Secularization Vindicated

Abstract

In the 1960s, it was taken for granted that modernization eroded religion. In the 1980s, this consensus was challenged by the rational choice, supply-side, or market model proposed by Rodney Stark and associates. In particular, they argued that the UK was hardly less religious then than it had been in 1880. Clive Field’s compendium of statistical data allows us to test Stark’s approach to the religiosity of the UK. We follow this with data on Europe and the USA. While we may still argue over some of the precise levers, there is now so much evidence in favor of the secularization approach that we regard it as vindicated.

1. Introduction

Since Bryan Wilson published his seminal Religion in Secular Society in 1966, the popularity of religion in the West and the popularity of the secularization thesis have enjoyed an ironic relationship. As we will demonstrate, religion has, as Wilson predicted, declined steeply in power, popularity, and plausibility. Yet the explanation for this decline as a long-run consequence of the growth of individualism, egalitarianism, and democracy; of increasing social and cultural diversity; of a shift from religio-ethnic to civic nationalism; and of the displacement of religious beliefs and activities by scientific knowledge and technological problem-solving, has also become less popular.

Wilson, like most social scientists of his time, took it for granted that modernization weakened the demand for religion. His treatment was essentially functionalist and structuralist. Technically effective solutions to problems that people had previously addressed through religion gradually shrank its remit and thus reduced its social presence and importance (and with it the status of the clergy). The need for the modern state to accommodate a variety of religions (and the right of people to have none) gradually reduced the state’s support for any particular church. Where early nineteenth century states generally promoted a shared hegemonic religion, twentieth century states scaled back such support: first generalizing it to a number of competing churches, and then pretty much abandoning it. Where church–state links remained, they were generally ceremonial or they were dependent on the churches providing some useful secular function, such as social care.

If utility and social structural support were two strands of the secularization thesis, a third was derived from the European phenomenological approach promoted by Peter Berger, who stressed the disconfirming effect of pluralism. When an entire society shares the same beliefs, they acquire immense plausibility from being embedded in every sort of social activity: from personal rites of passage, to the marking of the agricultural seasons, to grand affairs of state. If, because of migration or the fragmentation of the dominant faith, the religious culture becomes variegated and fragmented, it loses its taken-for-granted-ness. The more that people are aware of alternatives (especially when they are carried by natives who cannot as easily as immigrants be stigmatized as low status), the harder it becomes to be dogmatic and doctrinaire. The result was a gradual shift towards a liberal ecumenism and then to a relativism that saw all religions as being equally plausible (and thus equally implausible). Dogmatism becomes particularly difficult if toleration results in inter-religious marriage. As David Voas and Alasdair Crockett demonstrated, the inter-generational family transmission rate for religion in the UK over the twentieth century was only 50 per cent. Unless a Christian family had four children, such a rate would not ensure replacement. The feedback effects of the drift away from religious affiliation, practice, and belief can be clearly seen in Voas and Crockett’s second finding: if only one parent is involved in religion, the transmission rate drops from a half to a quarter (Voas and Crockett 2005).

Roy Wallis (who had been taught by Wilson) and Steve Bruce summarized the secularization thesis as postulating that religion is undermined by modernity except where it finds important social roles additional to mediating between humankind and God (Wallis and Bruce 1991). They grouped such roles under the twin headings of cultural defense (where, for example, war between neighboring states of different religions gives an added premium to group loyalty) and cultural transition (where migrant groups rely on the émigré religious institution to provide facilities to ease their accommodation to the new world).

In the 1960s and 1970s, most criticisms of the secularization approach were technical and peripheral to the main case. David Martin, for example, argued that the religiosity of pre-modern European peoples had been exaggerated, that many secularizationists had an ideological commitment to secularism, and that the terms used in the explanation were too vague to be useful (Martin 1969). But then he produced one of the most nuanced descriptions and explanations of secularization, and, rather than take his own advice to abandon the term ‘secularization’, used it in the title of his A General Theory of Secularization (Martin 1978). Callum Brown argued that a sweeping sociological explanation of the decline overlooked much important local variation and that the roots of the decline lay not in the nineteenth century but in the 1960s, but he also deployed extensive statistical evidence to demonstrate the collapse of religion in the UK; his principal argument with Wilson and Wallis and Bruce was the date of the onset of this collapse (Brown 1987, 1992).

