Argentina 2 waves · 2008–2019
Macro-region survey religion (CEIL-CONICET Encuesta Nacional sobre Creencias y Actitudes Religiosas), 2008 and 2019, across the six standard macro-regions — survey estimates with sampling error, not census counts (Argentina's census dropped religion after 1960). National margin ±2% at 95%; per-region subsamples near 400 carry a larger, unpublished margin. The 2019 affiliation share is a disclosed lower bound (the source chart prints only values above 2%), and no cross-wave change metric ships. 2019 reuse CC BY-NC-SA 4.0 via the peer-reviewed article, aligned with the project's own licence; reuse of both waves confirmed by CEIL-CONICET (2026-07-13), with the requested Informe 25 citation carried in the attribution.
Bahrain 1 wave · 2020
National census religion (Information & eGovernment Authority, iGA), 2020, on a flat two-line frame (Muslim, مسلم; Others, أخرى) with no no-religion category; the map shades the reference-group Muslim share and its exact complement, the Minority share, which is arithmetic on the published categories and not a no-religion measure. Bahrain publishes census religion at national geography only; the map shades one national polygon. Governed by the Bahrain Open Government Data License v1.0. 2020 only, no change metric; the 2010 census total is held.
Bangladesh 2 waves · 2011–2022
District (zila) census religion (Bangladesh Bureau of Statistics, BBS), 2011 and 2022 on the stable 64-zila frame, on a flat five-category frame (Muslim, Hindu, Christian, Buddhist, Others) with no no-religion category; the map shades the reference-group Muslim share, its exact complement the Minority share (arithmetic on the published categories, not a no-religion measure), and the change in the Muslim share across the two waves. Bhola's 2011 composition is withheld on a source-PDF defect and reads no-data. BBS asserts copyright with no open-reuse licence located; the derived summaries ship with attribution while a reuse ask is outgoing, and the page reverts in a single commit if BBS objects.
Barbados 1 wave · 2010
Parish census religion (Barbados Statistical Service, BSS), 2010, from the 2010 Population and Housing Census Volume 1 Table 02.06 — an integer full-count of the Tabulable Population (226,193), both margins exact. No religion is the map-worthy signal: 20.6% nationally, from 16.3% in St. Philip to 25.7% in St. Joseph, with urban St. Michael second. One census wave, no change metric; the 2021 census collected religion but published no religion table. BSS Open Licence Agreement; boundaries CC BY 2.5 Generic.
Belize 3 waves · 2000–2022
District census religion (Statistical Institute of Belize, SIB), 2000–2022 on the six districts, with a wave-to-wave headline change metric. The national no-religion share rose 9.4% to 15.5% to 31.0%, among the sharpest secularisation series in the corpus. 2000 is an integer full-count; the 2010 and 2022 cells are SIB weighted non-integer counts, reconciled float-exact and disclosed. Fine-denomination change is withheld across the 2000-to-2010 frame break; the headline shares stay comparable. No open-data licence is stated on the SIB source; the summaries ship with attribution while a reuse ask is outgoing. Boundaries geoBoundaries BLZ ADM1 (CC BY 2.5).
Bhutan 11 years · 2013–2023institution register
Dzongkhag administrative register counts of religious institutions (National Statistics Bureau, Dzongkhag at a Glance), 2013–2023. The counts describe buildings, never people; Bhutan collects no census or survey data on personal religious affiliation. Each dzongkhag office and edition defines religious institution locally, so counts are not comparable across dzongkhags or necessarily across years; register revisions and redefinitions are flagged in the popup series. No reuse terms stated; used with attribution pending confirmation. Boundaries geoBoundaries BTN ADM1 (CC BY 3.0).
Bosnia and Herzegovina 1 wave · 2013
Municipality census declared religion (Agency for Statistics of Bosnia and Herzegovina, BHAS), 2013, on the 142-unit post-Dayton frame (141 municipalities plus Brčko District); 20 verbatim categories, both margins exact to 3,531,159. The two non-response lines stay in the denominator, and affiliation and no-religion shares need not sum to 100. The 2013 census was disputed — the Republika Srpska institute rejected the BHAS methodology; the map renders the BHAS state publication as the official record, stated neutrally. No open licence stated: BHAS attribution, reuse confirmation pending. 1991 held (that census measured ethnicity, not religion).
