Data & API

Freedom House expert assessment; higher scores mean more freedom, not a percentage of people or sites. Year denotes the edition, assessing June 1 of the previous year through May 31 of that year. Total is access (0–25) plus content (0–35) plus user rights (0–40). Country coverage is neither a random sample nor the entire world. History starts in 2019, when the source reversed its scale; the 2025 production process was reduced due to budget constraints. Detailed questions cover 2025 only; gaps remain missing. Question codes/scales follow the Freedom House methodology; different scales are not percentages.

Published numerical ratings, attributed to Freedom House. No open-data license asserted; publisher terms apply.

CSV 1

Metadata

Freedom House expert assessment; higher scores mean more freedom, not a percentage of people or sites. Year denotes the edition, assessing June 1 of the previous year through May 31 of that year. Total is access (0–25) plus content (0–35) plus user rights (0–40). Country coverage is neither a random sample nor the entire world. History starts in 2019, when the source reversed its scale; the 2025 production process was reduced due to budget constraints. Detailed questions cover 2025 only; gaps remain missing.

Published numerical ratings, attributed to Freedom House. No open-data license asserted; publisher terms apply.

CSV 1

Metadata

Cases documented by Access Now / #KeepItOn in the reporting calendar year. Ongoing restrictions count in each year they persist, not only when they begin. Country denotes the affected location, not necessarily the imposing actor. Detection coverage varies; missing country-years are not filled with zero. Counts do not measure affected people, disconnected hours or censorship severity. Subcategories overlap. A subcategory zero means no explicit positive flags among the country-year records; unknown attributes remain in the detailed data.

CC BY 4.0 — Access Now; underlying linked evidence retains its own terms

CSV 1 · CSV 2

Metadata

Cases documented by Access Now / #KeepItOn in the reporting calendar year. Ongoing restrictions count in each year they persist, not only when they begin. Country denotes the affected location, not necessarily the imposing actor. Detection coverage varies; missing country-years are not filled with zero. Counts do not measure affected people, disconnected hours or censorship severity. Subcategories overlap. A subcategory zero means no explicit positive flags among the country-year records; unknown attributes remain in the detailed data. One row per affected country/territory and annual source-sheet row. Multi-country cases repeat by affected location; the world count deduplicates record_id within each year. record_id and source_row are snapshot-specific and do not link events across years. Somaliland stays a region and New Caledonia a separate territory. Source dates, duration and status are retained without correction and are not converted to disconnected hours. Platform flags denote targeted service restrictions, not the loss of all apps in a full-network shutdown.

CC BY 4.0 — Access Now; underlying linked evidence retains its own terms

CSV 1

Metadata

Table A1 of Laeven–Valencia WP/26/94 (May 2026), 164 rows including borderline episodes. Events, not a complete country-year panel. Nicaragua 2018, Vietnam 2022 and Sri Lanka 2023 are explicitly borderline. Some country footnote numbers conflict with the legend; those classifications remain unresolved. Angola, Ghana and Republic of Congo dates differ between Tables 1 and A1; both are retained without selecting a true date. End dates marked 5/ are capped at five years, not proven actual resolutions. January 1 encodes a year, not a known start day. US and Swiss events in 2023 do not meet the authors’ systemic-crisis criteria and are not inserted.

IMF data terms: numerical data with attribution to Laeven and Valencia (2026)

CSV 1

Metadata

The same approved BIS archive as the existing annual real-price series. Coverage/breaks in dimensions, source flags in quality. Nominal and real domestic-currency indices, 2010 = 100. Real prices are CPI-deflated; rents, costs, mortgages and taxes excluded. Missing observations remain missing. Original retrieval timestamp was not recorded; source URL and verified SHA256 are retained.

BIS terms of use: reuse with attribution

CSV 1 · CSV 2 · CSV 3

Metadata

All published months since 1950; cutoffs vary by country. A normal, B methodology break, E estimate, P provisional. B stops drawdown tracking. Zero index observations remain as source_zero_index but are excluded from calculations; their cause is unknown. The euro-area aggregate is not treated as a country; gaps remain missing.

