The continuing decline of American religiosity

August 14, 2014 • 6:27 am

The Richard Dawkins Foundation website highlighted a post by Tobin Grant, a professor of political science at Southern Illinois University whose interest is the sociology of religion, and who writes about it at the site “Corner of Church and State” at the Religion News Service.

Grant’s post reports 61 years of measuring “religiosity” (the degree of religious belief) in the US, using statistics he developed in a 2008 paper (reference and free download below). In that paper, Grant combined 14 indices of religiosity into one, and developed a way to not only present that statistic in a way comparable among years, but to check its reliability. (You can read about the “validation” of his measure, the Aggregate Religiosity Index [ARI] in the paper at the bottom.

The components of the ARI are the indices below; the “correlation in the right column is the correlation of each component of the index with the ARI as a whole, showing how well each one individually could represent the whole:

Screen Shot 2014-08-14 at 7.33.00 AM

In his 2008 paper, Grant showed a general decline in religiosity between 1952 and 2005: while religiosity rose for a while in the mid 1950s, it declined (irregularly) until 2005. His original graph showed this:

Screen Shot 2014-08-14 at 7.35.42 AM

Grant has continued to monitor American religiosity yearly, and now has 8 more years of data. That data show that the decline of religiosity Grant perceived in 2008 is continuing, and at an increasing rate. Here’s the graph from his new website post. (The normalization to a mean of 100 has changed in a way he doesn’t describe, but the index itself and the data are the same as presented in the graph above—with the addition of 8 more years.)


Grant concludes that we’re in the midst of what he calls “The Great Decline” of American religiosity. And about time, too! His summary:

The graph of this index tells the story of the rise and fall of religious activity. During the post-war, baby-booming 1950s, there was a revival of religion. Indeed, some at the time considered it a third great awakening. Then came the societal changes of the 1960s, which included a questioning of religious institutions. The resulting decline in religion stopped by the end of the 1970s, when religiosity remained steady. Over the past fifteen years, however, religion has once again declined. But this decline is much sharper than the decline of 1960s and 1970s. Church attendance and prayer is less frequent. The number of people with no religion is growing. Fewer people say that religion is an important part of their lives.

In 2013, we saw continued declines in religiosity. The importance of religion in people’s lives? Down. Church attendance? Down. People who say they are “no religion”? Up. The result: 2013 had the lowest level of religiosity of any year we can measure.

Overall, the hundreds of survey measures point to the same drop in religion: If the 1950s were another Great Awakening, this is the Great Decline.

Now we all know of the famous increase in “nones” in the U.S. (people who profess no formal affiliation with a church); that category is in fact fastest-rising category of “belief” if you consider “no affiliation” a form of “belief” like Catholicism or Judaism.  Between 2007 and 2012, for instance, “nones” rose from 15% to 20% of the American population, an increase of 33% from their original value.

While that has distressed some religious people, many have tried to dismiss it by saying simply that the “nones” are still religious, but just can’t find a church that satisfied their spiritual needs.

Grant’s data, however, shows that this explanation doesn’t wash: the decline in religiosity as measured by the ARI, while partly reflecting a decline in church attendance, also reflects a decline in “feeling close to God” (the component of the ARI most correlated with its overall value), “frequency of prayer,” and assessment of the importance of religion as a guide to life. Those are all forms of religion one can practice without formal affiliation with a church.

So while the rise in “nones,” and the decline of ARI, might reflect a replacement of formal religion with informal religion (i.e., a belief in God but no church membership), it almost certainly reflects a general secularization of the country as a whole. This conclusion is supported by data Grant gave in a post on August 1, which shows that while formal church membership and attendance are indeed decreasing, so are people’s assessment of “religion’s importance in life” and “religion’s relevance for today.”  The latter two statistics reflect people’s general feelings about the value of religion, not whether or not they belong to a church. And all of it is dropping like a stone.