Bulgaria 3 waves · 2001–2021
District census religion (National Statistical Institute, NSI), 2001–2021; a voluntary question whose published basis differs in every wave (full population, then respondents, then excluding additions without religion information), and the waves are separate snapshots with no change metric.
Cabo Verde 2 waves · 2010–2021
Concelho census religion or spirituality (Instituto Nacional de Estatística, INE), 2010 and 2021 on the 22 concelhos across ten islands, resident population aged 15 and over, voluntary question. Both waves reconcile integer-exact (336,049 / 352,494). The secularisation gradient is the signal — national no religion 10.8% to 15.6%, urban São Vicente 30.9% to 38.2% against a near-zero rural Santiago interior. The 2021 census renamed and split fine categories; fine-category change is withheld across the instrument break while the headline shares are read on the year selector. INE states no explicit reuse licence; the derived summaries ship with attribution while the position is reviewed. Boundaries geoBoundaries CPV ADM1 (ODbL).
Cambodia 2 waves · 2008–2019
Province census religion (National Institute of Statistics, NIS), 2008–2019, from the 2019 census Table 2.5.1 on a re-based four-category frame (Buddhist, Muslims, Christians, Other) with no no-religion category; the map shades the reference-group Buddhist share and its exact complement, the Minority share, which is arithmetic on the published categories and not a no-religion measure. 2008 is NIS's retabulation onto the current 25-province frame; both waves therefore share one geometry, and the printed one-decimal cells reconcile within a derived 0.20 pp rounding bound. A Buddhist-share change metric is offered but withheld for Mondul Kiri and Ratanak Kiri, whose highland Other-share shifts the report attributes to reclassification. NIS asserts all rights reserved; the derived summaries ship with attribution (PI ruling, 2026-07-11).
Chile 2 waves · 2002–2024
Commune census religious affiliation (Instituto Nacional de Estadísticas, INE, via BCN Estadísticas Territoriales), 2002 and 2024 on the current 346 communes, population aged 15 and over, with a wave-to-wave change metric on the stable frame. Chile has a real Ninguna (no religion) line: the national share climbs from 8.3% to 25.7% across the pair, among the strongest secularisation series in the corpus. Five communes are null in 2002 (four created in 2004, plus Puchuncaví, a series gap) and the four 2004-parent communes render no change across the reclassification. Reused under CC BY-SA 4.0 (share-alike); the mapped shares are the project's derivation, not INE's own analysis.
Cook Islands 3 waves · 2011–2021
Island census religious-affiliation overlays (Cook Islands Statistics Office), 2011–2021, across twelve census islands; census affiliation that measures neither belief nor practice. The Cook Islands Christian Church share fell sharpest on Rarotonga, and the boundary is a mixed-licence composite (geoBoundaries islands, OpenStreetMap northern atolls).
Côte d'Ivoire 1 wave · 2021
Sub-prefecture and commune census religious affiliation (INS/ANStat RGPH 2021), on an ordinary-household denominator; the source's own printed arithmetic differs by one to three people across many rows and ships unchanged with disclosure, and the Bona National Park boundary unit carries no census data. 2021 only, no change metric.
Finland 36 waves · 1990–2025membership register
National administrative religious-community membership from the Population Information System (Statistics Finland, StatFin table 11rx), 1990–2025; a register construct published as a national total only, measuring neither census affiliation, belief, nor practice. “Persons not members of any religious community” keeps the source name as register non-membership and is not a no-religion measure; the tap-for-composition popup carries the 23 mutually exclusive verbatim register categories. StatFin data CC BY 4.0; the geoBoundaries ADM0 boundary is ODbL 1.0 (OpenStreetMap), with a CC BY 4.0 national-vector swap recorded as deferred.