OECD data terms: reuse with attribution; original-provider rights retained

CSV 1 · CSV 2 · CSV 3 · CSV 4 · CSV 5 · CSV 6 · CSV 7

Metadata

CAPE divides price by trailing ten-year average real earnings; TR CAPE adjusts for payout changes and reinvestment. Excess CAPE yield compares inverse CAPE with a real long-term rate, not a promised return. Price is normally a daily-close monthly mean; the author interpolates earnings and dividends between quarterly or annual observations. High CAPE does not date a crash. The latest month may be incomplete. Author note: Oct price is Oct 2nd close | Sept/Oct CPI estimated | Oct GS10 is Oct 2nd value

Author-published numerical data with attribution; no explicit open redistribution licence stated. Original workbook remains private.

CSV 1 · CSV 2

Metadata

Direct Federal Reserve H.15, with retrieval date and version hash. Constant-maturity yields are percent per year, not bondholder profit. Negative real yields are valid. ND never becomes -9999 or zero. Spreads use paired months; annual spread means require 12 months. Inflation compensation includes risk and liquidity premiums. Inversion does not establish a crisis.

US government data, public domain; attribution to Federal Reserve Board / US Treasury

CSV 1 · CSV 2

Metadata

WDI/WFE/IMF definitions accompany indicators. Capitalisation is a stock, traded value a flow, turnover a ratio; none equals portfolio return. USD amounts are nominal. Missing observations remain missing. Capitalisation/GDP is not proof of a bubble. World aggregates are from WDI, never a sum of countries. India, 2025: WDI reports USD 10.56 trillion of capitalisation, while NSE Market Pulse January 2026 reports about USD 5.3 trillion for India. The reason is unresolved. That year’s capitalisation and capitalisation/GDP are retained in the typed dataset with quality=source_discrepancy but excluded from annual charts and automatic analysis pending review.

World Bank data terms: CC BY 4.0, with original-provider exceptions

CSV 1 · CSV 2 · CSV 3 · CSV 4

Metadata

Australia, 2021 Census: usual residents excluding overseas visitors. Denominations by birthplace and state; overlapping totals/subgroups are not additive. Secular Beliefs and Other Spiritual Beliefs and No Religious Affiliation includes secular beliefs, other spiritual beliefs and the literal No Religion, so described response. Inadequately described and Not stated remain separate. Random disclosure-control adjustments may cause small differences in totals. Affiliation does not measure practice.

CC BY 4.0; based on Australian Bureau of Statistics data

CSV 1

Metadata

Pew weighted adult survey responses, rounded to integer percentages; surveyed countries are not a representative sample of the world. Most observations are from 2024; US animal/nature/object spirits and afterlife are from 2023, while affiliation, importance, prayer, attendance and spirituality beyond nature use the 2023–2024 RLS. The table retains the full survey period and English response wording. Affiliation, beliefs and practice are different concepts. Do not multiply adult percentages by all-age population; 100 minus affiliation does not identify atheists. Thirteen missing cells remain missing: 12 questions not asked in Tunisia and one Bangladesh response excluded after a translation error. Rounded zero is not evidence of no people; small differences are not declared significant without uncertainty estimates. Russian labels are Macroscope’s and have not been reviewed or endorsed by Pew.

Pew Research Center terms of use; attributed numeric extracts, not a CC-licensed dataset; https://www.pewresearch.org/about/terms-and-conditions/

CSV 1

Metadata

Brazil preliminary 2022 Census sample results, residents aged 10 and older only. Major groups by sex and age; total age means all 10+, not all ages. Catholic, Evangelical, Spiritist, Umbanda/Candomblé, Indigenous traditions, other religions and no religion; don’t know and no declaration remain separate. Evangelical denominations are not available in this release. Published shares retain their age/sex universe, not an all-age population denominator.