The truly radical challenge came from the work of a small number of US students of religion who took Alexis de Tocqueville’s comparison as their starting point. Tocqueville noted that France had a single hegemonic religion and it was unpopular; the American colonies had a wide variety of churches and religion was much more vibrant there (de Tocqueville [1835, 1840] 2000). Borrowing the basic principles of liberal economics, Rodney Stark argued precisely the opposite of the secularizationists. What needed to be explained was not a decline in demand for religion; as he and William S. Bainbridge had argued at length in their theory of religion, the demand for supernatural beliefs and solutions was stable because it met basic human needs (Stark and Bainbridge 1985, 1987). What varied was the ‘supply’ of religion. Far from being a threat to religion, a diverse religious economy provided the most fertile context for religion because it allowed everyone to find a faith that suited them. Competition forced religious providers to work hard to attract adherents and it reduced their costs. In an impressive number of papers, Stark, Roger Finke, and Laurence Iannaccone argued that, just as competition improved the quality of cars, reduced their price, and thus increased car ownership, so a competitive religious economy made a country more, not less, religious.

Stark’s alternative attracted some support in the USA, where the growth of conservative denominations and sects in the second half of the twentieth century was accompanied by the development of ‘the New Christian Right’ (NCR). Led by a number of skilled Republican activists, the NCR, from the Reagan era on, mobilized conservative Christians to support Republican candidates and causes, not on the classic agenda of a small state, aggressive foreign policy, and low taxes, but on socio-moral principles: in effect, a culture war (Hunter 1991). Scholars outside the USA were less impressed by the rational choice, supply-side, or market model of religion, in good part because it hinged on the notion that people’s religiosity was a matter of adult choice. The idea of a religious seeker, trying a variety of faiths in order to maximize his or her utility, perhaps fitted Californians. It made much less sense in most European countries, where people acquired their faith in the same way that they acquired their nationality and their language: by birth and childhood socialization.

2. Religion in the UK

We will return to the question of how well the Stark model actually fitted the US data. First, we consider one of Stark’s most audacious claims: that the UK in 1990 was not significantly less religious than it had been in 1851 (Stark and Iannaccone 1994, 1995). Thanks to the extraordinary work of Clive Field in collecting, assessing, and aggregating every possible source of statistical data on religion in the UK, we are now able to produce a far more accurate description of historical change than Stark et al. managed, and we can also see how well their assessment fits with what has happened over the subsequent thirty years (Field 2015, 2017, 2019, 2022).

Field divides his data into 21 ‘key performance indicators’ (KPIs). These are personal salience of religion, religious affiliation, religious membership, observed religious service attendance, claimed religious service attendance; baptism, weddings, and funerals; private prayer, the centrality of the Bible, the changing audiences for religious broadcasts, and the observance of Christian festivals; the effectiveness of religious socialization; belief in life forces, belief in an afterlife, and a mixture of ‘alternative beliefs’; attitudes to religion and attitudes to churches and clergy; and Sunday observance, the political consequences of religion, and religious prejudice. One of the virtues of Field’s work is that he explains in detail why certain KPIs are difficult to assess and summarize; for the sake of brevity, we will list only those for which there are reliable and consistent data.

In assessing the following data, it is important to note that the British population rose from around 40 million in 1900 to around 60 million in 2000. Hence, for any index of religious sentiment to remain on par, it would have had to increase by 50 per cent over the twentieth century.

Personal salience of religion: Because it is known only from surveys of variable quality, most of them non-recurrent, it is difficult to make much of data on personal self-description as religious or not, but the European Values Survey (EVS) shows that the proportion of Britons describing themselves as non-religious rose from 40 per cent in 1981 to 62 per cent in 2017. The recurrent British Social Attitudes survey (BSA) has the non-religious growing from 25 per cent in 1991 to 45 per cent in 2018 (Field 2022, pp. 49–50). Non-binary questions that added the category of ‘spiritual’ did not stem the decline. As Field notes, ‘the principal effect of including it in religiosity questions was merely to split the religious “vote” … without increasing the aggregate of religious and spiritual persons’ (Field 2022, p. 54). Attempts to add flesh to self-description as spiritual by counting acts predicated on New Age notions show contemporary spirituality falls very far short of compensating for the decline of conventional religion. According to the 2001 Scottish Social Attitudes survey, interest in serious divination was rare: 70 per cent of Scots had never tried any and only 2 per cent had tried some and thought it ‘very important in living my life’. Yoga and meditation were similarly unpopular: 78 per cent had never tried either and only 3 per cent thought them very important. The most popular realm was complementary medicine, but 55 per cent had never tried any and only 5 per cent thought it very important (Bruce and Glendinning 2003, pp. 86–115).

A detailed two-year study of the small English town of Kendal showed that only 1.6 per cent of the population engaged in what Paul Heelas and Linda Woodhead called ‘holistic spirituality’ activities in a typical week and half of those people denied that their yoga, meditation, Taizé singing, reiki healing, and the like were done for spiritual reasons. It seems reasonable to conclude that most of what from a distance looks like alternative forms of spirituality is actually a secular interest in health and wellbeing (Heelas and Woodhead 2004; Voas and Bruce 2007).