Georgia 3 waves · 2002–2024
Region census religion (National Statistics Office of Georgia, Geostat), 2002, 2014 and 2024 on the 12-region frame; every wave reconciles exactly at both margins (4,371,535 / 3,713,804 / 3,929,581). Category frames differ per wave and are never merged; the comparable spine is the five great-tradition shares, and no cross-wave change metric is offered. 2002 no religion is null (the residual folds no-religion and other-religion together inseparably) and 2002 affiliation is a recorded lower bound; 2014 carries verbatim ≤10-cell suppression. Territorial scope rendered as published: Abkhazia and South Ossetia are not enumerated. Geostat Terms of Use open reuse, accepted.
Grenada 1 wave · 2021
Parish census religion (Central Statistical Office of Grenada, CSO), 2021, from Table 23 of the PRELIMINARY census results — an integer full-count over 26 verbatim categories across the seven parish units, integer-exact to 108,279 at every margin. The corpus's first preliminary-data ship: the counts are official CSO figures but may be revised, and the product regenerates when the final 2021 report lands. Roman Catholic (31.5%) and Pentecostal (19.9%) lead nationally; the no-religion share sums the two distinct no-religion lines (Atheist + No Religious Affiliation, 6.0%), with Not Stated held in the denominator and outside both shares. The Town of St. George rolls into St. George parish (disclosed); six of seven parishes carry an under-ten category count. One census wave, no change metric (2001 and 2011 religion is national only). CSO Open Licence Agreement; boundaries geoBoundaries GRD ADM1 (CC BY-SA 2.0).
Guinea 1 wave · 2014
Regional census religious-affiliation overlays for the eight administrative regions, 2014, from the Institut National de la Statistique (INS), RGPH-3 Table 5.10. A single census wave in percentages only: the source publishes one-decimal shares with no counts, and no change metric is offered. N'Zérékoré is the sharpest published gradient, 85.8% affiliated and 14.2% no religion, against the near-uniform 99% Muslim north. CC BY 4.0 via the INS Accord de licence de données ouvertes.
Guinea-Bissau 1 wave · 2009
Regional census religious-affiliation overlays for the eight administrative regions plus the Sector Autónomo de Bissau, 2009, from the Instituto Nacional de Estatística (INE), RGPH-2009 Quadro 4. A single census wave in percentages only: the source publishes one-decimal column shares on the Guinean-nationality universe with no counts, and no change metric is offered. A large Não declarado (Not declared) non-response share, 10.5–25.7% by region, is the gap between the shown shares and 100, and the affiliation shading tracks that non-response gradient more than a difference in religiosity. Census reuse is under review (INE all-rights-reserved footer, no located open-data licence); boundaries geoBoundaries GNB ADM1 (ODbL 1.0).
Guyana 1 wave · 2012
Regional census religion (Bureau of Statistics, Guyana), 2012, from Census Compendium 2 Table 2.19: integer counts across thirteen categories closing exactly at both national margins to 746,955. The no-religion share varies more than sixfold, from 1.15% in Upper Takutu-Upper Essequibo to 7.25% in Upper Demerara-Berbice; the source table prorates non-response, and the two shares sum to 100 by construction. One census wave, no change metric. BoS Open Licence Agreement; boundaries geoBoundaries GUY ADM1 (ODbL).
Hungary 3 waves · 2001–2022
County census religion (Hungarian Central Statistical Office, KSH), 2001–2022; a voluntary question with shares among stated responses, and non-response grew from 10.8% to 40.1% across the waves — cross-wave changes may reflect changing respondent composition as well as changing affiliation.
Iceland 29 waves · 1998–2026membership register
National register membership in religious and life-stance organisations (Statistics Iceland), 1998–2026; a register construct published as a national total only, measuring neither census affiliation, belief, nor practice. “No religious organisation” keeps the source name and is not a no-religion measure.