IBGE public aggregate statistics; attribution to IBGE, SIDRA table 9537; no uniform CC licence asserted

CSV 1

Metadata

Canada, 2021 Census table 98-10-0353-01: weighted 25% sample estimates for people in private households, excluding collective dwellings. Religion by age and gender; totals overlap with subgroups. Christian includes its listed denominations; do not add them to the subtotal. No religion and secular perspectives is not synonymous with atheists. Nonresponse and collective-dwelling residents are not inferred as zero. Census rounding and disclosure rules apply.

Statistics Canada Open Licence; adapted from table 98-10-0353-01

CSV 1

Metadata

Historical all-age estimates for 2010 from Pew’s 2011 release, 232 countries/territories. Pew’s 2025 revision does not update this detailed split: do not splice into the revised Christian total series. Affiliation, not practice. Protestant includes Anglicans and independent churches; Orthodox includes Eastern and Oriental Orthodox churches. Other Christians include, among others, Latter-day Saints and Jehovah’s Witnesses under the source classification. Counts of at least 10,000 are rounded to 10,000; <1,000 and <10,000 remain bounds in the typed table only. The source states that <10,000 means 1,000–9,999, not zero. World is the sum of the five regions in the same release. Netherlands Antilles and Channel Islands are not assigned to a single present-day country. Geography follows the 2011 release; South Sudan is separate. Macroscope analysis. Pew Research Center bears no responsibility for the analyses or interpretations of the data presented here. The opinions expressed herein, including any implications for policy, are those of the author and not of Pew Research Center. Pew Research Center has published the original content in English but has not reviewed or approved this translation.

Pew Research Center terms of use: attributed numeric extracts and Macroscope calculations, not a mirror of reports or respondent data; https://www.pewresearch.org/about/terms-and-conditions/

CSV 1

Metadata

Pew estimates for all ages in 2010 and 2020 from the same revised 2025 release. Harmonised census/survey religious identification, not practising believers. Unaffiliated does not mean atheist or without beliefs. Other religions include folk traditions, Sikhs, Jains, Baha’is and others. The 2026 appendix displays shares below 0.1% as <0.1%; the interactive table displays populations below 10,000 as <10,000; these are retained as bounds in dimensions.operator and omitted from exact chart points. Other counts are rounded to 10,000 and shares to 0.1 percentage point. No intervening years are filled. Affiliated is a derived subtotal of the six religious groups and must not be added to them. Regions/world overlap with countries and must not be summed. Diversity ranks describe only the seven broad groups; a larger rank means less diversity, not less freedom.

Pew Research Center terms of use; attributed numeric extracts, not a CC-licensed dataset; https://www.pewresearch.org/about/terms-and-conditions/

CSV 1 · CSV 2

Metadata

England, Wales and their combined area are overlapping geographies, not the whole United Kingdom. All usual residents, voluntary question on March 21, 2021. No religion and Other religion include their nested responses; do not sum hierarchy levels. Atheist/Agnostic are explicit recorded responses, not everyone holding these beliefs. Religion not stated is distinct from No religion. Disclosure control may prevent exact reconciliation.

Open Government Licence v3.0; ONS / Nomis

CSV 1

Metadata

EVS 2017, ZA7500 v5.0.0: published weighted tables, fieldwork 2017–2021. Percentages use valid answers to each question; do-not-know and refusal responses are excluded. The gweight calibration adjusts age, sex, education and region. Rounded weighted valid/missing frequencies are retained; they are neither population counts nor actual interview counts. Q17 distinguishes a religious person, a non-religious person and a convinced atheist: self-description, not affiliation. Q20 “do not know what to think” is not relabelled agnosticism. Q13a describes only respondents naming a denomination, not all adults. Great Britain excludes Northern Ireland and is retained as a region. Chart year is the fieldwork end year; the full period is retained in the table. Rounded 0.0% is not an exact zero; blank cells remain absent. Small differences are not declared significant without uncertainty estimates. Source: EVS, GESIS (2022), EVS 2017 Variable Report, GESIS-Variable Reports 2022/05; EVS (2022), ZA7500 v5.0.0, doi:10.4232/1.13897. Only selected aggregates from the public report are republished, not microdata.

