2013-08-19

The share of Americans who claim no particular religion doubled from 7% to 14% in the 1990s, as sociologists Michael Hout and Claude Fischer reported in an influential 2002 article based on the General Social Survey. A decade later, the Pew Research Center found that one-in-five U.S. adults (and fully a third of those ages 18-30) have no religious affiliation. Despite the rapid growth of the unaffiliated, Gallup editor Frank Newport cited survey data in a recent book to explain why he thinks “God is alive and well” in the United States. These findings raise many questions, including: What are the reasons for the rise of the religiously unaffiliated? Can organized religion thrive in the United States if growing numbers claim no religion? Is America, as a whole, becoming less religious or more religious? And how different, religiously, is the millennial generation from baby boomers and other recent generations?

On Aug. 8, 2013, the Pew Research Center brought together some of the leading experts in survey research on religion in the U.S. for a round-table discussion with journalists, scholars and other stakeholders on the rise of the religious “nones” and other important trends in American religion. The edited transcript is below.

Speakers:
Claude Fischer, Professor of Sociology, University of California, Berkeley
Michael Hout, Professor of Sociology, New York University
Frank Newport, Editor-in-Chief, Gallup
Greg Smith, Director of U.S. Religion Surveys, Pew Research Center, Religion & Public Life Project

Moderator:
Alan Cooperman, Deputy Director, Pew Research Center, Religion & Public Life Project

Audio and Slides:

You can find a recording  of the event as well as all the slides on Audio and Slides: Religion Trends in the U.S..

ALAN COOPERMAN, PEW RESEARCH CENTER: Thank you all very, very much for coming. I’m Alan Cooperman. I’m the deputy director of the Pew Research Center’s Religion & Public Life Project. As most of you know, the Pew Research Center is a nonpartisan organization – or a “fact tank” – that does not take positions on public policy issues. This event is part of an occasional series of round-table luncheons in which we bring together journalists, scholars and policy experts for timely discussions on topics at the intersection of religion and public affairs. The Pew Research Center’s Project on Religion [& Public Life] has been polling on religion now for about a decade, but we are the new kids on the block. The General Social Survey, conducted by the National Opinion Research Center [NORC] at the University of Chicago, has been asking religion-related questions since 1972, four decades now. And the Gallup organization has been polling on religion in America for even longer than that. So we are very honored today to have with us the leading experts on religion in these long-running surveys to talk with us about the important trends they see, and to discuss some of the big questions in American religion, such as whether the U.S. public is becoming more religious or less religious, and what might be the political causes or consequences of those trends?

Now, you have the speakers’ biographies in front of you in the packets, so I won’t go into great detail, but because scholars and journalists are sometimes less familiar with each other than they should be – and because one of the purposes of these luncheons is to bring the two communities together – I will say what everyone in the academy already knows, which is that professors Michael Hout of New York University and Claude Fischer of the University of California at Berkeley have had a very great impact on the field of the sociology of religion in the United States. One of their seminal papers way back in 2002 in the American Sociological Review noted a sharp rise in the percentage of Americans who do not identify in surveys with any particular religion. The size of that group, sometimes colloquially called the “nones” – N-O-N-E-S – had at that time doubled from about 7% to about 14% of the U.S. public, and in all of our surveys – in GSS, in Gallup and in Pew Research surveys – those numbers have continued since then to climb. Mike and Claude, in their paper, very persuasively, I think, showed that the increase in the size of this religiously unaffiliated population had both a generational component and a political component, and they offered the hypothesis that politics might actually be one of the factors driving the rise of the “nones.” Now, a few months ago, Mike told us in an email that he and Claude re-examined – using the GSS and more recent data – looked deeper at the causes of growth in the unaffiliated, and they have promised to share their latest thinking on that with us today.

Now, you’re all undoubtedly familiar with the Gallup poll and with its editor-in-chief, Frank Newport. In addition to overseeing all of Gallup’s surveys, which is an enormous endeavor, Frank has a particular interest in religious trends, as he showed in his book, which came out last December, “God Is Alive and Well: The Future of Religion in America.” Frank’s book took a different perspective on religious trends in America, that is, I think, in many ways a valuable counterpoint or corrective to the emphasis on the “nones,” and we look forward to hearing more about that. And, of course, we at the Pew Research Center also have some interesting data on religion in America, and I can’t think of a better person to discuss it than Greg Smith, our director of U.S. religion surveys.

So our format for this event is simple: Mike and Claude are going to speak for about 15 minutes on their research on religion using the General Social Survey, and then Frank will offer his insights on trends on religion in America in Gallup surveys, also for about 15 minutes. Then Greg will follow by briefly discussing the Pew Research Center’s findings on the rise of the religiously unaffiliated and other important trends. Greg has the home-court advantage, so he gets less time. Then we’ll invite the rest of you around the table to join us in the conversation. I should point out, this event is on the record and we are taping it. You are free to live tweet it, if you like. We’re taping this event. We expect to have a transcript on our website in a week or two. With that, I’ll turn it over to Mike and Claude.

CLAUDE FISCHER, UNIVERSITY OF CALIFORNIA, BERKELEY: [Thank you for the invitation to present on this topic. We much appreciate the important role that Pew Research has taken in disseminating social science research. My task here is to provide some background, some context, before Mike gives you the latest, hot data just off the grill.