Indonesia 1 wave · 2010
Province census religion (Badan Pusat Statistik, BPS, Sensus Penduduk 2010), 33 provinces on the 2010 frame, verbatim seven-religion frame (Islam, Kristen, Katolik, Hindu, Budha, Khong Hu Chu, Lainnya) with no no-religion category; the map shades the reference-group Islam share and the Minority share, the summed share of the six other named religions — arithmetic on the published categories and not a no-religion measure. Islam runs from 9.1% (Nusa Tenggara Timur) and 13.4% (Bali) to 98.2% (Aceh). BPS asserts copyright with no open-reuse licence located; the derived summaries ship with attribution while a reuse ask is outgoing, and the page reverts in a single commit if BPS objects. 2010 only (SP2020 dropped the question), no change metric.
Iran 3 waves · 2006–2016
Province-concordance census religion (Statistical Centre of Iran, SCI), 2006–2016, on 28 analytical units; the record uses the SCI recognised categories, and identities outside them, including unrecognised religious minorities, do not appear. No explicit SCI reuse licence located; published with attribution under project-lead approval.
Jamaica 1 wave · 2011
Parish census religious affiliation (Statistical Institute of Jamaica, STATIN), 2011, from Volume 6 Table 5 — integer counts across the 14 parishes, exact to the 2,684,115 religion universe at every printed margin. The printed frame is heterogeneous and ships verbatim: eleven parishes carry 19 categories, Saint James and Saint Catherine 20, and Saint Andrew 23 (small-group breakouts printed only there; the parish Other lines absorb the rest by source design). Seventh-Day Adventist (12.0%) and Pentecostal (11.9%) lead nationally; 21.3% report no religion, the slot assigned by role across three printed spellings, with non-response held in the denominator and outside both shares. One census wave, no change metric (2001 religion is national-only; 2022 has no religion release yet). STATIN Terms & Conditions of Data Use; boundaries geoBoundaries JAM ADM1 (CC BY-SA 2.0).
Kazakhstan 2 waves · 2009–2021
Regional census religious affiliation (Bureau of National Statistics, BNS), 2009 and 2021, each wave on its own boundary vintage: 16 regions in 2009 and 17 in 2021 after the 2018 South Kazakhstan split into Turkistan and Shymkent; the year slider switches the frame. Count-valued and exact at both margins (16,009,597 / 19,186,015). The Refused-to-indicate residual jumps from 0.5% to 11.0% between waves and is disclosed on every surface; a lower 2021 affiliation share reflects more non-response rather than measured secularisation. No cross-wave region change. BNS open reuse with attribution; boundaries geoBoundaries KAZ ADM1 (ODbL) and OCHA COD-AB (CC BY-IGO).
Kiribati 6 waves · 1990–2015
Island census religious-affiliation overlays (Kiribati National Statistics Office), 1990–2015, the deepest island time series in the Pacific stack; census affiliation that measures neither belief nor practice. Kiribati straddles the antimeridian, and 2020 religion was published at national level only.
Kosovo 2 waves · 2011–2024
Municipality census religious affiliation (Kosovo Agency of Statistics, KAS), 2011 and 2024 on the 38-municipality frame, from one machine-readable KAS table; both waves reconcile exactly (1,739,825 on the 34 enumerated municipalities in 2011; 1,585,566 on all 38 in 2024). The four northern Serb-majority municipalities were unenumerated in 2011 (boycott) and only partially enumerated in 2024; they render as no-data in 2011 and as published in 2024, and no cross-wave municipal change is claimed across the coverage break. Open KAS reuse with acknowledgement; boundaries geoBoundaries XKX ADM2 (CC BY-SA 2.0, share-alike carried).
Malaysia 4 waves · 1991–2020
Administrative-district census religious affiliation (Department of Statistics Malaysia, DOSM), 1991, 2000, 2010 and 2020 on the stable printed 160-district frame. Islam (63.5%), Buddhism (18.7%) and Christianity (9.1%) lead nationally in 2020. The map shows religious affiliation, the fused No Religion/Unknown share, and wave-to-wave affiliation change. Percentages use DOSM's printed religion-block denominator (the printed district population total can differ, and every deviation is recorded, never repaired); 74 district-waves remain null as printed or unpublished. Attribution-based reuse under DOSM's terms of use; the OSM-derived boundary ships share-alike under ODbL 1.0.