Claimed Religious Affiliation: This is difficult to assess for the same reasons as the previous KPI: the relevant surveys vary in quality and use different question wordings, but the BSA was consistent from 1983 to 2018 and it shows steady growth of those claiming no religious affiliation over that period from 32 per cent to 53 per cent. As the non-Christian category only rose from between 1 and 3 per cent in the 1980s to 8–10 per cent in the period 2017–2019, we can see the decline of Christian identification.

As measured by the census of the population, which uses a slightly different question, the number of self-identified Christians in England and Wales has been falling rapidly: from 72% in 2001 to 59% in 2011 and then 46% in 2021. Viewed across generations, the decline is even more dramatic. Nearly 80% of elderly people tick the Christian box on the census; for people in their 20s, the figure is only 30%. A majority of these young adults say that they have no religion, and we know from other studies that they are unlikely to acquire one in later life. Immigration and higher fertility took the Muslim share of the population from 3% to 6.5% during the period between 2001 and 2021, but this rise is obviously dwarfed by the Christian decline.

Religious membership: Two of the largest British Christian bodies—the Church of England and the Catholic Church—were late in recognizing a category of members, preferring to see themselves as serving entire communities. Nonetheless, the data are clear. In 1924, when the Church of England created ‘electoral rolls’, there were 3.5 million members; in 2019, there were only 900,000 (Field 2022, p. 72). If we take confirmation as a measure of the largest possible Catholic membership, we can note that there were 71,956 Catholic confirmations in 1970 in England and Wales and only 24,133 in 2019 (Field 2022, p. 358). The Methodists had 617,018 members in 1970 and only 164,024 in 2020. Peter Brierley estimates that overall Christian church membership in Britain declined from 7.7 million in 1970 to 3.9 million in 2020 (Field 2022, p. 327). So, over a period when the population grew by around 50 per cent, church membership halved.

Religious Service Attendance Observed and Claimed: Stark and Iannaccone try to inflate the impact of their membership figures by adding, ‘Moreover, the British may be far less inclined than are Americans and Canadians to actually see to it that they are signed up as church members, since a far larger percentage of the British population claims to attend church with some frequency than are counted on church rolls’ (Stark and Iannaccone 1994, p. 243). There is, of course, an alternative interpretation: that membership figures are reasonably accurate and that, in responding to surveys, people exaggerate their church attendance. The recurrent British Social Attitudes survey (which has the advantage over discrete surveys of repeatedly using the same methods) shows that claimed once-a-month-or-more attendance declined from 21.3 per cent in 1983 to 17.0 per cent in 2012 (Clements 2014).

But we need not speculate as we do have a number of counts of actual attendance. The 1851 Census of Religious Worship attached to the normal British population census a series of questions about church attendance that were sent to every known place of worship, and census officials worked hard to ensure completed returns. Because it asked for reports of numbers at each service, the census cannot tell us attender numbers but we can be sure of the upper end (if no one attended twice) and lower end (if those who attended any outlet that had more than one service did so twice) of the range, which can be summarized as 60 and 40 per cent. For convenience, we will say that 50 per cent of the population attended church on Sunday 30 March 1851.

A century later, the Scottish figure was 26 per cent of adults. More recent figures (from Peter Brierley’s censuses) show that the 1980 figure was 17 per cent and that for 2016 was 7 per cent (Field 2022, pp. 99, 336). The Church of England’s own data show a fall in ‘usual Sunday attendance’ from 1,541,828 in 1970 to 690,000 in 2019; that is, it was halved.

Baptism: Field aggregates the data from all the major churches to conclude that the baptism rate in Britain in the 1960s was around 85 per cent of live births; in 2018, it was no more than 25 per cent (Field 2022, p. 111).

Weddings: Despite there being good reasons why people who are not particularly religious might want a scenic wedding in an ancient church, religious weddings showed a similar decline. In 1838, almost all weddings in England and Wales were solemnized in places of worship. In 2017, the figure was 23 per cent. Scotland showed an even greater decline: the proportion of marriages solemnized by the Church of Scotland fell from 47.8 per cent in 1970 to 8.6 per cent in 2019, and the Catholic Church—the second-largest denomination—saw its rate fall from 16.4 per cent to 3.5 per cent (Field 2022, p. 115). Both were overtaken by Humanist ceremonies.