In 2002, Mike and I published an article that did two things: One, it reported a notable increase in the percentage of respondents] in the GSS [laughter] who – shall I continue? OK – who when asked the question, “What is your religious preference? Is it Protestant, Catholic, Jewish, some other religion or no religion?” answered with the last option, “No religion.” That percentage had doubled from 7% to 14% in the 1990s. Two, the paper offered an explanation of the increase in what we called “nones” that focused on changes in politics, rather than changes in faith. We argue that a big part of the increase was liberals and moderates declaring no religious preference as a way of rejecting the growing connection between churches and conservative politics, especially conservative cultural politics on topics such as the family, women and sex. The “nones” were saying, in effect, if that is what religion means, count me out. This interpretation is consistent with findings that some people who are “nones” vary from year to year, but I want to reiterate an important point, that the finding has to do with the declaration in a survey of “I have no religion,” that’s very different than assuming something about people’s religious behavior or religious theology.

OK, let me put this trend – this increase from 7% to 14% and we will see on further – into context. Four comments here. One, Americans have long been, and still remain, the most religious people among the peoples of the Western nations, both in faith and in practice, especially compared to Western nations that are predominantly Protestant, as the U.S. is. Second, church membership and religious activity have waxed and waned and cycled over American history, but over the long run, over two centuries, both membership and activity have in net increased. Americans’ level of religious involvement peaked in the 1950s. Despite recent downward tendency, religious involvement is still higher than it was a century ago. Third, Americans’ relationship to their churches exemplifies the volunteerism that is central to our culture; that is, it is not that Americans are individualistic, they do believe in community, but it is a community that is voluntarily chosen and rechosen every day. When Americans are disappointed in their churches, they exit and usually look for another church. Recently, for many nonconservatives, it is organized religion as a whole that has disappointed them, and many of the disappointed now opt for labeling themselves as “spiritual, but not religious.” Finally, politics and religion have long mixed. Typically, Americans’ religious identities drive their politics, but often it is the other way around. We know that to be true in other countries, such as when the Irish embraced the Catholic Church as a way of expressing their opposition to England, and the Poles did the same with respect to Soviet rule. There are signs that now, in the United States, we are seeing political polarization starting to drive people’s religious identification. A question for future research is whether it will also end up driving faith itself. Now, I’ll turn it over to Mike.

MICHAEL HOUT, NEW YORK UNIVERSITY: Thank you, Claude. I’m also very happy to be here today, and the agenda is pretty simple. We’re going to update the trends through 2012 – the most recent GSS data that we have – and then to try to begin a discussion of what is behind the generational change, which we identified in the 2002 paper as accounting for about 40% of the overall trend, and we’ve now – We’re going to update that estimate to about 60%, actually, of the total trend – as that has to do with the fact that the so-called greatest generation is passing away and being replaced by millennials who have much less attachment to organized religion and organized anything else, as near as we can tell.

First, on the trend, the blue is what we had under observation in the 2002 data, that is to say they trend up through the year 2000. The little dots are the actual percentages in a survey; the vertical bar showed the 95% confidence interval around that; and what we see here is a trend line that we estimate using a kind of moving average called a “locally estimated regression.”

In contrast to that – The little – the line here with the dashed lines – skinny, tight – are the estimates that we published in 2002, and we probably over fit the data a bit – we ran the flat line too far and thus, put this line too steep. If you just make a projection of this upward, we should see about 27% of Americans had no religion in the 2012 survey and, in fact, what we observe is closer to 19%. So what the new estimates provide is, first of all, an earlier starting point and, most importantly, a much more gradual trend over time than we saw in the other one, and that gradual trend is one of the fingerprints of generational trend and is – generational change and generational succession – and is therefore part of the basis for our decision that – or our estimate that the generational trend accounts for about 60%, rather than our previous estimate of 40% of the secular trend over time. This is what we had in the way of a political gap to look at the – in the year 2000, that was in the 2002 paper.

Since then it has continued to widen. There’s still not much of a trend toward no religion among people who describe themselves as politically conservative, whereas for those who describe themselves as politically liberal, it is continuing upward and is in the neighborhood of 40% now.

The trick here – and I think everybody can understand that. All right. That’s the politics, and Claude articulated our argument on that pretty clearly just a minute ago. Without a Ph.D. in demography, the generational argument might be a little hard to fully understand. If generations account – generational succession completely accounted for the trend over time and the fraction of Americans who have no religious preference – now when we break out Americans by the year in which they were born, what we would see would be a bunch of flat lines, the ones where the oldest cohorts ending at some point, when there aren’t any longer enough of them who are – and they’re being replaced by people who initially are too young to talk in a survey of adults, but we start to hear from, and that’s – If cohort completely explained everything, these lines would all be flat. In fact, we see a lot – to a first approximation, these for older people are, in fact, flat. For the baby boom generation, we actually saw a movement toward religion as they married and started having kids. We all quit church in college, but then it came time to maybe baptize that kid and some of us made the move back, but there is, net on that, a slight upward movement after about 1986 for the baby boom cohorts showing a movement away from organized religion; we call that a period of vet.  It’s much stronger for the so-called Gen X’s born 1966 to 1975. This, we actually had very little data on this cohort, 1976 to ’85; if anything, it was moving away – toward religion, maybe echoing some of this life cycle stuff.