Micronesia (Federated States) 3 waves · 2000–2023
State census religious affiliation from the FSM Statistics Division (and SPC SDD for the 2000 state reports), 2000, 2010 and 2023, for Yap, Chuuk, Pohnpei and Kosrae. Every wave counts all persons of all ages with no universe break; broad affiliation is therefore comparable across all three waves. The substantive story is denominational, Kosrae overwhelmingly Congregational against Catholic-majority Yap and Chuuk. The no-religion slot combines the source's No Religion and Refused lines. Kosrae's 2023 No religion/Refused cell is asterisk-suppressed and renders as no-data, and the 2023 Table B6 carries a one-person national total gap and further asterisk suppressions, all disclosed and never repaired. No reuse licence is stated on the FSM census tables; the derived summaries ship with FSM Statistics Division attribution while a courtesy reuse ask is outgoing and revert in a single commit if the division objects.
Moldova 3 waves · 2004–2024
Raion and municipality census religion (National Bureau of Statistics of the Republic of Moldova, BNS), 2004, 2014 and 2024 on the stable 35-unit right-bank frame; every wave reconciles exactly at both margins (3,383,332 / 2,804,801 / 2,409,207). Category frames differ per wave and are never merged; each wave's verbatim categories ride the popup composition block, and no cross-wave change metric is offered — the no-religion block is not cell-comparable across waves, and the not-declared line stays in the denominator. Territorial scope rendered as published: the right bank only is enumerated, and Transnistria and Bender render as no-data, never zero. BNS Terms of use, Creative Commons Attribution 4.0.
Montenegro 3 waves · 2003–2023
Municipality census religion (Statistical Office of Montenegro, MONSTAT), 2003–2023, on the stable 21-municipality 2003–2011 frame; the 2023 census reports 25 municipalities, aggregated to their historical parents as complete partitions (Petnjica into Berane, Gusinje into Plav, Tuzi and Zeta into Podgorica). Municipality population totals are published in full and reconcile exactly; for 2023, MONSTAT suppresses small category cells and the affiliation and no-religion shares are published-cell lower bounds where suppression applies (247 persons across 17 of 21 municipalities), disclosed and never redistributed.
Myanmar 1 wave · 2014
State/region census religion (Department of Population, Myanmar), 2014, from the Union Report Volume 2-C (Religion), Table 1: seven verbatim categories on the enumerated count basis for the 15 units, both margins exact to the enumerated Union total 50,279,900. An estimated 1,090,000 persons in Rakhine (plus Kayin 69,753, Kachin 46,600) were not enumerated; the enumerated results are the count basis, and the report's estimated-total presentation (denominator 51,486,253, Union Islam 2.3 → 4.3 per cent under its stated Rakhine-Islam assumption) is carried only as Union-level context, never blended. The Department of Population states the enumerated Rakhine and Union religion profile is inconclusive. The 2024 census is a frame-and-universe break, deferred. No stated reuse terms: Department of Population attribution, licence under review. Boundaries OCHA COD-AB (source MIMU, CC BY-IGO).
Nepal 1 wave · 2021
District census religion (National Statistics Office of Nepal, NSO), National Population and Housing Census 2021, Table 5, on the ten-category official frame printed verbatim (Hindu, Bouddha, Islam, Kirat, Christian, Prakriti, Bon, Jain, Bahai, Sikha) with no no-religion category; the map shades the reference-group Hindu share and its exact complement, the Minority share, which is arithmetic on the published categories and not a no-religion measure. The composition is sharply regional: Rasuwa is 70.8% Bouddha, Panchthar 55.7% Kirat, Rautahat 22.5% Islam. The NSO confirmed reuse on 2026-07-13, requesting citation and acknowledgment; the derived summaries carry NSO attribution. 2021 only, no change metric; the 2011 district wave is held.