Funerals: In 1900, almost all funerals were religious events. In 2018, only 41.7 per cent had a religious officiant and most of those took place in a crematorium rather than a church. The changing nature of such events can be described with data from Co-op Funeralcare, by far and away the UK’s largest provider, with over 100,000 funerals a year. In 2011, it reported that only two thirds of its funerals followed the rites of a particular religion. Humanist celebrations counted for 12 per cent and 21 per cent of its funerals were ‘contemporary’: a personalized celebration of the life of the deceased. Only a third of funerals now have only religious music, the remainder using contemporary or classical music or a mixture of both (Bruce 2020). The top three funeral songs in 2009, according to a separate Co-op Funeralcare study, were My Way (Frank Sinatra or Shirley Bassey), Wind Beneath My Wings (Bette Midler or Celine Dion), and Time to Say Goodbye (Sarah Brightman or Andrea Bocelli).

Private prayer: With the usual caveats about inconsistencies in question wording, Field reports that, in 1950, 48 per cent of a sample survey claimed to pray regularly. In 2013, no more than 20 per cent claimed to have prayed in the previous month (Field 2022, p. 126).

Belief in God: Stark and Iannaccone report 74 per cent of Britons claiming to believe in God and only 4 per cent saying that they are convinced atheists (1994). It would be generous to call this a deviant reading of the data. The first poll known to ask about God offered a binary response: in 1961, 91 per cent of Britons said that they believed in God. By 1987, believers were down to 69 per cent. The BSA’s recurrent surveys (which offered six nuanced responses) showed that belief in God—including only sometimes—had fallen from 61 per cent in 1991 to 43 per cent in 2018 (Field 2022, p. 403). Non-belief had grown over the same period from 10 per cent to 26 per cent and the agnostic response from 14 to 18 per cent; that is, almost half of respondents were now non-theists.

Attitudes to Religion: Most commentators would have described British attitudes to religion in the last quarter of the twentieth century as benign indifference: religion was fine—so long as it does not interfere with me—because it taught children morality and gave comfort to the bereaved. Hostility has increased markedly since. A ComRes poll in 2006 asked if people thought religion was a force for good: 53 per cent agreed. The same question in 2017 showed only 37 per cent in accord (Field 2022, p. 423).

Attitudes to Clergy: As it reflects either the popularity or the power of religion, the number of clergy (which can be measured with considerable accuracy) is an important index. In 1900, there were some 45,400 clerics in the UK; in 2000, there were only 34,160. In 1900, the Church of England had 21,000 full-time paid parochial clergy. In 2000, there were 9538 and a fifth of them were unpaid part-timers (Brierley 2001, Table 2.7). In 1900, Scotland had around 3600 Presbyterian clergy; in 2000, it had around 900 (Brierley 1985, p. 55).

Status is harder to measure but there is no doubt that it has declined severely. There were poorly paid clergy in the nineteenth century—especially in working-class areas of cities—but for the most part the clergy were of sufficient status and sufficiently well-rewarded for the church to offer a respectable profession for the younger sons of the minor aristocracy and the gentry. Even in the 1930s, the vicar of the very ordinary parish of Gosforth, Cumbria, lived in a very large house, kept servants, and possessed a stable of horses. In 2001, the average Anglican clergy salary was around GBP 17,000, which, even with free accommodation, was well below the national average for graduates (ThisisMoney 2013). An Office of National Statistics listing of 400 occupations by salary has the clergy at 293, its GBP 21,485 far below that of ‘health practitioners’ at GBP 54,684. Since 1983, Ipsos MORI has tracked public trust in various professions (Ipsos MORI 2013). In 1983, 83 per cent of the public trusted the clergy to tell the truth. In 2013, the figure was 66 per cent, well behind doctors (89), teachers (86), and judges (82 per cent).

The Power of Religion

Overlooking their failure to recruit their own children, religious people often complain that their faith has been ‘marginalized’, the active verb implying deliberate action by secularists. Actually, religion in Britain is still granted a degree of respect by governments and public institutions that is far above what its current support merits, yet, despite this, every change points to religion’s declining influence.

The secularization of education, for example, has continued apace. Even in nominally church schools, dogmatic religious education has been replaced by education about a variety of religions, all presented as being equally valid. The ancient practice of calling for divine blessings on institutions (such as opening council meetings with prayer) has either been neutered (by inviting representatives of all faiths to lead such prayers) or discontinued. Laws making blasphemy a criminal offence were repealed in England and Wales in 2008. Such protections of the sabbath as bans on Sunday trading have been removed.