Published aggregate results with attribution; source microdata are not redistributed. GESIS terms: https://www.gesis.org/en/institute/data-usage-terms

CSV 1

Metadata

India, 2011 Census, all residents: broad religions and detailed Other religions and persuasions, by sex and urban/rural residence. The Appendix publishes detailed responses exceeding 100 nationally; totals and nested categories are not additive. Atheist is only a literal recorded response, not an estimate of all nonbelievers; Religion not stated remains separate. These are 2011 counts, not current estimates. Annexure Sunni/Shia figures are excluded because nearly all Muslims did not specify a denomination there; they cannot establish the full denominational distribution.

Census of India / ORGI public aggregate statistics; primary attribution; no uniform CC licence asserted

CSV 1

Metadata

Pew 2024 estimates of international migrant stocks at seven dates, 1990–2020. Incoming means migrants living in a country; Outgoing means people from that country living abroad. These are not annual flows or shares of all residents. Based on international migrant data and estimates of religious composition, not a census of practising believers. Incoming and Outgoing overlap; world totals must not be added to countries. Counts are rounded to 10,000; smaller counts remain <10,000 bounds and are omitted from exact chart points. Shares use migrants in the same direction as denominator, rounded to 0.1 percentage point; 0.0% does not establish absence. Shares for 22 Incoming records with zero total are undefined and omitted. Vatican City Incoming is missing. Unaffiliated does not mean atheist. This release is not spliced with the revised 2025 all-population estimates.

Pew Research Center terms of use; attributed numeric extracts, not a CC-licensed dataset; https://www.pewresearch.org/about/terms-and-conditions/

CSV 1 · CSV 2 · CSV 3 · CSV 4 · CSV 5 · CSV 6 · CSV 7 · CSV 8 · CSV 9 · CSV 10

Metadata

Published weighted responses among Muslim adults aged 18+, Q31 and the published Q32 subtotal; not shares of all residents or all-age population counts. Chart year is the fieldwork end year; exact dates are in the table. Fourteen African samples date to 2008–2009 and the remaining samples to 2011–2012. Sunni/Shia self-identification differs from Pew’s 2009 ethnographic classification: “just a Muslim” is not reassigned to either. Zero is a rounded survey answer, not proof of an absent community. Ahmadiyya, Alevi, Bektashi and Aliran Kepercayaan are volunteered (VOL.) responses. Sufi affiliation is a separate question and overlaps Sunni/Shia identities; do not add it to Q31. Thailand covers only five southern provinces and is a region, never a country chart point. Chad and DR Congo samples are disproportionately urban; some countries exclude inaccessible areas. Several major Muslim populations, including India, Iran and Saudi Arabia, were not surveyed. No world total can be inferred. See the source methodology for country-specific sampling exclusions and margins of error. Macroscope analysis. Pew Research Center bears no responsibility for the analyses or interpretations of the data presented here. The opinions expressed herein, including any implications for policy, are those of the author and not of Pew Research Center. Pew Research Center has published the original content in English but has not reviewed or approved this translation.

Pew Research Center terms of use: attributed numeric extracts and Macroscope calculations, not a mirror of reports or respondent data; https://www.pewresearch.org/about/terms-and-conditions/

