Field of Science

Is religion going to die out in 9 countries? Well that depends...

This week the BBC picked up on a mathematical analysis, first published back in January, which predicted that religion is on the road to oblivion in 9 Westernised countries (Australia, Austria, Canada, the Czech Republic, Finland, Ireland, the Netherlands, New Zealand, and Switzerland).

Unsurprisingly, the report went viral. But is the study any good? Well, you can read it for yourself, as the report (which hasn't been peer-reviewed), is posted on arXiv. Here's my take on it.

The first thing to remember is that, like any mathematical model, it's only as good as the assumptions that are built into it. These include the explicit assumptions, but also implicit assumptions that the modellers make.

In this study they began by assuming that religion is defined soley by self-reported identification with a religious group (i.e. Catholic, Protestant, etc - taken from census data), and that there are only two things that make religion attractive:
  • The more members a group (like a religious group) has the more attractive it is.
  • The more useful it is to potential members, the more attractive it is.
They also assume that religion is fixed and unchanging, that all religions are essentially the same, and that all people are the same.

They express this mathematically, and then tweak the parameters until the equation fits the data. Perhaps the most remarkable thing is that this model actually does fit the data, as shown in the figure.

What the graph shows is all the data from the 9 countries showing religious affiliation over time, rescaled and time shifted so they can be plotted on a single graph.

The only way they get the model to fit the data was to assign a value that made religion less useful than non-religion.

So, what this model suggests is that religious decline can be explained simply by the fact that people think that being non-religious is more useful - for whatever reason - than being religious.

That's a quite remarkable idea, although not unreasonable, if you think about it. What's more, if their assumptions hold, then religion will steadily decline and eventually disappear.

But will their assumptions hold?

Well, no. For starters, religion is not a fixed quantity. Every generation creates its own religion, and the religions of the future will have greater utility in the modern world. Then, too, people are different, so the picture is inevitably going to become patchy.

More importantly, I actually could imagine that religious affiliation might pretty much die away. We might see a day when almost no-one can be bothered with Churches and Mosques. But will that mean no more religion?

It depends how you define religion, of course. Many of the social and psychological traits that make up Western religion will still exist. They'll just be put together in different ways.

Will we call these new social constructions religion? Well, that's up to you!


Creative Commons License This article by Tom Rees was first published on Epiphenom. It is licensed under Creative Commons.

7 comments:

  1. That's a quite remarkable idea, although not unreasonable, if you think about it.

    Does that mean you have a suggestion for what is so useful about being non-religious?

    The more members a group (like a religious group) has the more attractive it is.

    So, if one were to model the frequencies of affiliation of different groups, should one put in a parameter that makes a conversion from one to the other more likely the larger to other group is?

    ReplyDelete
  2. Paper Title:
    "A mathematical model of social group competition with application to the growth of religious non-affiliation"

    That helps me understand some of the assumptions you speak of.


    Defining Religion
    I loved your last statement:

    "Many of the social and psychological traits that make up Western religion will still exist. They'll just be put together in different ways."

    And of course that holds for Eastern Religions also. Indeed, non-affiliate people and even atheists have strewn together all sort of social, psychological (and may I add) beliefs that could be called socio-psycho religions depending on your definitions.


    Ironical Funding of the Paper
    The James S. McDonnell Foundation
    Founder of McDonnell Aircraft Corporation --> McDonnell Douglas.
    Interestingly, from Wiki, "McDonnell was, by some accounts, a believer in the occult, as shown by giving many of his aircraft occult names such as phantom, demon, goblin, banshee, and voodoo."
    Also, he set up the "McDonnell Laboratory for Psychical Research" which the Amazing Randi approached to help them see through their biases but was at first rejected but later instituted exposing fake psychic powers. (of course) (source)

    Now, I don't know how this informs their choice of grant selections, but I thought it would be fun spice to add to the stew.

    His son writes of JS McDonnell:
    "While living in Chicago and working for Western Electric during the summer after his sophomore year, he spent every spare moment in the library reading the works of William James and Frederic W.H. Myers' book, Human Personality and It's Survival of Bodily Death. He became convinced that all mental and physical activity, including the so-called paranormal phenomena, had their basis in the performance of the underlying neuronal systems."

    Finally, this quote is interesting:

    "In closing, I believe what expressed Mr. Mac's philosophy and approach to life most succinctly and elegantly is a prayer he created and painstakingly improved throughout his lifetime.