What we see now in the – oh, and that explained 40%. Flat lines would’ve explained 100%; because they dip and rise, it’s only 40%. What we see since is a lot of flatness for the baby boomers and the millennials, and a movement away again among the so-called millennials and this group that we have born since ’85 that we have very few observations on. We’ve simply plugged in their average to show that the average is higher, but we really can’t estimate a trend for them.

Meanwhile, we lose track of those born 1900 to 1915, and we’re starting to lose track of the 1916 to 1925 cohort, and so the new estimate is that this generational succession, the fact that these people are replacing these people in the population, accounts for 60% of the overall trend away from organized religion. What explains these cohort differences, they are – it’s just another clock. We’ve replaced the clock that runs in real time with one that tracks people’s birth rate. We really have – there’s no substance to birth – we give them all these names: greatest generation, millennial, and so on, but there isn’t much in the way of substance behind that.

Bob Putnam and Dave Campbell published “American Grace” a couple of years ago [in 2012] and they put forth the argument that there was a culture shock and two aftershocks; and we set about trying to test that in our current work. The shock was Woodstock and everything that came before and after – that’s the cover from the album – sex, drugs and questioned authority, and it turns out this looms much larger, actually, than these other two. The aftershocks – Moral Majority, traditional values, church-based, that’s the political push that Claude was talking about; and the second aftershock is the changing status of gays in American society, the trend toward embracing gay members of society and accepting it not just as a lifestyle but as an orientation and as an essence of being; and that’s the second aftershock. So what we’ve done is come up with a couple of survey instruments or survey questions that measure, first of all, the sex, drugs and attitudes toward authority as well as some of these – actually, this is in the other part of the paper that I’m not talking about today – but also attitudes toward gay sex; that’s in the GSS item. But we also invoke it over to Bob Putnam and Campbell, an academic literature that focused on the – I’ve got a cover of Dr. Spock’s 35-cent manual for young parents [laughter] – that so much emphasized treating children in a level conversation, emphasizing thinking for yourself, rather than “I’m the dad here. Listen to me,” and which undermines both authority and especially the teaching authority of religious leaders. So we then composed – oh, and by the way, there’s also the other potential explanatory factor, which might be secularization. I was just leafing through the Pew [Research “nones” report] book here, these trends of reproduced error from different data; so it’s always good when we echo one another. No trend at all in very traditional ideas about heaven, life after death, miracles, hell – hell is not as popular as heaven; there are some people who believe in heaven but not hell – but there’s no trend in any of these things.

Younger people do have doubts about whether God exists, although there’s not much of a trend toward atheism or agnosticism among them to the extent to which there’s any going on; it’s for people born since ’75, that baby boomers moved away from – well, in the earliest cohorts, half of people thought the Bible was literally the word of God. By – those born in the ’50s and ’60s, it was one-third, and there’s been no change since then. So we’ve combined all this information into a secularization index and put that in our statistical model as well. So our statistical test consists of – We control for standard demographic characteristics, putting dummy variables for each single year of the General Social Survey, and then a thing called the “cohort random effect” that I’m not going to tie you down with here; I could put you all to sleep right after lunch – but it’s a way of marking how big the cohort’s succession factor is. Then we refit that that model with cultural attitudes, the values measured having to do with whether people think it’s more important to parents to instill thinking for themselves in children rather than obedience in children, and our secularization index, and if these things account for the cohort difference, then a coefficient representing how spread out these cohorts are will trend toward zero and be replaced by these three factors. Also, they’ll get a regression weight, a coefficient that represents how important it is and we’ll – we can talk about that. Oh, yes, there’s one important point about it, that’s why I’m in this slide.

These are just – we’re just measuring differences between cohorts; we’re not entering a person’s actual answer to a question about “Do you think it’s more important for a kid to think for himself or obey her parents?” or “Do you think that gay sex is always right or always wrong?” but rather the cohort’s representation. So these are zeitgeist measures as a way of getting away from the prospect that having a religion is going to influence how people answer this question, but the cohort’s average is going to reflect the overall view of people of that generation. So when we run a statistical horse race, here is what we get. First, in the null model, the measure of the spread of the cohort random effects is substantial and statistically significant. When we enter our three explanatory factors, it’s still statistically significant, but only 20% is big. So we’ve accounted for about 80% of the cohort with our three variables – actually, with two of the three. It’s just the countercultural attitudes regarding sex and drugs, and the value on thinking for oneself, rather than obedience. The secularization, even though we have leaned heavily over and cherry-picked a couple of indices that are really strongly indicative of secularization, comes through with absolutely no effect; the standard error and the coefficient are the same size to a first approximation.

So we start here, we both confirm and extend the results, [laughter] shifting a 40% to a 60%, and our overall conclusion with respect to the cohort effects – and we don’t have time for politics today – is the baby boomer, later generations develop values and attitudes that undermine traditional authority, and secularization is left out of our story by the statistical horse race. With that, I’ll stop.

COOPERMAN: That’s very, very provocative. I’m sure there’ll be a lot of questions about it, but let’s hear from Frank, from Greg, and then we’ll open up the discussion.

