er cohort or else a resurgence of Christian affilation among future cohorts.


Martin Bento 07.31.15 at 1:42 am

Are the new agers “religious”? Are people who believe in reincarnation, supernatural beings, astrology, crystals,telepathy, etc. “religious”? Arguably not; one could say they are just superstitious, and “new age” encompasses a huge spectrum anyway, but what they are not is materialist atheists like Dawkins wants. So a decline in religious affiliation may not mean a wider embrace of the materialist world view, and an increase in “atheism” in the West may just mean a decline is judeo-christian notions, not in a supernatural world view generally.


Witt 07.31.15 at 2:29 am

I’m not so sure about that. It’s not just immigrants, but also their children.

I just spent a while poking around looking for data on second-generation immigrant religiosity in the US, but I couldn’t find anything representative or reliable. This very comprehensive Pew report on second-generation Americans has very limited data on religion. And this atmospheric Huff Post op-ed is really just a slice of life.

I did find this rather helpful literature review/study on immigrant religiosity generally, but significantly the data set is sharply limited — only legal permanent residents (“green card” holders).

I also found a Pew blog post claiming that newer immigrants are more likely to be religiously unaffiliated, but I think there are significant problems with their data collection (namely, they only interviewed people in English and Spanish, and immigrants of non-Christian faiths in the US are disproportionately likely to avoid disclosing their religion).


Witt 07.31.15 at 2:30 am

Shorter me: Inconclusive!


bad Jim 07.31.15 at 5:52 am

World population ≠ US population. Americans are rapidly becoming more secular; the differences between the 2007 and 2014 Pew surveys are startling. Twice as many godless!

But that’s a change within a stable population. The countries with the highest birthrates tend to be religious, since religion and socioeconomic insecurity are tightly related: “the sigh of the oppressed creature”, usw.


ZM 07.31.15 at 6:26 am

“Sorry guys, I have definite plans to check out in at most 5 years, if nature doesn’t do it for me. You are free to worship my corpse, if you send a check for the taxidermy in advance.”

No! — bob mcmanus — No! In order to save you from this terrible fate of dying in a mere five years time from either natural causes or by your own hand — thus leaving the internet impoverished without your commentary — I will suggest a super national wartime mobilisation effort to solve climate change and other sustainability & resource problems! You must not spiral into despair!


bad Jim 07.31.15 at 8:25 am

My parents were agnostic, but I suffered a mild case of religion due to my exposure to my grandparents and school prayer. One of my brothers caught a worse case from his friends, and his children were infected by his late wife. They seem to have recovered.

Maybe I’m too fond of my science-loving family, but I’m inclined to doubt that many of those who’ve acquired the knack of critical thinking will be easily induced to abandon it.


Lee A. Arnold 07.31.15 at 10:21 am

If “religion” is defined as “organized religion with a theology”, then it will slowly lose ground to science, as it has done, in fits and starts, for the last 350 years. If “religion” is defined as a “reason for living”, then it will remain. If religion is defined as “change in being”, then it is in a new era of metamorphosis, not a cessation.

Big things which will have effects upon the near future: 1. Smartphone videos increase further awareness of violence and poor living conditions, and increase disgust with the status quo. 2. Social and emotional effects of medical treatments for disease and extension of lifespan. 3. Psychedelic drug use, which can have an enormous religious (in every way) impact upon individuals, has already returned to 1960’s levels (and use appears to be much higher among Millennials than Gen-X’ers in every developed country, although data are sparse). 4. Environmental concerns cause holistic emotions beyond scientific rationalism.


Lee A. Arnold 07.31.15 at 10:57 am

It’s possible therefore that the Pew trends to 2050 are likely to happen, due to Muslim demography, but that after that time, global religious affiliation will begin to recede due to increasing modernization. I think that this is another reason to watch trends among European Muslim youth in particular, because right now they are “wandering between two worlds, one dead / The other powerless to be born”.


ZM 07.31.15 at 12:10 pm

“Maybe I’m too fond of my science-loving family, but I’m inclined to doubt that many of those who’ve acquired the knack of critical thinking will be easily induced to abandon it.”

They are not necessarily opposites though — I just saw on Twitter that Peter Singer is asking Pope Francis to work together for animals


anon 07.31.15 at 7:48 pm

Large numbers of people in the Former Soviet Union, Former East Germany, and Former Peoples Republic of China (now Crony Capitalist) are seeking solace in Christianity.

Perhaps that accounts for those numbers.


ccc 08.02.15 at 11:37 am

@Quiggin: Do any of these works discuss the possibilities of radical longevity enhancement through biotech/biomed breakthroughs? I’d argue there is a fair chance that we’ll see treatments that will lengthen lifespan for 70 year olds in 2030 by another 20 years. And a fair chance that additional breakthroughs are made during those 20 years. And… Which sums up to non-zero chance that adults now living might just make it to escape velocity. That could be good in a lot of ways but would likely lower the pace of secularization as old habits die hard and the older people get the older their habits will be.


William Berry 08.03.15 at 12:04 am


Given that the coming* Great Human Die-back might be well underway by 2050, any increases in longevity among rich Westerners will probably be more than offset.

*Assuming that we, as a species, continue in the usual way of muddling along with our heads in the sand/ up our asses (appropriate metaphor according to taste).


Meredith 08.03.15 at 4:23 am

If some surveyor were to ask me about religion in a yes-no silly way, I would gag. Are there no scholars of religion on the CT list? No, there are not. Who even knows what the profile of “religion” will look like in 50 years? As an American, I see religion around me every second, not least in atheists (the people who can’t stop talking about god). We’ll see. (Or we won’t. We’ll be dead.)


Eli Rabett 08.04.15 at 4:07 pm

A speculation: In the last 50 years, the ability of humans (well at least some humans) to explain and manipulate our planet and particularly the biosphere which supports us has grown to the point where religion is irrelevant to that understanding as a practical matter. In the last 20 years the availability of that information has grown to the point where that knowledge is available to most. What remains for religion is ethics, but even ethics does not require religion.


Justin 08.05.15 at 1:47 pm

@John & Witt

Witt is right in that only looking at the share of foreign born population under-counts the effects of 2nd generation Americans. But I was also considering native religious populations in addition to immigrants. One Mormon or Jehovah’s Witness or southern Baptist family with six or seven kids (starting at 20) can offset a lot of people like me who are >30 with no kids. To really model this you’d need some data on birth rates vs. affiliation, age of parents at birth vs. affiliation, and conversion rates, but I haven’t found much.

Comments on this entry are closed.