Professor Ceiling Cat predicted this, of course, but it’s also been predicted by many other sociologists and my pal Steve Pinker. As we become more embedded in the wider world, and connected with European countries that have become more secularized, we will undergo the same secularization as they. This is almost inevitable given the spread of Enlightenment values, of respect for science, and a realization that it’s time to put away our superstitions and belief in childish things. We no longer need religion in the U.S., either as an explanation for the great mysteries of life (many of which are being solved by science), or as a form of social glue to give us solace.  As social welfare and general empathy increase, religiosity (a divine substitute for what governments and friends can do for you) will also decline. The opium of the people will no longer be needed when you can get that “fix” from your government and community providing you with free health care, maternity and paternity leave, and help when you’re old or ill.

Although we won’t reach an ARI of zero in my lifetime, I’m confident that, unless the Caliphate takes over, the U.S. will someday become as nonreligious as Denmark, Sweden, and France.  Thank God for that!


Grant, J. T. 2008. Measuring aggregate religiosity in the United States, 1952-2005. Sociol. Spectrum 28:470-476.

87 thoughts on “The continuing decline of American religiosity

  1. Would love to see Mr. Pinker (or yourself) put violent crimes on the chart of ‘The Great Decline’. To see violent crime go down with religiosity would be fantastic! Morals from religion? Don’t make me laugh.

      1. And I would not be surprised if places where there ARE high levels of religiosity, there are also higher levels of violence.

        1. High religiosity is statistically linked to high violence worldwide. There is also a strong correlation with low per capita income, although there are two major outliers – USA and China.

  2. I often take comfort in knowing that eventually religion will exist only on the fringes. It is, however, a very slow process and in the interim religious fanaticism remains the single greatest danger in the world. I just wish more people would come to that realization sooner rather than later.

    1. One of the problem here is that many people think their religion is the solution to the problems cause by other people’s religions.

  3. I admire your optimism. But I’m sure it would just take one big war or another massive catastrophe for it to rise again. And I fear this century will be a century of massive catastrophes.

    Sorry, feeling particularly pessimistic today, just finished reading the Wired interview with Edward Snowden and all the regular distressing news.

    1. Of course the general decline can’t guarantee that a cataclysmic event won’t throw a wrench into the works and these events are often tricky to predict.

      I’m hoping that people become more reasonable, expect the same reason from their leaders, and we are able to negotiate around conflict.

      But of course, you never know and the world seems really nuts right now.

    2. Pessimism comes from only seeking negative data points. Any long-term trend will have short-term reversals, so you can’t draw any conclusions from them.

      From my view, the world becomes a better place with each passing year, and each generation better than the next.

    3. In the 1780’s the U.S. was created by an unprecedented group of Enlightenment figures. Today our government cannot seem to accomplish the simplest thing without copious references and deference to God.

      Also, modern secularism is a relatively very new condition in Europe; it’s hard to be convinced that it’s here to stay, and it seems to me that plenty of cracks are developing already.

      And climate change, overpopulation, and vanishing resources are hardly a recipe for peace and humanism.

      But then, Pinker’s a lot smarter than I am. (Re: Better Angels.)

      1. Yes, but Pinker also says that we can’t predict that the trend will continue and predicting influencing events is tricky stuff. Those things you mentioned could all get us in trouble.

        One thing I often wonder about is if you are in a period of long peace, could you produce more peace loving folk or at least more empathic folk? I often wonder if those that came during wars were doomed. People may have selectively mated with those with sociopathic tendencies because they offered survival & protection (I’ve had narcissists on my side & it’s great – until it isn’t). Further, people probably didn’t get the nurturing they needed emotionally and nutritionally….