FRANK NEWPORT, GALLUP: Thank you very much for having me here, Alan and everybody else involved here at Pew [Research], and it’s a pleasure to be on the program with such brilliant scholars in this area as Claude and Mike there. So let me just see if I can, in a few minutes here, add just a few viewpoints from a different perspective. It’s all very preliminary, but it does give us a way of looking at the situation involving the “nones” that come at it from a slightly different perspective.

Part of what I’m looking at here takes off from a comment that Claude made in his introduction there, which is to keep in mind that we’re talking about, in religious identification, one measure of religiosity, but it is not necessarily correlated with or the same as many other possible measures of religiosity. We’re talking about a self label in a question in a survey, which is, “What is your religious identity?” We have our question wording in this panel here – excuse me, in the paper. We’re talking about one question where the respondent is asked to choose among those labels to self-report for themselves to a survey interviewer, and although all the other questions of religiosity that we measure in a survey context are also self-reports, they can be different, because some of them were asking a respondent to self-report on past behavior, most notably, church attendance; others were asking respondents to access cognitive states like importance of religion; and then there is another series of measures that we saw on Mike’s chart there of belief in various aspects of religion, which are the same thing, asking people to assess what’s going on up in the brain. But when we asked religious identity, in some ways it’s different because we’re asking people to publicly put a label on themselves in a given arena, and I think that there – and that’s what I want to talk about here, is that part of what we may be seeing here is a change in the way that people choose to label themselves, rather than something which represents a more fundamental change in some of the other measures of religiosity that we can look at.

Now, I give as an example, LGBT status, which without resorting to data – this is a hypothetical – it’s certainly possible that the underlying percentage of Americans who on a variety of personal measures would be LGBT, the percentage who would self-identify as LGBT, over time could rise for no other reason than the social and normative environment in which people are asked to self-label could change. In other words, it’s much easier, we can presume, in this hypothetical situation, for a person to say, “Yes, I’m LGBT,” today than it might have been in the ’50s or ’60s, even though that same person in terms of personal behavior and lifestyle was exactly the same. So that’s kind of what I want to look at here, based on some data, and see if we can get some preliminary interest in and focus on some aspects of this labeling hypothesis.

I’m using a telescoped microcosm of data, which may or may not be legitimate. We have longstanding Gallup data, but we don’t ask questions – GSS, of course, has the tremendous virtue that it’s been held quite constant in many of the ways questions are asked over time, and although a lot of what Gallup has asked had been quite constant – like “Do you approve or disapprove of the performance of the president?” – a lot of religious measures have varied over the years, and they’re not always in the same survey and in the same context as other measures of religion, so it’s somewhat more difficult to go back in our data – two decades ago, ’50s, ’60s, and ’70s – and replicate a wide variety of questions asked in the same way that we might be asking today. But starting in January – Jan. 2, 2008 – Gallup began its daily tracking program. For a variety of reasons, we’re very fortunate to be able to do it and basically, we interview 1,000 people a day, which means we’ve now interviewed up through the end of last year more than 1.7 million people and luckily, we were able to embed in that survey, among a variety of other questions, several – not a huge battery, but several – questions on religion which we can then look at and analyze. So what I’ve chosen to do here is look at about 300,000+ interviews that we conducted in 2008, fast-forward to the exact same questions asked using the same methodology in 2012, a five-year period, and saying, Aha! If we can look at some changes over this five-year period, maybe we’ve provided some insights into what might be happening had we been able to go back to the ’50s or ’60s, and compare it to where we are now, which we’re not, because we didn’t have the same battery of questions back then.

So we’re looking at a microcosm of five years and asking: To what degree can we learn by understanding the changes over these five years? – or understand better what might have been happening in the long-term graph that we just saw from GSS, and of course, we see it in our own data. We did ask religious identification enough over the years that we’re able at Gallup to track the same thing. In fact, I found the survey in the ’50s where it was zero percent “nones.” How’s that? [Laughter] I mean literally, it rounded down to zero. So it’s amazing that back when the Gallup interviewer came a-calling – and it was in person in the ’50s – literally it looks like almost every single respondent chose a religious identification other than “none,” and there’ve been some changes in the question, but that was the bottom line.

So let’s look at the data here, or as Warner Wolf used to say, “Let’s go to the video,” [laughter]. Let’s look at the overall nature of the sample. Again, here’s the enormous sample sizes. So that’s good, we can actually – in a normal survey of 1,000, this is error noise in a lot of instances, but with larger samples, we think it represents something. The “nones” have gone up. The latest GSA [GSS] is closer to 19%. In 2012, we were at 18%, which is an interesting question in and of itself, Pew [Research] had reported quite famously 20%, all in the same ballpark. But it’s another interesting question about why they’re not exactly the same in these large sample surveys, which is a totally different issue, but nevertheless, they’re all in the same ballpark. But in 2008, we were at 15% and now we’ve had an increase of 3.2% in these very large samples over this five-year period of time, and, of course, it’s a zero-sum game, so we have the concomitant decrement in the people who gave a religious identity over the same period of time.