On almost every contentious issue where the Christian churches have had a distinctive position, they have lost the argument. Shortly after Religion in Secular Society was published, homosexual acts between consenting adults aged 21 and over were legalized; since then, there has been a steady increase in the public acceptance of gay rights. Abortion was legalized. Obstacles to divorce were gradually reduced until such arguments were made irrelevant by the willingness of people to cohabit and to raise children outside the bounds of matrimony. When Wilson wrote, it was still common to refer to such arrangements as ‘living in sin’. Now, they are entirely commonplace and the Christian preference for monogamous life-long heterosexual marriage no longer hinders such arrangements as the allocation of welfare rights and spousal benefits.

Instead of continuing the list, we can just note the significance of the 2010 Equality Act, which made it illegal to discriminate in the provision of goods and services against people on the grounds of age, sex, race, disability, marriage, pregnancy, sexual orientation, gender reassignment, or religion/belief. Although religion is a protected status, it is only one of nine, and subsequent legal testing has made explicit what was implicit in the legislation: to borrow the card game analogy, religion is no longer trumps and exemption from the law on religious grounds can be claimed only on the narrowest of matters. Religious organizations are not required to change their views, but they cannot impose them on others or act in ways that simply express their disapproval of particular classes of people. So, churches can, for example, exclude homosexuals from their clergy; they cannot exclude them from such non-clerical positions within their organizations as administrator or secretary.

Field concludes his magisterial compendium of statistical data as follows:

In sum, the empirical evidence reviewed in this book offers, within its defined parameters, strong confirmation that secularization remains valid as the dominant narrative and direction of travel in Britain during the half-century from 1970. The process may be incomplete, but it has now taken such deep root, including in the formerly extremely religious nations of Wales and Scotland, that there seems little prospect of it being reversed.

(Field 2022, pp. 296–97)

Now, we can return to Stark’s supply-side proposition that religious pluralism is associated with religious growth rather than decline. Far from the national church monopoly that Stark and Iannaccone suppose, Britain already had considerable diversity in 1851. For example, of the 132 congregations listed in the census returns for Leeds, the established Church of England had only 31: fewer than a quarter. There were also Independents (11 congregations), Particular Baptists (9), Scotch Baptists (1), General Baptists (2), ‘other’ Baptists (1), Quakers (1), Unitarians (3), Wesleyan Methodists (26), New Connexion Methodists (13), Primitive Methodists (13), Wesleyan Association (10), New Church (1), Brethren (2), Roman Catholics (2), Mormons (2), and six ‘isolated’ congregations. Nor was such variety a feature only of cities: in the county of Cheshire, for example, only 31 per cent of places of worship were Church of England. Since then, the religious marketplace has become even more diverse with the addition of nineteenth century innovations (such as Christian Science, Spiritualism, and the Jehovah’s Witnesses) and twentieth century novelties (such as white British Pentecostalism, independent charismatic churches, and Black Pentecostal churches). Few people joined them but all the new religious movements of the 1960s were available in Britain. Britain also now has significant numbers of Jews, Hindus, Buddhists, Muslims, and Sikhs. In short, Britain has become more religiously diverse but all the measures of religious interest and involvement have declined drastically.

3. The Rest of Europe

Europe has long been a key test-bed for theories of religious change. Emile Durkheim (1858–1917) and Max Weber (1864–1920) put the impact of modernization on religion at the heart of the new discipline of sociology. The proponents of the market model argue that the quasi-monopolistic character of many European state churches supports their supply-side theory (Stark and Iannaccone 1994). Other scholars have offered alternative accounts (Davie 2000, 2002). Two issues are particularly important in the debate: is there a single story that applies to most or all European societies, and how should we view the many people who are neither committed churchgoers nor overtly unreligious?

The unevenness of the European religious landscape is obvious. Catholic, Protestant, and Orthodox churches dominate in some places and are nearly absent in others. Religious participation started to decline more than a century ago in some countries and is only just beginning in others. The various dimensions of religiosity have changed at different rates in different places: belief and practice have dropped to low levels in Scandinavia, for example, but affiliation remains high. Opponents of the secularization thesis have pointed to this diversity in levels and trends as evidence that a different paradigm is needed.

Iannaccone (2003) argues that trends in religious involvement are too varied to be explained by standard models of secularization. Writing in the same year, Greeley was characteristically combative: ‘A single, one-directional model does not begin to cope with the variety of religious phenomena in Europe. … “secularization” … is patently a useless theory because it says too much and hence fails to subsume a wide variety of interesting data’ (Greeley 2003, p. xi).

An implication of this argument is that the secularization paradigm would seem all the more impressive if it was able to account for such diversity of levels and trends. If the explanatory framework can accommodate dozens of countries with different languages, histories, and cultures, it would pass an important test. Rather than being fatalistic about the difficulty of generalizing, social scientists should aim to understand why so many separate societies seem to be experiencing religious decline. The alternative is to devise a whole set of distinct national stories to explain a common phenomenon.