CSV 1

Metadata

New Zealand 2013, 2018 and 2023 censuses: self-identified religious affiliation of the usually resident population; CEN23_ECI_017, national total, all ethnicities and genders. The 28 categories include totals and nonresponse. Up to four affiliations are counted per person; communities overlap and must not be added as independent groups. Christian nfd is an unspecified Christian response, not all Christians. No religion is broader than atheism. Object to answering and Not elsewhere included remain separate. Jedi in 2013 is _U.., omitted rather than zero. Fixed random rounding applies; components need not sum exactly. Our shares use count / published Total - religious affiliation × 100, with denominators 4,242,048 (2013), 4,699,755 (2018), 4,993,923 (2023): 2013 no religion is 38.55%. The published 41.9% uses the separately retained Total stated denominator of 3,901,164. 2018/2023 missing answers were supplemented by historic census responses and imputation: a zero residual does not imply complete response. In 2023, 84.4% came from current responses, 9.2% from historical censuses, and 6.5% from imputation (rounded shares). Changes in missing-data treatment affect comparability even with a common all-resident denominator. Some detailed categories were reclassified after 2018; population change is not individual switching. Public browser table text captured on 7 October 2026; the HTTP CSV download did not complete. This work is based on/includes Stats NZ’s data which are licensed by Stats NZ for reuse under the Creative Commons Attribution 4.0 International licence.

CC BY 4.0; this work is based on/includes Stats NZ data; https://www.stats.govt.nz/about-us/copyright/

CSV 1

Metadata

Northern Ireland usual residents, 2021. 113 leaf responses plus a total. Detailed names are published from 10 responses; smaller groups are pooled. Current religion, not religion or religion brought up in. Atheist/agnostic/humanist are literal answers, not exhaustive belief estimates. This geography is not the whole UK; Not stated is separate.

Open Government Licence; NISRA

CSV 1

Metadata

Northern Ireland published historical series, 1861–2021. None/Not Stated is combined and cannot be labelled atheist or unaffiliated. Historical geography/counting and the special 1981 treatment are in MS-B22 notes; modern totals can differ from MS-B21 due to disclosure protection. Spreadsheet proportions are converted from 0–1 to 0–100; no non-census years are filled.

Open Government Licence; NISRA

CSV 1

Metadata

Pew coding of published materials for 2023, 198 countries and territories. Scores from 0 to 10, not population percentages. GRI covers government restrictions; SHI covers hostile acts by private individuals and groups. They measure different things and must not be summed. North Korea is not scored due to insufficient reliable observations; missing is not no restrictions. Released June 15, 2026.

Pew Research Center terms of use; attributed numeric extracts, not a CC-licensed dataset; https://www.pewresearch.org/about/terms-and-conditions/

CSV 1

Metadata

Scotland published census percentages for 2001, 2011 and 2022, not the whole UK. Religious groups and No religion by age. Percentages are not converted into exact counts. Pagan became a separate 2022 response; the missing 2011 comparison is not zero. Not stated remains separate.

Open Government Licence v3.0; National Records of Scotland

CSV 1

Metadata

Historical 2009 expert/ethnographic estimates, not a current census or survey responses. Denominator: all Muslims in the respective country, not its total population. Ranges are the source’s likely estimate ranges, not statistical confidence intervals; Pew does not quantify a margin of error. Lower and upper range endpoints have separate indicators; no midpoint is calculated. Approximate (~) estimates are separate; <1%, <5% and similar are bounds in the typed table only. Missing -- is not zero. The remainder is not automatically called Sunni, which would conceal other traditions. Share of the world’s Shia and Shia share of a country’s Muslims have distinct denominators. Macroscope analysis. Pew Research Center bears no responsibility for the analyses or interpretations of the data presented here. The opinions expressed herein, including any implications for policy, are those of the author and not of Pew Research Center. Pew Research Center has published the original content in English but has not reviewed or approved this translation.

Pew Research Center terms of use: attributed numeric extracts and Macroscope calculations, not a mirror of reports or respondent data; https://www.pewresearch.org/about/terms-and-conditions/

CSV 1

Metadata

Percentages of ALL adults in each country, surveyed in 2024 (US 2023–2024), published April 23, 2026. Raised, remained, left and joined compare childhood with current identity, not annual flows. Categories overlap: retained affiliation is included in raised. Rounded numbers must not be added to reconstruct current Catholic/Protestant shares. Rounded zero does not mean no people; differences are not declared statistically significant.