North Macedonia 1 wave · 2021
Municipality census religious affiliation (State Statistical Office), 2021, from census table T-06: integer counts across thirteen verbatim categories closing exactly at both margins, and across the City of Skopje ten-municipality partition, to 1,836,713. The gradient is sharp: Saraj is 93.6% Muslim in the source table, and central Skopje's Centar carries the highest no-religion share at 4.5% against the national 0.48%. The administratively-sourced component (132,260 persons, 7.20%) renders as its own published category, never repaired. One census wave, no change metric; 2002 held on the pre-2004 boundary frame. SSO cite-the-source clause; boundaries OCHA COD-AB ADM2 (CC BY-IGO).
Norway 21 waves · 2005–2025membership register
Diocese Church of Norway membership register counts (Statistics Norway), 2005–2025. SSB publishes no diocese population denominator; the map shows counts only. The register records administrative membership; it does not record census affiliation, attendance, or belief.
Pakistan 1 wave · 2023
District census religion (Pakistan Bureau of Statistics, PBS), Census 2023, Table 9, on the eight-category official frame printed verbatim (Muslim, Christian, Hindu Jati, Qadiani/Ahmadi, Scheduled Castes, Sikh, Parsi, Others) with no no-religion category; the map shades the reference-group Muslim share and its exact complement, the Minority share, which is arithmetic on the published categories and not a no-religion measure. The record is rendered verbatim and the contest reported: Qadiani/Ahmadi is the state's post-1974 enumeration label for the Ahmadiyya community, regarded by many Ahmadis as pejorative and contested by the community. The tables cover the four provinces and the Islamabad Capital Territory; Azad Jammu and Kashmir and Gilgit-Baltistan are outside the published religion tables and are not mapped. PBS asserts no open licence; the derived rates ship with attribution while written confirmation is sought, and the page reverts in a single commit if PBS objects. 2023 only, no change metric.
Palau 3 waves · 2005–2020
State census religious affiliation (Palau Office of Planning and Statistics), 2005–2020 on 16 states. The universe breaks across waves: 2005 counts all persons with a real no-religion category, while 2015 and 2020 count adults 18 and over and fold no-religion into Other, making affiliation flat at 100% by construction so that it carries map signal only in 2005; no change metric is offered.
Paraguay 1 wave · 2002
Department census religion (Instituto Nacional de Estadística, INE), 2002, on the 18 first-level units (17 departments plus the Asunción capital district), from a parametrically pinned REDATAM cross-tabulation (base CPV2002) of ‘Religión que profesa’ recoded to the published cuadro P16 seven-group frame; the seven recoded national group totals equal the published P16 totals to the person, reconciling to 3,892,603. The No informado non-response group stays in the denominator and outside both shipped shares, so affiliation and no religion need not sum to 100. Boquerón, in the Mennonite Chaco, is the one Evangelical-majority department and carries the highest no-religion share (10.34%). One census wave, no change metric (2012 and 2022 asked no religion; 1992 is not in REDATAM). Census reuse accepted under the Paraguay open-data licence (Decreto 4064 / Ley 5282/2014); boundaries geoBoundaries PRY ADM1 (CC BY 4.0, source DGEEC).
Peru 2 waves · 2007–2017
Department census religious affiliation (Instituto Nacional de Estadística e Informática, INEI), 2007 and 2017 on the 25 departments, population aged 12 and over, with a wave-to-wave change metric on the stable frame. Four verbatim INEI categories (Católica, Evangélica, Otra, Ninguna); no religion runs from 1–2% in Piura and Apurímac to 11.3% in San Martín (2017). 2017 carries exact published counts; the 2007 category counts derive from published one-decimal percentages under a recorded rounding bound. INEI states no explicit reuse licence; the derived summaries ship with attribution while the position is reviewed.
Philippines 1 wave · 2020
Province and highly urbanised city census religious affiliation (Philippine Statistics Authority, PSA), 2020, on a household-population denominator that differs from total population; highly urbanised cities are folded into their host province and the BARMM Special Geographic Area is reconciled nationally only, not mapped. 2020 only, no change metric.
Poland 10 waves · 2014–2024attendance count
Diocese Catholic Mass-attendance head count (ISKK dominicantes, communicantes), 2014–2024 (2020 not counted); a practice count, not census affiliation. 17 of 41 diocese boundaries are approximate digitisations, seven flagged low confidence.