        1. Thought-provoking as always, Diana.

          Regarding sociopathy–I was taught that variation tends to remain in a population in proportion to the frequency that it is adaptive. To go out on the un-evidenced limb that Jerry so loves (does WP have an eye-roll emoticon?), it’s always made sense to me that that would apply to sociopathy–too much and society would collapse, but at the right level, it comes in handy some times…

          (I know it sounds hopelessly facile in summary like that, but the only alternative is something thesis-length or worse…Your proposals about the effects of early conditioning on later actions are certainly testable, though, and already yielding predictive results. Perhaps you should take a mid-career switch to the study of human behavior.:))

          1. What’s so funny about your comment re: human behaviour study is that I am in midst of a career change and a few weeks ago, after 8 years with my company, was one of the last people, in a dreadful 3 year process, to have their position declared redundant. The company packages you out and they give you career support so you are partnered with a company that helps you redo your resume and support you though training (videos or Web dial in) about interviewing, becoming an entrepreneur, networking, etc. They also have assessments and mine listed life and social sciences at the top for careers that I am suited to.

            1. Diana, first of all, I am SO sorry that you’ve been going through this! I’ve watched this process (“right-sizing” when my husband’s company went through it) unfold before and it’s so cold-hearted and repugnant. (Have you read Ehrenreich’s Bright-Sided? She has some cogent commentary on the process.)

              (No wonder your physical maladies have been exerting themselves!)

              I’ll be eager to learn just what direction you do go in, and certainly wish you all the best in the process! Keep us apprised!

              1. Thanks! It’s really a bit of a relief as the 3 years of watching your friends being walked out one by one was pretty horrible. The good thing is, I received many notes telling me how shocked they were and that I was well respected. I had lunch with my former manager last week and she told me they aren’t even looking at people and they are getting rid of the best ones.

                I’m thinking of doing more writing. Breaking into the field is the trick and I still need to work on my writing centric resume (I just finished my regular business analyst/process one and I’ve had some interest in contract work there). I have a friend who is a professional writer who is helpful (though her expertise is financial writing which I find repugnant for no good reason other than my brain somehow shuts down when I have to deal with anything to do with finances or their systems).

              2. Well, here at WEIT you come across as a polymath, someone who would be good at whatever she sets her mind to. I have no doubt that (to the extent it’s possible in today’s economy*) you’ll land on your feet and rise to the top of whatever path you choose.

                (*Sorry for the proviso; at least in the Republican plutocracy that is the States these days, merit doesn’t always mean a lot.)

              3. Thanks! I’m not really a polymath, esp the math part. 😉 but it’s nice that you think so. 🙂

            2. Very sorry to hear this, Diana. I teach science (REAL science!) for an organization that is Christian and conservative, and I never know if or when the ax is going to fall. Good luck embarking on your writing career. To help cheer you up, a story from an episode of The Simpsons I saw the other day: Child Welfare has reason to believe that the Simpson kids are being neglected. They send two agents to the Simpson home to inspect. Among other horrors, they discover “toilet paper hung in improper overhand fashion”.

              1. Ha ha! Yes I did see that episode of the Simpsons and laughed pretty hard.

                The first paranoid thought I had when I got axed is “maybe it’s the atheism” since I don’t hide it on social media and if people ask, I tell them. The only reason I thought that was because my VP is from the south BUT he did always like me so I don’t think he is like that.

              2. Your VP is from South Canada? I’ve heard about those people. Living near the US border probably messes them up. When I get fired or leave my current job, you’ll know because I’ll start posting under my full name. Guess I’m paranoid too!

              3. LOL no the southern US. He doesn’t strike me as religious though. I don’t want to work for anyone that would want me in the atheist closet but I’m lucky in that Canadians tend to see religion/lack of religion as private so really the only thing they might be annoyed with is I write atheist things but I don’t do anything like that at work. I also hope to be around more science-y people where there are more atheists that won’t really care about religion.

  4. I would content that the rise of the Caliphate will only hasten the minimization of religiosity in America. Fundamentalism is no longer kosher and its inherent ‘extremeness’ is recognized as a translatable feature to anyone’s religious beliefs. This is making people think twice about adopting or maintaining any religious belief system.

    One of the major educational benefits that atheism brings today is that religion = religion; if you belong to one religion you need to park your belief system next to the other crazies.

    1. “I would content that the rise of the Caliphate will only hasten the minimization of religiosity in America.”