Pew Research Center terms of use; attributed numeric extracts, not a CC-licensed dataset; https://www.pewresearch.org/about/terms-and-conditions/

CSV 1

Metadata

WVS7 published tables of Joint EVS/WVS v5.0.0, 66 country/territory samples, 2017–2023. Separate EVS samples are excluded to avoid duplication. Percentages use the full weighted sample, including do-not-know, refusal and other missing responses in the denominator; this differs from EVS valid percentages. gwght retains the original national WVS weights, not population-size country weights. The weighted TOTAL is retained and is not the actual interview count. F034/Q173 distinguishes a religious person, a non-religious person and an atheist; the original WVS response does not say “convinced” although the joint harmonized heading does. Affiliation, self-description and belief are distinct questions; there is no separate agnostic category. The year is the survey reference year from the year variable, not the publication year; India is 2023. Great Britain and Northern Ireland remain separate regions. Published dashes remain unprinted cells, not exact zeros. No confidence intervals or representative global aggregate are available in these tables. Source: EVS/WVS (2024), Joint EVS/WVS 2017–2022 Dataset v5.0.0, doi:10.4232/1.14320; tables dated 2024-06-30; WVS7 v6.0, doi:10.14281/18241.24. Selected published aggregates for non-commercial analysis; original microdata are not redistributed.

Published aggregate results with attribution; source microdata are not redistributed. GESIS terms: https://www.gesis.org/en/institute/data-usage-terms; WVS non-commercial publication terms: https://www.worldvaluessurvey.org/AJDocumentationSmpl.jsp?SAID=0&CndWAVE=7

CSV 1

Metadata

ECB SPACE 2024, chart tables 41–42, euro-area adults aged 18+. Ownership uses all adults; purposes use crypto owners only. The ownership question became standalone in 2024, potentially affecting responses and limiting comparisons with 2022. 2024 fieldwork: September–November 2023 and April–June 2024; Germany and Netherlands use national surveys. Croatia has no 2022 observation: its Excel zero is excluded. Euro area is a region, not the world.

ECB statistics; attribution, survey methods and source notes apply: https://www.ecb.europa.eu/services/disclaimer/html/index.en.html

CSV 1

Metadata

FCA / YouGov Cryptoassets consumer research 2025, page 24, figure 16: current ownership among UK adults aged 18+, not any use over a year. Online panel; general-population sample 2,199 in 2024 and 2,353 in 2025. 2025 fieldwork: 5 August–2 September. Offline population is not represented. Verified published points only, no interpolation.

FCA / YouGov published aggregate facts; attribution; report copyright retained by FCA

CSV 1

Metadata

Federal Reserve SHED table 33, all US adults aged 18+. Use during the preceding 12 months, not ownership at the survey date. Online survey; responses overlap and must not be summed. Not an all-age estimate. Year is the survey year, not publication year.

Federal Reserve public report; cite SHED, table 33 and survey year

CSV 1

Metadata

Each metric retains its own latest complete day; dates can differ. This is not a live price feed. A selection of 30 assets, not the whole market. Complete months: price, capitalization and supply at month-end; volume, transactions and fees summed; daily addresses and hash rate averaged over every day. Gaps remain missing. Units, metric definitions and day counts are retained. Exchange turnover is not on-chain transfer volume; addresses are not people. CC BY-NC 4.0 does not grant commercial reuse.

Coin Metrics Community: CC BY-NC 4.0; https://creativecommons.org/licenses/by-nc/4.0/

CSV 1

Metadata

A selection of 30 assets, not the whole market. Complete months: price, capitalization and supply at month-end; volume, transactions and fees summed; daily addresses and hash rate averaged over every day. Gaps remain missing. Units, metric definitions and day counts are retained. Exchange turnover is not on-chain transfer volume; addresses are not people. CC BY-NC 4.0 does not grant commercial reuse.