    "Universal Creative Spirit – We thank you for the gift of conscious life on Earth with the opportunity to explore, create, develop, and grow in spirit and the opportunity to nurture all living things and take charge of the creative evolution of same. Hallelujah!" (source: Foundation website)

    So it seems he was an unaffiliated but religious in ways, eh?

    ReplyDelete
  3. "The only way they get the model to fit the data was to assign a value that made religion less useful than non-religion"

    This is a completely unresonable assumption (at least, to me). If you're a real beliver, could you thing of something more usefull than eternity in what your particoluar religion defines as heaven ? Of course, not. Utily may explains conversions. Utility cannot explains lack of belief. Not believing make your perceived life worse (although it can make your real life better). It is only when you're become more ready to accept reality for what it is, that you're ready to abandon religion.

    So I'm not thinking that religion will ever die out.

    ReplyDelete
  4. It is an interesting model, no matter how the end result would be, if the data is valid. One (other) caveat is that they put in effort to vary and test the model, but the test is the usual by eye only fit in preliminary research.

    So, what would be the problems with it?

    "For starters, religion is not a fixed quantity. Every generation creates its own religion, and the religions of the future will have greater utility in the modern world."

    This remind of the distinctions between absolute and relative fitness in evolution. Here they study relative utility, so this shouldn't be a problem for their model. (But for their current tests of it, for the reason you mention.)

    On a relative scale, the changes of religious and non-religious utility should change according to the environmental changes, while the "alleles" utility ratio is more conserved. If any religion would have swept the world, it should likely already happened.

    "Then too people are different, so the picture is inevitably going to become patchy."

    I would put this into the category of simplified away non-linearities that are bound to pop up when approaching the tails of the process (small amounts of non-religious or small amount of religious). The non-empty set of "unaffiliated" in historical time validates these effects.

    So yes, if unaffiliated was never extinct, don't expect affiliated to be. But they could approach that ~ 0 population of the unaffiliated tail.

    @ Anonymous:

    Your completely unreasonable assumption seems like a completely unreasonable assumption to me: the model works, so its assumptions have already been tested as "good (enough)".

    In reality you try to insert a theological analysis into an empirical discussion. That will never work, simply because it can't.

    As for that theology, it is well known that the actual utility of old and tired Pascal's wager is zip. The many religions make it so. It is then only a perceived utility that have to be embedded into the actual utility of the chosen religion itself.

    ReplyDelete
  5. Bjørn, things that are useful about being non-religious include having Sundays free to do other things, fewer restrictions on sexual conduct, better opportunities of divorce, freedom from a hierarchical, patriarchal organisation, and in general no need to adhere to old moral codes. Basically, utility should be interpreted in a general sense as whatever it is that makes people prefer non-religion (or vice-versa).

    With regard to your second, yes, I guess so. They only model two groups - the only effect of the group size thing is to introduce a delay in the inevitable domination of the highest utility group.

    ReplyDelete
  6. Torbjörn, I agree with all you write, with the caveat that religion, at some point in the future, could have greater utility than non-religion. A new religion could arise. More likely something or some collection of things that are very like religion but which are not 'official' religions could appear. They could lose some of the the baggage of 'religion' and so have greater utility, while keeping the useful bits.

    ReplyDelete
  7. Thanks for this blogpost--good to see that some corners of science media are willing to do a little critical analysis of this, rather than breathlessly accepting the paper's results. The idea that the evolution of the proportion of religious people can be explained by a utility function that depends on the current prevalance is interesting (to someone like me who knows a bit about statistics but nothing about academic study of religion), but there's no attempt to show their model is "better" than others in any sense other than simplicity, and the numerical predictions should be assumed worthless until at least one of them comes true.

    To put it another way--in all of the countries considered in this study, the percentage of non-heterosexuals has continually increased. One could then, I assert, easily fit a logistic curve (the religion paper's model) to the data, implying that heterosexuality was doomed to extinction. Now, no one would believe this, except maybe the BBC. The problem with the sexuality and religion models is that, beyond a certain point, there's no data at all. There is no region in the study in which the percentage of religious people is below 40%, so there's no way of knowing what will happen if the percentage of religious people falls below 40%--unless you have sound theoretical assumptions. As you've shown, one can object to the assumptions here without much effort.

    That the same model fits for all regions--if you let yourself do things like rescale time--doesn't show much other than non-religiousness is on the increase in those regions. Even that's partly a matter of selection--if data from, say, China existed and were included in the study, the end result would, at minimu, be a lot less tidy.

    ReplyDelete

Markup Key:
- <b>bold</b> = bold
- <i>italic</i> = italic
- <a href="http://www.fieldofscience.com/">FoS</a> = FoS