      My thought, too, although there will be percentage of Christians who have their beliefs reinforced by the Caliphate threat.

  5. That downward progression starting in 1992 (fig.4) is when I purchased a Genie account and discovered the internet. And that’s where I met my first atheist. I was incredulous.

    1. The protection that has been afforded to religion as the sole bastion of wisdom has taken a noticeable decline corresponding to the internet. Unless someone turns it off, the internet may very well be the ultimate death of religion. Thank you, Al Gore.

        1. Agree. There are just as many (no doubt many more) websites that pump religiosity as those that debunk it. “We” just don’t frequent those much.

          It may be becoming more socially acceptable to identify as a “not” but that doesn’t necessarily imply an actual change.

          1. We have the advantage, however. No matter how a kid is raised in America, he or she will be exposed to Christianity. The internet may be the first, or only, exposure a kid raised Christian (or Jewish, or Muslim, or…) will have to atheism.

              1. Unless they are particularly rebellious! However, I do know a kid who is deeply anti-authority except when it comes to religion. His family is part of a hyper right wing branch of Catholicism. Oddly, there is zero discipline at home and all the kids are a directionless mess. Go figure.

    2. I was recently listening to an interview with Hemant Mehta where he credits his childhood exposure to early atheist webpages (this would be around 1997, or so) to his deconversion from Jainism.

      It’s great that the internet makes it so much easier to break out of the bubble of religious information that so many of us are born into.

      1. Yep, I fondly remember reading The Internet Infidels or The Secular Web site back in, oh say, mid 1995. I am sure that is what Hemant started reading as well.

  6. To be “that guy”, can scientists please use proper Y-axes anchored at 0 (or whatever the lowest value is) when producing graphs? One can make the magnitude of a change appear however big one wants by truncating the scale. It’s just bad practice to create scientific graphs without the full scale represented.

    1. That would make an awkward graph without changing the facts. Proper form is to ‘use your space’, as I always tell my students.

      1. I agree. Truncated scales don’t change the facts, but allow the reader to evaluate the variances more accurately (or here: the small between-year ups & downs, relative to the overall trend).

        I too tell my students that truncated scales are ok.

      2. But visually, using a truncated scale does “change the facts”, or at least our perception of them, in that it over-emphasizes variance.

        1. I disagree. An unbiased reader/reviewer will not “over-emphasize” variance because both the visual impression of the treatment means/differences and the visual impression of the variances change; that is, the relationship between variance and treatment-differences does not change with truncated scales, but it is visually easier to assess that relationship.

          Of course, readers who pay attention ONLY to visual differences in treatment means are fooled by truncated scales, whereas readers should pay attention to p-values, and whether the right kind of stats-test was applied in the first place.

          1. the relationship between variance and treatment-differences does not change with truncated scales

            But it does if one is talking about an absolute scale. Consider if the data in Figure 2 weren’t between roughly 70 and 80, but instead between 100.070 and 100.080. I could provide the same graph, just changing the Y-axis, but certainly visually emphasizing the tiny variability in these data compared to their mean value would be grossly misleading.

            And I think that kind of presentation has real-world consequences. Imagine, for example, showing two graphs to two groups of people, one graph being the unaltered Fig. 2, and one with the full scale on the Y axis. I would be willing to bet a substantial sum of money that if you later asked each group to estimate the proportion decline in religiosity from highest to lowest, the group that saw the truncated axis version would provide a much larger estimate, and a far less accurate one.

            Graphs are a means to convey data in a visual fashion. If they don’t use the whole scale, then they are not actually presenting all the data, but are instead rhetorical devices.

            1. I completely agree, Tulse. Even worse are presentations in which Y-axis scales vary from graph to graph, relative to the highest and lowest values of any given plot.

              1. A graph may be technically correct, but intuitively wrong. If the general visual impression given by a graphic representation is confusing rather than quickly illuminating, it fails in its sole function. One may as well just present a table of data.