Coin Metrics Community: CC BY-NC 4.0; https://creativecommons.org/licenses/by-nc/4.0/

CSV 1 · CSV 2 · CSV 3

Metadata

Only the named Curve pool on Ethereum, not the whole protocol or DeFi market. Volume and trader-paid fees sum daily USD estimates from the primary Curve API, requiring every day and prices for every token. Fees are not DAO profit. Macroscope computes reserve value as the sum of token balances times their USD prices in the Curve snapshot; stablecoins are not assumed to equal one dollar. The last calendar day snapshot retains its source timestamp and block, not necessarily the final second of the year. LP tokens and underlying assets must not be added without checking double counting. Zero USD turnover without prices is excluded; native balances remain available. Missing observations remain missing.

Primary Curve public API observations; attribution to Curve. Macroscope calculations; no blanket third-party data licence asserted.

CSV 1

Metadata

Only the named Curve pool on Ethereum, not the whole protocol or DeFi market. Volume and trader-paid fees sum daily USD estimates from the primary Curve API, requiring every day and prices for every token. Fees are not DAO profit. Macroscope computes reserve value as the sum of token balances times their USD prices in the Curve snapshot; stablecoins are not assumed to equal one dollar. The last calendar day snapshot retains its source timestamp and block, not necessarily the final second of the year. LP tokens and underlying assets must not be added without checking double counting. Zero USD turnover without prices is excluded; native balances remain available. Missing observations remain missing.

Primary Curve public API observations; attribution to Curve. Macroscope calculations; no blanket third-party data licence asserted.

CSV 1

Metadata

Each series has its own latest complete day. Own growthepie data, CC BY 4.0. A named chain, not the whole crypto economy. Transactions and fees are summed only over complete periods; active addresses are a mean of daily values, not unique people or annual unique addresses. Bots and multiple wallets per person are not removed. Stablecoins are the USD value of supply on the chain at period-end, not payment volume. Missing values remain missing; cross-chain address counts do not give unique users. Upstream-provider metrics available through the same API are excluded.

growthepie: CC BY 4.0; https://www.growthepie.com/data-terms

CSV 1

Metadata

Own growthepie data, CC BY 4.0. A named chain, not the whole crypto economy. Transactions and fees are summed only over complete periods; active addresses are a mean of daily values, not unique people or annual unique addresses. Bots and multiple wallets per person are not removed. Stablecoins are the USD value of supply on the chain at period-end, not payment volume. Missing values remain missing; cross-chain address counts do not give unique users. Upstream-provider metrics available through the same API are excluded.

growthepie: CC BY 4.0; https://www.growthepie.com/data-terms

CSV 1

Metadata

One Uniswap v2 USDC/WETH pool on Ethereum. Macroscope calculations from contract state through Ethereum JSON-RPC (dRPC). Reserves are getReserves at the final block before month-end, verified against adjacent block timestamps. Token addresses and decimals are checked by contract calls. Each reserve retains native USDC or WETH units; USDC is not assumed equal to USD. This is not the entire protocol, total DeFi TVL, turnover or a yield estimate. Year-end observations start in 2020; monthly observations start in 2025. Swap history is not published: tested free RPCs did not supply a reliable complete period without additional keys or quotas.

Macroscope aggregates of public Ethereum contract state, read through dRPC; not a third-party market database.

CSV 1

Metadata

• msw_per_capita — Municipal waste per person, kg per person a year · CSV
• msw_landfill_share — Municipal waste landfilled, % of treated · CSV
• msw_recycling_share — Municipal waste recycled, % of treated · CSV
• msw_compost_share — Municipal waste composted, % of treated · CSV
• msw_incineration_share — Municipal waste incinerated, % of treated · CSV

• packaging_waste_per_capita — Packaging waste per person, kg per person a year · CSV
• plastic_packaging_per_capita — Plastic packaging waste per person, kg per person a year · CSV
• packaging_recycling_rate — Packaging recycling rate, % of packaging waste generated · CSV
• plastic_packaging_recycling_rate — Plastic packaging recycling rate, % of plastic packaging waste generated · CSV
• waste_all_per_capita — All waste per person (all activities and households), kg per person a year · CSV
• waste_nonmineral_per_capita — Waste excluding major mineral waste per person, kg per person a year · CSV
• household_waste_per_capita — Household waste per person, kg per person a year · CSV
• manufacturing_waste_per_capita — Manufacturing waste per person, kg per person a year · CSV