Voas (2009) took up this challenge. He looked at five-year birth cohorts covering much of the 20th century for all the countries included in the first round of the European Social Survey. Religiosity (measured using a scale based on current or past affiliation, attendance at worship services, private prayer, self-assessed religiosity, and the subjective importance of religion) declines from one cohort to the next in all of the countries apart from Israel. And although these countries ranged from very religious (Greece) to very unreligious (Czech Republic), the rate of decline was remarkably constant.

As with the rise of the NCR in the USA, the promotion of a shared religious identity by some authoritarian conservative governments (and the restoration of the hegemonic churches in post-communist states) has distracted scholars from the steady decline of personal involvement in organized religion. Voas (2009) found that the drop in religiosity over two generations—between people born in the late 1920s and the early 1980s—was essentially the same over the score of countries covered by the survey. In consequence, the relative positions of these nations in a European league table of religiosity was maintained over the course of the 20th century. There was a small amount of convergence; the decline was slightly higher in countries that were more religious at the outset.

To put the matter differently, differences between countries relate mainly to the timing of the onset of secularization. The generation gaps within countries are very similar across Europe. What varies is not so much the pattern of religious change as the progress of the various social forces that undermine attachment to tradition. Across the continent, people born early in the 20th century are relatively religious, and those born late in the century are relatively unreligious.

This fact has tempted many observers to suppose that religiosity increases with age. If this is so, then churches have no cause for alarm: young people might appear to be unreligious, but it is just a stage they are going through. Overall religious involvement will not decline, because people will return to church when they start families or recognize their own mortality or experience some other effects of the life course.

The evidence points in the other direction, however. We now have data going back half a century from Eurobarometer, a few decades from the European Values Study and the International Social Survey Program, and a couple of decades from the European Social Survey, to say nothing of the many high-quality national surveys (including ALLBUS in Germany). Generation gaps are persistent. Religious decline is real and continuing. Several recent books offer good overviews of the evidence; Pollack and Rosta (2018) are particularly thorough.

Interesting discussions are possible about the contributions of period and cohort in the process, but it is clear that age plays a negligible role. Multiple studies using both repeated cross-sectional surveys and panel surveys support the hypothesis that religious affiliation, practice, and belief typically stabilize when people reach their mid-20s. Some individuals become more or less religious in later life, but, within each birth cohort, these gains and losses tend to balance out (Te Grotenhuis et al. 1997; Voas and Crockett 2005; Crockett and Voas 2006; Voas and Chaves 2016; Wolf 2008). Secularization is produced by cohort replacement: older, more religious people die and their less religious grandchildren come of age. The composition of the population can change even if no individuals change.

But perhaps our focus on averages has led us astray. If we think of people falling into one qualitative category or another, rather than being located somewhere on a spectrum from highly religious to completely secular, things might look different. Take, for example, a simple three-fold classification of religious, secular, and whatever falls in between. There can be no dispute that the religious category is shrinking while the secular one is growing. The trajectory of the intermediate group seems more puzzling. It is becoming larger in some countries (such as Greece and Italy, where religion is strong) and smaller in others (such as Sweden and the Czech Republic), while staying comparatively static in places like Switzerland and Germany.

Was Andrew Greeley right, then, that it is not possible to generalize about religion and religious change in Europe? Even scholars sympathetic to the idea of secularization have been struck by the patchwork of different levels and kinds of religiosity (Draulans and Halman 2005; Halman and Draulans 2006). If countries with different histories and cultures show unpredictable variation in the size of the groups such as nominal Christians, the spiritual but not religious, and so on, searching for a general theory might seem misguided.

The middle ground between religious commitment and full secularity needs to be understood. Competing schools of thought characterize this group in different ways. Secularization theorists tend to see such positions as staging posts on a route with traffic going mainly in one direction. Some sociologists of religion maintain that these intermediate points (invisible religion, alternative spirituality, believing without belonging) can be destinations rather than halfway houses. For the market theorists, the territory is populated by consumers waiting for a better product.

Voas (2009) argues that this middle ground of fuzzy fidelity is the natural consequence of a general phenomenon: religious involvement starts to wane before many people become wholly secular. During a transitional period—which is protracted, because of the generational nature of religious decline—the intermediate category swells in size. It grows, plateaus, and then slowly dwindles, following a common trajectory. The onset of religious decline varies, but once secularization is underway, it tends to follow a familiar path at a relatively predictable pace. Variations in the European religious landscape conceal a common underlying pattern. Although the fuzzy category is very large—often around half the population in contemporary Europe—it is destined to be eroded as secularization proceeds and more and more people give up all forms of religious involvement.