Now, what’s interesting is across these same five years, we did not see a concomitant increase in three other indicators embedded in exactly the same survey of the negative valence of three other indicators of religiosity. The number of people who said religion was not important in their daily life – actually, this is the percentage – excuse me, this is the percent “important,” it’s mislabeled; I mean to flip it, but it’s a zero-sum game, so it’s the same thing, and I’ll show you that in the chart in a moment – change very, very little here. [The chart below has been updated to reflect the correct percentages for “religion not important.”] The people who’d seldom attend religious services – I’ll show you that again in a moment – changed very little; and then a composite index we do at Gallup where we put the two together, we classify people in three groups: highly religious, moderately religious, unreligious – changed very little; and then we have the 3.2% change in “nones.”

So conclusion number one is that we had a – as these samples go, significantly larger change in “nones” than we did in these other three indicators. Same groups of people, same samples interviewed in exactly the same periods of time. And therefore we said, “All right. What’s going on?” and we broke out the sample – and I’ll do this for all three of those indicators – but we looked at the group of people who said religion was important in ’08 and ’12, then we looked at the group of people – This is not a panel survey – [it’s] two major cross-sectional surveys, keep in mind – the people said religion was not important, and lo and behold, what’s important here is that we had relatively little change in the self identification as a “none” among that group of people who said religion was important, which is 60-something percent of the population. All of the change, essentially, or most of it, came of those people who in both surveys said religion was not important, a change of 7.3%. And you do the division and take into account the weighting of the sample size, so that’s where you get the 3.2%.

So I’ll pause for a moment. I’m going to show you the other indicator, but this is probably the heart of the matter as far as what I’m talking about here today – is that the change came in self-identification is a “none” among people who in both surveys were already not very religious. So the suggestion is – the hypothesis is, if what we’re seeing here is a change in labeling, rather than a change in underlying religiosity – at least to some degree, that’s what seems to be contributing to it – phrased differently as I did a moment ago in my LGBT example, it can be hypothesized that these people for whom religion wasn’t important now feel freer to tell a survey interviewer that they don’t have a religious identity. Or there are other cultural forces at work which make it easier for them to say, “Yeah, I don’t have religious identity” in 2012 than they did in 2008, even though underneath it all, they were – religion was not important in both samples. In other words, if it had been a panel holding constant for the moment, test, retest and all that, we would have supposed exactly the same – people wouldn’t have changed in religion “important,” but they did change, in fact, in labeling. So that leads us to the conclusion that we are seeing the change in self-labeling on religious identification that is not consistent with a change in self-reports and other measures of religiosity.

Now, I’m showing you the same thing; this is a – just to show you what happened overall, there was very little change here, only 0.2% in the percent of the sample who said that religion was not important between the two samples.

So just again, this held constant where we had a 3.2% jump in “nones” over the same period of time. This is the same on our Gallup measure of religious service attendance; between the two samples, you can see this is the percent who were “nones” [in each group]. Very little of those people who said they attended services at least once a week, very few – which is an interesting study in and of itself – in either sample said that they were “nones.” But if you notice, the big change in the percent “nones” came among the “nevers.” In other words, these are people – a small percent of the sample, but these are the people – they are relatively small, not small and big in terms like those down there, but we had a seven-point jump in the percent in both samples who never attend church, who said that they had no religious identification.

So again, it’s a second piece of evidence along the same route that the first piece of evidence on religion’s importance led us to, and that is that we are seeing the change in this microcosm five-year period of time along people who are not religious in either sample. It’s just something about – and as I say, at the end of the paper – I’m not sure exactly what that is, and that’s room for additional exploration and probably taps into a lot of the other things that the other scholars here have been looking at. Something has changed over time which makes these people who never attended church to begin with feel freer to tell a survey interviewer, “Yeah, I don’t have a religious identification.” In other words, in ’08, 7% more of them said, “I’m Catholic, Protestant, Jewish, Baptist,” whatever it was, than was the case in 2012. Now, over this period of time, we had a slightly higher change in the religious attendance, but not much. This was fairly constant. The “nevers” went up slightly, but none of these changes in religious attendance approximated as we saw in that first chart, the 3.2% that we found in “nones.”

Just to reinforce the same thing, this is our Gallup grouping of people. It’s just a restatement of what you’ve seen because these categories are based on an algorithm with those two – not a fancy algorithm – but an algorithm of those two questions on religious importance and church attendance, and by grouping these people, we see the same thing among the people we classify as very religious. Basically, these are people for whom religion is personally important and who are frequent religious service attenders – very little change in the percent “nones” – the percent change “nones” as we would imagine – because you’ve just seen previous slides – came among those people who are unreligious.

So all of this compounds to the same basic conclusion, that we have seen the change – at least in these five years – in the already unreligious who are just changing the way they label themselves rather than changing the labels.

Now, there’s a lot of issues here that we can talk about, and just because I don’t go to church and say religion is not important doesn’t mean that I couldn’t have shifted even more to not going to church or religion was even less important in those two times, because these are not highly refined categories, they’re broad categories. But what we can say is that we did not see the change in these broad categories of religiosity over time. We did see the change in the “nones,” and the “nones” came for the most part among people who are unreligious both in ’08 and in ’12, but for some reason in ’12 felt – or just to say it behaviorally, for some reason in 2012, told survey interviewers – 300,000 of them – that they had no religious identity to the tune of about 7.5% more than they did back in 2008.