Saint Kitts and Nevis 3 waves · 1991–2011
National census religious affiliation (Department of Statistics, St Kitts and Nevis), 1991–2011; religion is published for the whole federation only, so the map shades one national polygon under the small-country clause. The 1991 wave carries the source's printed one-decimal percentages exactly; 2001 and 2011 derive from printed counts. Reuse under the DOS-SKN Open Licence Agreement; the 2011 religion-table total (47,195) differs from the resident-population figure (46,398), recorded without repair.
Saint Lucia 3 waves · 2001–2022
District census religion (Central Statistical Office), 2001–2022; the source's own category cells differ from its printed district totals by one to three people and ship unchanged with disclosure, the 2022 wave is Provisional Results, and cross-wave change is withheld because the category frames differ.
Saint Vincent and the Grenadines 2 waves · 2001–2012partial layer
Census-division Christian-denomination shares for 2001 and 2012 across the thirteen divisions. Every percentage uses the division's total population as its denominator, and the printed tables omit the non-Christian, no-religion, and not-stated remainder at division grain, so rows do not sum to 100 by source design; the percentage-only Christian layer is hatched and explicitly tagged as partial. Statistical Office data used with attribution pending confirmation; boundaries United States Census Bureau Demobase, a US-government work.
Seychelles 1 wave · 2022
District census religion (National Bureau of Statistics, NBS), 2022, from the Population and Housing Census Table B4.1 — an integer full-count of the all-households population (102,612), every district row and religion column exact. The 2022 report publishes no standalone no-religion line (folded into Other), and the map therefore shades only the religious-affiliation share, with the 17.2% residual (11.5% Missing) disclosed. Hindu affiliation concentrates on the plantation and outer-island geography (Cascade 20.1%, Other Islands 49.4%); Anglican strength peaks on Praslin (Grand Anse Praslin 21.4%). One census wave, no change metric. Census data © NBS, reproduced with attribution; boundaries OCHA COD-AB SYC ADM3 (CC BY-IGO).
Solomon Islands 2 waves · 2009–2019
Province census religious-affiliation overlays from the Solomon Islands National Statistics Office (SINSO), 2009–2019, across the ten provinces; two open province waves with exact margins (515,870 and 720,956). Affiliation is near-universal, and the substantive story is denominational: Isabel overwhelmingly Church of Melanesia, Choiseul and Western United Church-led, Guadalcanal Roman Catholic-led, Malaita split between Church of Melanesia and South Sea Evangelical, and Rennell-Bellona heavily Seventh Day Adventist and South Sea Evangelical; that composition rides the downloadable area summary verbatim. No SINSO reuse terms are stated; the derived summaries ship with attribution while a courtesy ask is outgoing, and the page reverts in a single commit if SINSO objects.
Spain 1 wave · 2025 survey estimate
Autonomous-community religious self-definition survey estimates (Centro de Investigaciones Sociológicas, CIS), Barómetro de Abril 2025 (Estudio nº 3505), a single wave on the 19 units (17 communities plus Ceuta and Melilla). Spain's census carries no religion question; every figure is a weighted (PESOCCAA) percentage with a wide 95% interval, never a count. Affiliated (practising and non-practising Catholics plus believers of another religion) and no religion (agnostics, indifferent/non-believers, atheists) sit either side of the self-definition item, with N.C. in the denominator and neither share. A distinct mass-attendance construct (monthly-or-more, weekly-or-more) rides the same wave on the asked believer subset, carried in the data and the overview. Under the PI small-cell ruling (2026-07-12) the unweighted respondent n is the wash denominator: self-definition washes five communities (Aragón, Asturias, La Rioja, Ceuta, Melilla). Single wave, no change metric (the CIS community design begins 2020). CIS open reuse via the datos.gob.es federation grant with citation (accepted); boundaries geoBoundaries ESP ADM1 (CC BY 4.0).