A surprising number of sociologists of religion claim that falling adherence to Christianity does not show that people are less religious. Instead, they say, religion is being transformed into something less formal, doctrinal, and institutional. If people have not defected entirely to secularity, it is because they still crave some sort of religion or spirituality. This hypothesis implies that fuzzy fidelity will continue to grow, or at least maintain its large share of European populations.

The evidence suggests otherwise. As predicted by Voas (2009), the fuzzy category levels off and then starts to fall as the secular share overtakes the religious. In time, the wholly secular proportion of the population becomes larger than the fuzzy group as well. We can see this outcome taking shape in countries where secularization is most advanced—for example, in France, Scandinavia, and Central Europe (specifically the Czech Republic and Hungary). Other countries appear to be heading in the same direction. Europeans are not looking for some form of faith detached from the old institutions; their casual, partial, and occasional religious involvement is a transient phenomenon. This hypothesis has been found to apply to the United States (Brauer 2018), to which we now turn.

4. The United States of America

The United States has long been held up as an apparent counterexample to the secularization thesis. Despite being in the vanguard of modernization (at least in their own view), Americans have been a remarkably churchgoing people by Western standards. Belief in God has been almost universal, and a sizable minority of the population claims that the Bible should be interpreted literally.

It is now apparent, however, that the United States has not escaped religious decline. Moreover, the proximate cause of this decline is cohort replacement, just as in highly developed countries elsewhere (Voas and Chaves 2016). America is not an exception: secularization is happening, and it is driven by the generational dynamic described above for Europe. Most attention has been given to ‘the rise of the nones’—the rapid increase in the proportion of adults with no religious affiliation—but a drift away from religious practice and belief is just as apparent. Although the change is gradual, there is no mistaking the writing on the wall.

The most important data source on beliefs and attitudes is the General Social Survey, which started in 1972 and has been conducted at least every other year since then. Sample sizes range from around 1500 to 3000.

Table 1. The religiosity of old and young Americans in 2018 (%).

	Born before 1950	Born during the 1990s
Has a religion	89	63
Claims to attend services weekly or more often	43	13
Has no doubt about God’s existence	72	40

Source: US General Social Survey 2018.

below shows the key indicators from 2018: affiliation, practice, and belief have all dropped from the high levels found among people born in the first half of the 20th century to the much lower levels characteristic of young adults today. The erosion is dramatic, with attendant feedback effects: each generation is raised in a social environment that is less religious than that experienced by their grandparents and parents.

The 2020/21 wave of the GSS shows even more striking growth in non-religion, but the COVID-19 pandemic had forced a change from face-to-face interviews and some scholars argue that the highly religious were undercounted. We can sidestep this debate; the picture was already clear by 2018.

While many people now claim to be ‘spiritual but not religious’ (SBNR), it is clear that diffuse spirituality cannot compensate for the decline in organized religion. No institutions are being created, and these individualized forms of belief and practice provide no basis for collective action. Their impact is generally limited, both personally and socially. Furthermore, SBNR seems to be less successfully transmitted from parents to children than conventional Christian religiosity. New generations will find their own paths to self-help and wellbeing.

Each successive cohort of Americans shows lower religiosity than the one before, and these generation gaps persist through adulthood: the United States is on course to repeat the experience of the rest of the Western world. Denial that religion and religiosity are declining in America has gone from being orthodoxy to an eccentricity. It is often still asserted, however, that the ‘nones’ are just ‘nothing in particular’; that is to say, they are religious despite not being currently attached to a specific church or denomination. But a Pew Research Center survey from 2015 suggests that around two thirds of nones defected because they stopped believing or started questioning, another large group dislikes organized religion, and only a tenth are inactive believers (Lipka 2016).

Commentators tend to focus on the astonishingly rapid rise of the nones in the United States, but secularization is not only or even mainly about the number of people who no longer belong. People with a religion become less religious. Churchgoers in Iowa today are more secular than their grandparents were; they are less likely to object to dancing, playing cards, or divorce, and more likely to go shopping on Sundays, or spend time on matters unconnected to personal salvation.

We do not dispute that religious participation has been and remains high in the United States compared to most (though not all) European countries. What we deny is the doctrine of American exceptionalism. Modernization creates problems for religion, even in the US. And the result is what we have seen elsewhere: a gradual decline in affiliation, attendance at services, and belief in Christian doctrine, driven by the replacement in the population of the religious old by the more secular young. The process may be slowed if cultural conservatives organize effective resistance, but currently the situation seems to be the reverse: religion is collateral damage in the progressive backlash against traditional values.