        2. The perception may be correct, but nearly every scientific plot makes use of the space available. Look at it this way, you get to see much higher resolution data. Would you rather see numbers in a list:




          1. Kevin, your example isn’t analogous. Instead the comparison is between:

            4.73, 4.72, 4.79, 4.76


            3, 2, 9, 6

            1. There are points to be made in favor of both truncated and non-truncated graphs under different circumstances.
              To me the key issue *here* is that the decline described by the lower graph is from about 80 to about 70, i.e. roughly 1/8. However, the graph makes it appear that the decline is 2/3. Sure, the scale sets you right, but you shouldn’t have to check the scale to see if the measure you are interested in (decline in religiosity) is visually represented accurately.
              Darrell Huff made this point in “How to Lie With Statistics” back in 1954 – and you can still get the book (reprinted as recently as 1993).

      3. My vote would be to “use the space”, but also use some sort of broken-line indicator on the y-axis to make it a bit more obvious that the graph doesn’t go to 0. (Also, on the first graph, it wouldn’t hurt to use bigger print on the y-axis numbers — I can’t actually read those numbers without leaning closer to the screen.)

    2. I would normally be exactly “that guy”, but not in this case. (I hate data plotted on axes in which I can’t tell what the scale means.) In this case however, the author notes,

      Because the index is a combination of different measures with different scales, the index produced by the algorithm does not have a specific scale. The numbers in the graph are roughly analogous to percentages but not exactly the same. The index’s numbers are like degrees Fahrenheit and Celsius, which are arbitrary marks on a thermometer that show the difference between heat levels but not absolute levels of heat show by Kelvins.

      …and he gives a link. The important plot is Figure 4, in which the standard deviation are shown and at least you can conclude at a glance that whatever “religiosity” is defined to be, at least we’re seeing a statistically significant shift over time. To learn more, the reader is going to have to decide to read more of the paper, to know what the aggregate index is.

      1. index produced by the algorithm does not have a specific scale

        But that means that we then can’t then make any claim about the magnitude of the religiosity decrease. Sure, it may be statistically significant, but if it is genuinely the case that “the index’s numbers are […] arbitrary marks”, then no claim at all can be made about the practical impact of the decline, merely that whatever magnitude of decline is statistically significant. Put more formally, we can specify it is statistically significant, but have no information at all about the effect size, which is what we really want to know.

  7. Theoretically ARI could dive to 0 (there are arxiv papers on that), but only if the social penalty of childishness et cetera gets high enough.

    In that context it is curious that the best correlation to ARI is sort of mistaking one’s consciousness for a magic agency (“feels close to God”). A mistake that sort of can be induced by electromagnetic stimulation (IIRC Dawkins has described his experience) and sort of shows up prominently in TMI when religious people are asked “what would your [magic agency] do” as they use the same brain parts as when they do other moral evaluation.

    As I remember it, it is about a year between realizing that Santa Claus is myth (at 5 years for me) and that religion is myth based. [Not that that means one can’t continue deliberate on religion.]

    Is “feels close to God” a very childish behavior? I think so.

  8. Does anyone know what country has “free healthcare”?
    I’d like to know how they swing it there. We could copy that model up here in Canada so we could apply the tax money allocated to pay for it to housing, education or day care.
    Something nice.

    1. I’d be pleased to pay the monthly cost of the your medical taxes if you’ll pay the $1556/mo that my employer and I pay for medical care for my wife and I here in the USA – and I’d be willing to trade the care each of us receives too.

  9. “Overall, the hundreds of survey measures point to the same drop in religion: If the 1950s were another Great Awakening, this is the Great Decline.”

    This quote taken from the study should have been worded to reflect that a rise in religiosity should be considered the Great Decline. Thinking and reasoning being the Great Awakening.

  10. How was this data taken? What are the demographics of the people represented by the data? The % of Hispanics is increasing, and I think think they tend to be Catholic.

