"Secularization" is one word pointing at three different things: the institutional separation of religion from politics, law and education; a decline in individual belief and practice; and the withdrawal of religion from public life. The evidence for each is of a completely different strength, and mixing them guarantees conclusions that contradict each other. One counterintuitive fact — the number of religiously unaffiliated people grew by 270 million in a decade, to 1.9 billion, and yet, on demographic grounds, that group's share of the world's population may well fall over the coming decades.
"By the 21st century, religious believers are likely to be found only in small sects, huddled together to resist a worldwide secular culture."— Peter Berger, The New York Times, February 1968; he publicly retracted the judgement in 1999
Secularization was not a theory built on data; it ran the other way. The "disenchantment" (Entzauberung, literally de-magicking) of Weber's 1917 lecture Science as a Vocation claims this: the world has come to be treated as something masterable in principle by calculation, so that one need no longer appeal to mysterious powers. That is a claim about how the world gets explained, not a headcount of believers — and the argument that followed has confused the two ever since.
The thesis hardened in the 1960s. Wilson's Religion in Secular Society (1966) and Berger's The Sacred Canopy (1967) made it a mainline of sociology: modernization brings functional differentiation, each sphere running on its own logic; pluralism then dissolves any monopoly on belief — once you know your neighbour believes something else, your own account drops from fact to choice. In 1968 Berger gave the Times his famous forecast.
By the late 1970s reality pushed back: the 1979 Iranian revolution, the organization of the American religious right, the explosive growth of Pentecostal churches in Latin America, the rise of religious politics in South Asia — all in places where modernization was advancing, not retreating.
Casanova's Public Religions in the Modern World (1994) made the decisive cut: it split "secularization" into three independent claims — differentiation, decline, privatization — and argued that only the first holds, while the third is directly refuted by "deprivatization," religion's return to public argument. In 1999 Berger edited The Desecularization of the World, writing that today's world is "as furiously religious as it ever was" and that the literature loosely labelled secularization theory is "essentially mistaken."
Taylor's A Secular Age (2007) changed the question: not how many believe, but what has changed in the conditions of believing. He distinguishes three senses — separation of church and state, the statistics of belief and practice, and the shift by which faith stops being self-evident and becomes one option among many. The third cannot be reached by a questionnaire, yet may be the layer that has really moved: devout and unbelieving alike now share one thing — both know they could be otherwise.
A decade of shifting shares. Pew Research Center's estimates released in June 2025 cover 2010–2020: Christians 2.3 billion, 28.8% of the world (30.6% in 2010); Muslims 2.0 billion, 25.6% (23.9% in 2010), a gain of 347 million — more than every other group's growth combined, driven by age structure and fertility rather than conversion; the religiously unaffiliated 1.9 billion, 24.2% (23.3% in 2010), firmly the third-largest category; Hindus 1.2 billion, flat at 14.9%; Buddhists down 0.8 points to 4.1%, the only major tradition whose absolute numbers also fell.
The centre of gravity is moving. In 2020, 30.7% of the world's Christians lived in sub-Saharan Africa, against 22.3% in Europe — the first time that has been true. Europe's Christian population fell to 505 million, down about 9% in the decade. One tradition, contracting at one end and swelling at the other.
Which way people move. Pew's March 2025 survey of 36 countries found that about one in ten adults under 55 who were raised in a religion has left it. Retention rates differ sharply: roughly 99% for Hindus and Muslims, 83% for Christians — for every 100 people raised Christian, 17.1 leave and 5.5 join, a net loss of 11.6 — and 78% for Buddhists, the lowest. Most of that movement does not go to another religion; it goes to "unaffiliated."
The most closely studied case is the United States. The unaffiliated rose from 16% in 2007 to 23% in 2014 and 29% in 2023–24, then held around 30% for the past five years. Pew's December 2025 report puts it this way: the decline has stalled, but there is no revival — and about 55% of young adults claim a religion, still well below their parents' generation.
What "religious population" measures is what people report. Iran's official census says 99.5% Muslim. In a 2020 online survey by GAMAAN with more than fifty thousand respondents, only 40% identified as Muslim and 32% as Shi'a, while 78% said they believed in God. That was a self-selected, non-probability sample and cannot be read as a census. But the size of the gap itself makes a point: where declaring an identity carries a cost, a questionnaire records what can safely be said out loud.
Projections do not splice. Pew's 2015 round projected that by 2050 Muslims (about 2.8 billion, 30%) would nearly match Christians (2.9 billion, 31%), with the unaffiliated share falling to roughly 13%. But that round's 2010 baseline is not the same as the 2025 estimates (the unaffiliated stood at 16.4% on the old basis, 23.3% on the new), so the two series cannot be joined end to end. The directional finding still holds: the unaffiliated are older and have fewer children — only about 9% of babies born in 2055–60 are projected to have religiously unaffiliated mothers — so even with a steady one-way flow of switchers, the religious share of humanity need not fall.
The argument only becomes tractable once the three claims are pulled apart.
Two explanations compete, and each accounts for about half.
Davie's "believing without belonging" describes Britain: far more people believe than attend. Its mirror image is the Nordic "belonging without believing" — very high church membership alongside low rates of belief, where membership works more like a cultural and fiscal inheritance. Identification, belief, and practice are three different things, and a number from any one layer cannot speak for the other two. That is exactly where national "religious population" figures are most often misread.
Misconception: modernization inevitably makes religion disappear.
Of the three claims, only institutional differentiation holds globally. Individual-level decline is real in Western Europe, but treating it as a universal law of modernization has been contradicted by developments since 1979 — and Berger himself withdrew the prediction.
Misconception: "religiously unaffiliated" means atheist.
Seven in ten American "nones" believe in God or a higher power and more than six in ten in spiritual forces; those who reject every such item are about a fifth of the group. The category records a relationship with institutions, not a worldview.
Misconception: the whole world is becoming less religious.
Regional differences outrun any global trend: Europe's Christian population fell about 9% in a decade while sub-Saharan Africa became the world's Christian heartland. And because of fertility and age structure, the unaffiliated share of the world may not rise at all.
Misconception: the U.S. is either secularizing to a majority or already in revival.
Five years of data support neither: the unaffiliated hold around 30% and observance shows no clear rise or fall. But young adults remain below their parents — a plateau and generational replacement can coexist, and the long run depends on whether this cohort returns once it has families.
The unaffiliated sit in the middle of the spectrum, not at its end. Day 21 laid out the range from monotheism to non-theism; the figures here add a layer to it. The world's third-largest "group" is not a community of belief but an unfilled form — it holds committed atheists alongside people who believe in God and walk through no door. Counting them wholesale as a secular bloc is as wrong as counting them wholesale as prospective believers.
The same thread as the last two issues. Day 23 argued that who may interpret matters more than what the text says; here the holders of that right are changing — institutional authority is leaking away and networks and algorithms have taken over most of the distribution. Day 24's lesson also stands: falling religious participation does not automatically reduce conflict.
Two opposite curves are true at once. By share, the religious population may well rise globally; by intensity and institutional authority, most wealthy societies are still declining. The hard part is not which is right but that the two point at different layers — demography counts births and inheritance, sociology counts belief and participation, and they have never moved in step. The next two issues step into that gap: Day 26 on new religious movements and modern spirituality, Day 28 on the worldview of the nonreligious.
This piece surveys the intellectual history of secularization, the current data, and how traditions have responded, from a neutral and comparative standpoint. It offers no praise, blame or ranking of any religious or non-religious position and passes no judgement on any living religious figure. Population figures are estimates from research organizations, vary by definition and method, and cannot be spliced across reports.