We still need to understand why secularization was late in coming to the United States. Some nationally specific factors—the size of the country, the requirements of immigrant incorporation, the existence of communal subcultures, the autonomy of local schools, and so on—may be relevant. But it would be a mistake to look for national causes when secularization is an international experience, and the same forces are at work across the post-industrial world.

5. Conclusions

For a long time after Wilson published Religion in Secular Society, reliable data on religious interest and involvement were sufficiently sparse that it was possible for scholars to deny the secularization of Britain by deflating the religiosity of the pre-industrial past and exaggerating the importance of the occasional sign of life in the present; for example, the decline of church attendance could be neutered by pointing to the apparent popularity of religious broadcasting on radio and television. And every innovation could be heralded as the first sign of revival. However, as Field’s massive compendium of data amply demonstrates, Pentecostalism, the 1960s new religious movements, the charismatic movement, and New Age spirituality have all failed to divert the declining trajectory.

As we have demonstrated, Stark was wrong about religion in the UK; he was also persistently wrong in repeatedly summarizing the secularization thesis as predicting the worldwide death of religion. The secularization approach is, in the first place, an attempt to explain the past of the West. If it is correct, then we can turn it into the prediction that any society experiencing the same changes will (with the two Wallis and Bruce exceptions) also become more secular. However, in simplifying the prediction, Stark missed two crucial points.

First, modernization, in the form assumed in the secularization approach, is not inevitable. China, for example, has developed a very successful manufacturing industry but, as its treatment of Falun Gong and its Uighur minority show, it is a long way from accepting the egalitarianism that is essential to the growth of tolerance as a response to diversity. Many African, Middle Eastern, and Asian countries show little sign of democratizing or of shifting from religio-ethnic to civic nationalism.

Second, there is an obvious and profound difference between the secularization of the West in the late nineteenth and twentieth centuries and the possible future of anywhere else. When the West changed, it did so de novo. But secular societies now exist and they can be taken by other societies as a model to emulate (for example, by Turkey under Ataturk), a fate to be avoided at all costs (the view held by such Muslim regimes as Afghanistan’s Taliban), or something in between.

We suspect that the Starkian distortion of secularization is based on a simple misreading of ‘irreversible’ as ‘inevitable’. There are no good reasons for thinking that secularization simply accompanies the passage of time, but there are many good reasons for thinking that popular secularization (as distinct, for example, from the elite imposition of secularity) will be difficult to reverse. These are given at length elsewhere (Bruce 2011, 2020) but can be briefly summarized as the high social costs of imposing a shared belief system on a culturally diverse democracy and the difficulty of religious conversion, to which we now turn.

Given that religious families are failing to recruit their own children at the necessary rate for stability, let alone growth, any religious revival will depend on recruiting the currently religiously indifferent. Although the new religious movements of the 1960s failed miserably to fill the gap left by the decline of Christianity, they did encourage many social scientists to study conversion. This literature is complex, but a general conclusion is that people are unlikely to be persuaded by some new belief system presented by people who are quite unlike themselves (Bruce 2018). Conversion requires that the targets see something of themselves in the evangelist. In the modern parlance, the new faith needs to be relatable: the targets need to think ‘That guy is like me but better’. People are very rarely persuaded by carriers of alternative religious faiths who seem foreign, alien, or exotic. Any young, religiously indifferent, majority-ethnicity person in the UK today who, for a moment, considers acquiring a faith will be confronted by the oddity of the religious: they are old women, they are inhabitants of the geographical peripheries, they are Pakistani, or they are West African.

Furthermore, conversion normally requires the development of an enduring positive relationship between the targets and the carriers of the faith. Given that the seriously religious are now only some 5 per cent of the UK population (and geographically concentrated), the odds of a non-religious person ever meeting a seriously religious person, let alone forming a positive social relationship, are slim. In such circumstances, it is difficult to see how significant numbers of people can be converted.

The alternative is imposition. It is possible to imagine a ruling class using the levers of power (even in a democracy: the Trumpian USA offers a glimpse) to impose a religion (or in the case of China and the Uighurs, a non-religion) on its citizens. But, as Iran is now learning, imposition may produce outward conformity but it does not persuade. Quite the opposite. As the post-1945 communist states of Europe demonstrated, imposition breeds contempt, cynicism, and subversion. For these reasons, we find it hard to imagine how largely secular societies such as the UK, Denmark, the Netherlands, or Australia can now experience a religious revival.

There is no harm in academic disagreement: social science, like its natural science counterpart, benefits from deviants challenging a comfortable consensus with difficult questions. However, there comes a point where the evidence is so clear that continued dissent is pathological rather than helpful. There is much yet to be learned about the precise details of secularization and there is every good reason to believe that local and national studies will improve our understanding of the relevant causal connections but there is no good reason to continue to doubt secularization.