Now, because the data here, I wanted to show – this is very little change in these over time. I wanted to show you just a couple of other things here that may get at these – food for thought before my time is up and I get the hook here, OK? We wanted to look at the change in “nones” just in these five years – change in religious affiliation. Basically – and I think you all will agree if you’ve looked at it – when you look at the change in “nones” in the long-term, the Catholics have been relatively stable. Most of the “nones” have come out at the hide of Protestants, or what we call “Protestant and other Christians.” In other words, that’s been the change over time. Catholics, for a variety of reasons, including immigration of majority-Catholic Hispanics over time, have been able to hold their own, but in these five years, they actually fell by a point, Protestants fell 2.2%, and others, there were slight changes over time in our religious identification; Muslims actually went up slightly; and this, of course, is the 3.2% that we’re seeing. But I thought this was – and we’ll come back to that in a bit.

Here is something that our large sample sizes allow us to do, which I always think is fascinating – we can actually look at religiosity by each age point. George W. Bush is 67; we know that because he just had a stent put in, right, in Dallas. So we can actually find 67-year-olds and look how religious they are, and compare that to George W. Bush as an example.

Trends are the same; this is a well-established pattern of the relationship between age and religiosity that we’ve all seen before. It starts out – these are negative numbers going up. People are more religious when we capture them at age 18 or 19 because they’re still in the home, we think, but as you’ve just said, everybody goes to college and ceases to be religious. I say everybody – people who go to Brigham Young and others may not, but for the majority of college students, we see – religiosity going up, and it comes down in some measures; it’s been plateauing in the 40s and 50s, and then after the 60s, it goes down.

I talk a lot about it in my book. This group of people who are 65 and over are going to double in the next 20 years, or at least just a couple of years – already the leading age of the baby boomers, including George W. Bush, are already past that point. But what’s going to happen is these baby boomers move to this cohort, right? That’s a fascinating question: Will it be generational, so that they, too, will become more religious, or will it be a cohort where they don’t? Maybe you all can answer that question, but that’s a real key to the future of religion, is what’s going to happen? My point here is there’s a pretty high correlation between all these measures. So cross-sectionally, the “nones” are getting at the same thing as these other measures, as we look at it by age, which I think is an interesting pattern, although there are some differences. Look at the 2008 and 2012, by age percent of “nones.” Statistically, there’s a main effect for a year, right? In other words, we can see that for – in every little – right here in the early 50s, and I’m not sure what’s going on here, but statistically, there’s a main effect for years, so no matter what your age, you’d drop between 2012 – back 2008 and 2012. Everybody at every age point became less religious.

So the 87-year – well, that’s a little aberration there, but 80-year-olds that we interviewed in 2008, now when we interviewed 80-year-olds in 2012, they’re more likely to be “nones” than they were back in 2008. You can notice, if you’re perspicacious, which fits into your point here, there’s a little wider gap there, and so I put this together here. This is the change in “nones” by age point between 2008 and 2012, and it was highest among the young generation. So we have more of a shift here.

In other words, people of this age now have shifted higher than average in terms of “nones” compared to 2008 than other people. These are baby boomers when they’re about 47, 48 and on, and I’m not sure why they didn’t shift, but we actually have some shifting up here as well, which is a totally separate question, but we are seeing some more change in that younger cohort than we do find elsewhere.

Just a couple of final things; I found this fascinating – this is back to my Catholic idea; this is cross-sectional. This is only 2012. This is age across the bottom axis, and then we see, of course, the percent of “nones,” which is here, declining with age and that’s the standard pattern that we see with measures of religiosity, and of course, what’s happening is, as I’ve said, it’s coming out of the hides of Protestants, right? Young people are significantly less likely to identify as Protestant or other non-Catholic Christian, and then that goes up as we get to older people cross-sectionally here, exactly concomitant with the decrease in the “nones,” and the Catholics have stayed fairly stable.

One hypothesis here is that Catholics – it’s more of an ascribed characteristic, as I’ve talked about, for Catholics, than it is for Protestants. “Protestants” is an amalgam of literally hundreds, if not thousands, of different denominations. You mentioned religious switching; I wrote a paper on that once. “Protestant” is a weird conglomeration of religions in this country where, I think, people may get habituated with the idea that they can change, and they can shift and so it may be easier, the hypothesis would go, for somebody who is a Protestant to change to a “none” than a Catholic, which it may be a little more like an ascribed characteristic sociologically, where if you’re Catholic, you’re a Catholic going forward. May be or not be the case, because of the Hispanic immigration, as I’ve mentioned earlier, that accounts for a lot of the stability in the percent Catholic in this country over time, but at least in this cross-sectional sample in 2012, we can see that, generationally, it’s the “nones” and the Protestants that are kind of playing musical chairs as we go across the age spectrum. I think that’s an interesting finding as well, and it leads to some speculation for some other things we can see going forward. I have the same data in here in a couple other ways, in some scatter plots, but it’s basically showing the same thing.