Sri Lanka 4 waves · 1981–2024
District census religion from the Department of Census and Statistics (DCS) across four censuses, 1981–2024, on the stable 25-district frame and a six-category religion frame (Buddhist, Hindu, Islam, Roman Catholic, Other Christian, Other) with no no-religion category; the map shades the reference-group Buddhist share and its exact complement, the Minority share, which is arithmetic on the published categories and not a no-religion measure. The 2001 census enumerated religion for only 18 of 25 districts; the seven northern and eastern districts render as no-data. DCS asserts all rights reserved with no open licence; the derived summaries ship with attribution while a reuse ask is outgoing, and the page reverts in a single commit if DCS objects.
Sweden 54 waves · 1972–2025membership register
National Church of Sweden administrative membership, 1972–2025; a single-denomination register construct measuring neither census affiliation, belief, practice, nor attendance, and a non-member supports no no-religion claim. Membership fell from 95.2% in 1972 to 50.7% in 2025; since 1996 no child enters membership automatically at birth. The source printed no metrics for 1981 (no population data were available that year), preserved as no data. No reuse terms are stated for the church series; used with attribution pending confirmation. Boundaries geoBoundaries SWE ADM0 (2019, Open Government Licence v3.0).
Taiwan 5 waves · 2020–2024register
County/city administrative register of registered temples and churches and reported temple followers (Ministry of the Interior), 2020–2024; a register construct, never census affiliation, belief, or attendance. The map shows reported temple followers and their year-on-year change; from 2014 the follower column counts temple followers only.
Thailand 1 wave · 2010
Province census religion (National Statistical Office of Thailand, NSO), 2010, on the 76-province 2010 census frame; nine verbatim categories, full-count. The 76 province totals sum to 65,981,653 against the WholeKingdom total 65,981,660 (−7), within the NSO's disclosed rounding residuals. The 2010 enumeration predates Bueng Kan province; the map ships the 76-unit frame with Bueng Kan unioned into Nong Khai, geometrically exact. Affiliation is above 99% nearly everywhere; the substantive story is compositional — in the deep-South provinces Islam is the majority declared religion (Pattani 84.4%) while Buddhism is the majority elsewhere — and it is preserved in the verbatim category counts. No open licence stated: NSO attribution, reuse confirmation pending. 2000 and 2020 held. Boundaries geoBoundaries THA ADM1 (ODbL).
Tokelau 3 waves · 2006–2016
Atoll census religious affiliation from the Tokelau National Statistics Office (TNSO) and Stats NZ, 2006, 2011 and 2016, for Atafu, Fakaofo and Nukunonu; CC BY 4.0. Affiliation is near-universal in every atoll and wave (Tokelau is effectively universally Christian, the no-religion count 0, 0 and 1 across the waves), and the near-flat headline is genuine; the substantive subnational story is denominational, Congregational Christian dominating Atafu and Fakaofo and Roman Catholic dominating Nukunonu. The 2022 census is licensed microdata only and pre-2006 waves publish no atoll religion table; both are held.
Trinidad and Tobago 1 wave · 2011
Municipality census religion (Central Statistical Office, CSO), 2011, from the Demographic Report Table 8: a count-valued cross-tab of the non-institutional population (1,322,546) across fifteen units, seventeen verbatim categories including Baptist-Spiritual Shouter and Orisha, never merged. The published table carries source-internal residuals (up to ±2 per margin, national total −1), rendered verbatim under the documented-discrepancy ruling. Affiliation runs 78.5% (Tunapuna/Piarco) to 90.8% (Penal/Debe), the gradient driven mainly by Not Stated; no religion peaks at 5.4% in Tobago. One census wave, no change metric. The CSO clause is permission-based; the summaries ship with attribution while a reuse ask is outstanding, and the page reverts in a single commit if the CSO objects. Boundaries OCHA COD-AB TTO ADM1 (CC BY-IGO).
Tuvalu 2 waves · 2012–2017
Region census religious affiliation (Central Statistics Division), 2012–2017, for Funafuti and the Outer Islands; near-universal affiliation makes the two headline shares vary little between regions, and the 2017 wave is a mini-census. The 2022 census reports religion nationally only; it is not mapped by region, and no change metric is offered.