  11. We need a longer time series to see what the real trend is! During the 1980s and ’90s I had the impression that a LOT of people had put a lot of decisions off (e.g. climate-change denial was entrenched) in case something were going to happen at the (2K) millennium. I found that disappointing and silly, but when I was a kid, some movie I watched in 1975 pushed my planning horizon out to 2011 (and I’m probably not unique). So, it’s not really surprising to see kinks in the curve around those dates.

    1. Ah, yes, Y2K. During that time I had a neighbor who was a retired engineer that was stocking barrels of drinking water and dried beans in his basement, preparing for the Y2K end-of-times.

  12. I wonder what the people who claim that “Christianity is not a religion, but a personal relationship with Jesus” answer to religiosity polls. Do they actually deny being religious, thus ending up with the nones? Surely they can’t deny they feel close to God.

    1. They might appear as anomalies like the “non religious” but “theist” category, where exists. Even _The Simpsons_ got that right – an early episode portrayed Homer as a nonreligious theist (though of a different sort from the “personal relationship with Jesus” folks).

  13. Hopefully the decline in religion will also be linked to more progressive legislation – further decriminalization of drugs and the drug war in general, reduction of the prison state, some restrictions on gun rights, single payer instead of Obamacare, closed loopholes and higher taxes for the excessively rich, more funding for science research and education, dedication to a non fossil fuel economy future and actual tackling of global warming etc., less demonizing of the poor, you get the point. A more reality based politics where everyone benefits.

  14. Religion is continuing to wither away… Nothing but a good thing, and flies in the face of those who say, “religion will always be with us,” and that humans need religion.

  15. From, I admit a fairly anecdotal perspective, over a 60+ year lifetime in New Zealand I suspect that a multiplier effect kicks in at some point, that is non-religious people become much more ready to admit the fact as social negativity decreases. Amongst my acquaintances many lifelong atheists would once have said they were Anglican or something despite never having been near a church in decades. At that time it seemed the normal thing to do – now the normal thing is to tick the no belief box

    1. Latest stats from the 2013 NZ census: 41.9% identified as no religion. Here in my hometown Wellington, it’s 44.2%.

      Almost halfway there!

      1. I think on reports that show the amount of atheists in various countries, NZ is one of the most atheist.

      2. In the UK, the census shows a rather higher degree of religiosity than the British Social Attitudes Survey does, ostensibly because of differences in the way the question is introduced and worded; in the census it’s more of a cultural background thing. Of course, HM Gov believes the census over the BSAS, so we’re in a pickle…


    2. I agree. Even in the UK, where the British Social Attitudes Survey shows that 50% are “nones”, there are folks who still seem to feel obliged to tick the “Christian” box.

      This was revealed by the RDF-sponsored Ipsos/MORI poll a year or so back.

      And also anecdotal evidence: When I went into hospital a while back, I overheard several older men (I was 49) respond to the admission-form question about religion with, “Church of England… I suppose.” None of them responded with certainty. (I just confused the nurse by saying, “Humanist”.)


      1. I’d hazard the hypothesis that many people don’t think of “none” as an option unless it is pointed out. Suppose you had a survey question that read “What colour is your hair?” I suspect that some bald folks would hem and haw over that one … (even if “none” was accepted). This is why some questionnaires are designed to *not* let freeform answers, so the “none” or “N/A” can be stated explicitly. Of course this is difficult with a religious affiliation question because there are so many denominations and listing them all is challenging.

  16. There is a superficial yet unsurprising correspondence between the ARI and the general economic situation – I overlayed BLS economic data: recessions and the rise and fall of the labor participation rate roughly follow the ups and downs of the ARI. Labor participation ticks up, religiosity ticks down and vice-versa; recession hits, a few more people get the faith. Idle hands and all that …

    The study may note the economics (tl;dr) – in addition I find the falloff from 1965 intriguing: did the Civil Rights Act deflate religiosity? Or was it the birth of SDS (did the Yippies engineer this)? It seems like the High Cold War of the late 50’s/early 60’s put In G-d We Trust on the money (five years after we got put “under” G-d in the Pledge) and buoyed against the trend … if the ARI graph is “right,” it’s interesting ground for speculation.

Leave a Reply