The bottom line out of this is that in a microcosm of five years in our samples, at any rate, we have seen the percent “nones” go up, which we would expect based on the chart we’ve seen and projections and every other bit of data we have. It looks like the significant majority of that increase in “nones” has come among people who were not religious in ’08 and who are not religious in 2012 either, based on extrapolations from two cross-sectional, not panel, surveys. So we’re seeing an increase among people who aren’t religious, which leads to the hypothesis I discuss in the paper, and it’s only a hypothesis because there are many other angles to come at it. But what we are seeing here, at least in the last five years, is people who are not very religious to begin with but who were clinging to a religious label earlier, are now finding it – freer, if that’s the word, or finding that they are in a situation when an interview calls them up, it’s easier for them to say, “Yeah, I’m none, I’m no longer going to cling to the label I cling to from early socialization because basically I’m not religious so now I’ll make all my labels conform.” So that’s a hypothesis that we can talk about and how that might extend back to where we are going back in time.

COOPERMAN: Terrific. Greg, let me tee one or two quick things up for you. Mike, Claude, Frank and Pew [Research] data all show a rise in the [share of the] American public that doesn’t identify with any particular religion. We’re all getting close to 20% of the general public, but there are little differences in the numbers in these very large surveys. Before you get started, do you want to offer a conjecture as to why these large surveys come up with slightly different numbers?

GREG SMITH, PEW RESEARCH CENTER: Well, yes, this is conjecture, without having looked into it in any great depth. But when I see different surveys come out with different estimates of the share of the population that belongs to one religion or another, the number one thing I look to is question wording. I think that question wording – especially about religion – can make a big difference. I think that – In fact, you see this in our own data. We actually changed the way that we ask about religion here at the Pew Research Center back in 2007. Up until 2007, when we asked about people’s religious affiliation, we asked a close-ended question – we provided a number of options for people to choose, but we did not provide people with the explicit option to say they had no religion. They could volunteer that, and if they volunteered that they weren’t religious or had no religion or were atheist or agnostic, we recorded them as such, but the burden was on them to volunteer it. We did not list it among our options.

We were watching over the years all these same trends that we’ve just been discussing in the growth of the rise of the “nones” and we became convinced that not to provide people with that option would – to continue to not provide that option would be to miss in some ways what was clearly becoming a growing and increasingly important way that people identify themselves religiously. So in 2007, we changed the way we ask about religion, and we added the option for people to tell us what their religion is – we gave people the option to say they had no religion. The question we now ask is, “What is your present religion, if any? Are you Protestant, Catholic, Mormon, Orthodox – such as Greek or Russian Orthodox – Jewish, Muslim, Buddhist, Hindu, atheist, agnostic, something else, or nothing in particular?” And we take three of those categories – atheist, agnostic and nothing in particular – and that’s what we call the “nones.” I think it’s pretty similar to most other surveys. But when we made that switch in 2007 – and that’s all we changed was the way we asked the question – we saw an immediate uptick in the percentage of people identifying as “nones” by about three or four points. It went from about 12% before the switch to about 15% or 16% right after the switch. So that’s what we’ve experienced, and I think – that’s why I say – that’s the first thing I look to, to try to understand differences.

[The following exchange has been edited for clarity.]

COOPERMAN: OK, just to clarify, in Pew Research surveys, that last response option is “nothing in particular,” while in General Social Surveys, Mike and Claude, the last response option is “no religion,” have I got that right?

HOUT: Yes.

FISCHER: “No religious preference.”

COOPERMAN: Wait, the response option offered to respondents in the GSS is “no religious preference” or “no religion”?

HOUT: The stem question is “What’s your religious preference?” The response options are: “Would that be Catholic, Protestant, Jewish, something else or nothing at all, no religion?”

COOPERMAN: OK, so in the GSS, it’s “no religion,” in Pew Research surveys, it’s “nothing in particular.” Frank, what’s the wording of the Gallup question about religious affiliation?

NEWPORT: We do not have atheists or agnostics, but it is “no religion.” “Catholic, Mormon … another religion, or no religion.”

COOPERMAN: So, to summarize, is it possible that one reason why Pew Research gets a slightly higher number of “nones” is because people can choose “nothing in particular”? Meaning, is it a reasonable hypothesis that “nothing in particular” is a slightly easier thing for people to say about themselves than “no religion”? [Nods of assent from the panel.] All right, Greg.

SMITH: That’s right, and I do think that’s the important point – what is the trend that we’re [all] seeing, [not whether there are small differences in the numbers we get]. If the estimate [for the percentage of “nones” in the U.S. public] is 20% or 19%, or 17% or 18%, to me, those are all very much in the same ballpark, and the question is: Are they all headed in the same direction or not? We’ve seen that they are.

OK. So what I thought I’d try to do is, number one, echo Alan’s gratitude to our guests. It’s a real honor for me to be on this panel and to participate with such esteemed scholars of religion, and I’ve really been looking forward to this session; I’m looking forward to our discussion. I will try to keep my remarks brief. What I thought I would try to do is to try and reconcile what might be two countervailing pieces of evidence. On the one hand, we have seen that the “nones” have been growing, and we have seen that there is reason to expect their ranks to continue to grow. If the change is being driven largely by generational replacement, and young people are more likely to be “nones” than older people, then as they age into the population, the country should become less religiously affiliated. At the same time, there’s other evidence that would lead us to expect maybe there will be a religious resurgence. If people get more religious – If people become more religiously affiliated as they age, then as baby boomers get older, as the 76 million baby boomers in the United States age and become retired, if they become more religiously affiliated, then that might lead us to expect that religious affiliation will increase, that we’ll see a reversal of some of these trends. I don’t know which of those two things will happen, but I can, I think, try and offer a couple of things that we might keep in mind to help us anticipate the future.

There’s a couple of things I think we should keep in mind. Number one is that – and we’ve talked about this – the religiously unaffiliated, the “nones,” are really a complex group. They are not wholly secular, as we’ve seen, and I think that’s an important thing to remember. But I would also point out – and I think this is just [as] important – that they are also not “seekers,” they are also not simply religious in new and different and alternative ways, and I’ve seen people make that mistake as well. Neither of those things is true. They’re not simply secular, they’re not alternatively religious or seeking, and I think we should keep that in mind. The second thing that I think we should bear in mind is that even though it’s true that Americans become more religious as they age, in some ways, religious affiliation is not one of those ways. And regardless of whether or not disaffiliation from religion is itself an indicator of secularization, I do think it’s reasonable to expect that a continuation of the trend toward disaffiliation could have serious consequences for religious institutions and for the way that religion is practiced in the United States.

So let me just try to clarify what I mean by some of that. Number one, the religious “nones” are not uniformly secular, and to equate them with nonbelievers would be a real mistake. We’ve talked a little bit about how we define this group in our surveys. If you tell us you are atheist or agnostic, or that your religion is “nothing in particular,” we count you as a “none.” But you can see here that most “nones” are not atheists or agnostics. Instead the big majority are people who describe their religion as “nothing in particular.” They have not sworn off religious belief altogether; they’re simply not personally associated with any particular religious organization.

We can also see quickly that many people who are religiously unaffiliated tell us they pray. Four-in-ten “nones” overall tell us that they pray at least once a month, and more than half of that big group of people who say their religion is “nothing in particular,” more than half of them say that they pray once a month.

We see the same kind of thing when we ask people, “How important is religion to you and your life?” About one-third of the religiously unaffiliated say religion is at least somewhat important to them, and that includes four-in-ten among those whose religion is nothing in particular.

Now, obviously, on these measures, the “nones” are substantially less religious than the public as a whole; that’s also worth remembering. But nevertheless, you can see that it would be a mistake to assume that this group consists entirely of nonbelievers or people who are not religious. It’s also not clear, as we’ve discussed, that the growing ranks of the “nones” is an inherent indicator of secularization, and this chart makes much the same point that we’ve already seen. It shows trends in religious disaffiliation by frequency of religious attendance over time.

We use religious attendance as a rough but very valuable indicator of religious commitment, and what we can see is that just over the last five or six years, among people who tell us they seldom or never attend religious services, there’s been a sharp uptick in the ranks of the “nones,” from 38% in 2007, all the way up to 49% today. At the same time, among those people who attend religious services more frequently, the trend line has been very stable; there’s been no change in the percentage who described themselves as religious “nones.” Now, over the same period of time, there really hasn’t been much change in how often people say they attend religious services. The number of people who say they seldom or never attend really isn’t changing very much. It’s just that these people are becoming more likely to describe themselves as religiously unaffiliated. So this is just more evidence of what we’ve already seen.

So it would be a mistake to conflate the “nones” with nonbelievers, and it would even be a mistake to infer that the growing share of the population that’s unaffiliated is necessarily an indicator of secularization. But I think this is just as important – It would also be a mistake to conceive of the “nones” as a group of spiritual seekers or practitioners of new and alternative forms of religion. As I’ve mentioned, I point this out not because anybody in this room necessarily makes this assumption, but I have talked to people over the years who do seem to make that mistake. This, I thought, was a really interesting question we asked on a survey last year. We asked those people who described their religion as “nothing in particular” – We asked them directly, “Would you say you’re just looking for the religion that would be right for you, or are you not doing this?” And the overwhelming majority, nine-in-ten, tell us, “I’m not doing that.”

There are “nones” who say that they believe in yoga as a spiritual practice and not just as exercise. I thought that was an interesting question. There are “nones” who tell us they believe in reincarnation, that they believe in astrology, but it’s not a majority of them. And they do not express these kinds of beliefs at rates any higher than what we see among the general public.

We also find – I think this is one of [Religion & Public Life Project Director Luis Lugo’s] favorite questions – that the religiously unaffiliated population really doesn’t tell us that they have any great need to belong to a community of like-minded believers. We ask people, “How important is it to you to belong to a group of people who share similar beliefs and values?” About a quarter of religious “nones” told us that’s very important to them, and that’s far below what we see among other religious groups.

So they’re not uniformly secular, but I also want us to remember that this does not mean that this large group of people is out there just searching, just waiting, just looking for the religious organization or religious community that’s right for them. They’re not doing that either.

Let’s turn and think about what will the future hold for American religion? Will “nones” continue to grow as a share of the population? Will the United States become more religious, or will it become less religious, or will it stay about the same? I, by my nature, am a pretty cautious person, so I try to stay away from too much in the way of prediction. I think it’s much easier to look backward, and if we do that, as Frank pointed out, we can clearly see that people do tend to become more religious as they age in some important ways, and I think prayer is a good example of this. These data come from the General Social Survey, and it’s just a little different way of classifying some of the generational cohorts than what we saw earlier. What you can see here – and what it shows – is the percentage of people who say they pray every day among different generational cohorts and how that has changed over time. And what the chart shows is that as a group of people gets older, they do become more prayerful.

Back in the la

Show more