• industrial_production_index — Industrial production index, index, 2021 = 100 · CSV
• manufacturing_production_index — Manufacturing production index, index, 2021 = 100 · CSV

• industrial_production_oecd — Industrial production (OECD index), index, 2015 = 100 · CSV

• manufacturing_share_gdp — Manufacturing, share of GDP, % of GDP · CSV
• industry_share_gdp — Industry incl. construction, share of GDP, % of GDP · CSV
• manufacturing_value_per_capita — Manufacturing value added per person, constant 2015 US$ per person · CSV

• manufacturing_value_per_capita_unido — Manufacturing value added per person (UNIDO), constant 2020 US$ per person · CSV
• manufacturing_share_gdp_unido — Manufacturing, share of GDP (UNIDO), % of GDP · CSV

• oil_production — Oil production, million tonnes a year · CSV
• gas_production — Natural gas production, billion cubic metres a year · CSV
• coal_production — Coal production, million tonnes a year · CSV
• lithium_production — Lithium production, thousand tonnes of lithium a year · CSV
• rare_earths_production — Rare earths production, thousand tonnes of oxides a year · CSV

• crude_steel_production — Crude steel production, million tonnes a year · CSV

• cement_production — Cement production, thousand tonnes a year · CSV
• aluminium_production — Primary aluminium production, thousand tonnes a year · CSV
• copper_mine_production — Copper mine production, thousand tonnes of copper a year · CSV

• ewaste_generated_per_capita — E-waste generated, kg per person a year · CSV
• ewaste_collected_per_capita — E-waste formally collected, kg per person a year · CSV
• ewaste_recycled_per_capita — E-waste recycled, kg per person a year · CSV

• msw_per_capita_waw — Municipal waste per person (World Bank), kg per person a year · CSV
• msw_uncollected_share — Municipal waste left uncollected, % of municipal waste generated · CSV
• msw_recycling_share_waw — Municipal waste sent to recycling, % of municipal waste · CSV

OECD model scenarios, not forecasts: baseline assumes no new policies; regional_action and global_ambition assume policy packages. mismanaged = dumped outside controlled landfills, incineration and recycling. 2019 is the model base year.

CC BY 4.0

CSV 1

Metadata

A World Bank model scenario (from GDP per person and population), not a forecast and not observations. World = sum of countries.

CC BY 4.0

CSV 1

Metadata

• rd_business_gdp — Business R&D, % of GDP, % of GDP · CSV
• rd_government_sector_gdp — Government-sector R&D, % of GDP, % of GDP · CSV
• rd_higher_education_gdp — University R&D, % of GDP, % of GDP · CSV
• rd_funded_by_business — R&D funded by business, % of GERD · CSV
• rd_funded_by_government — R&D funded by government, % of GERD · CSV
• rd_funded_from_abroad — R&D funded from abroad, % of GERD · CSV
• rd_spending_ppp — R&D spending, billion dollars, billion USD PPP, 2020 prices · CSV
• researchers_per_thousand_employed — Researchers per 1,000 employed, per 1,000 employed, FTE · CSV
• researchers_fte — Researchers, thousands, thousand full-time equivalents · CSV
• gbard_ppp — Government R&D budget, billion dollars, billion USD PPP, 2020 prices · CSV
• gbard_defence_share — Defence share of government R&D budget, % of GBARD · CSV
• gbard_space_share — Space share of civil government R&D budget, % of civil GBARD · CSV
• triadic_patent_families — Triadic patent families, families per year · CSV
• gbard_gdp — Government R&D budget, % of GDP, % of GDP · CSV

• rd_business_per_capita — Business R&D per inhabitant, euro per inhabitant · CSV
• researchers_headcount — Researchers, thousand people, thousand people · CSV

• women_researchers_share — Women among researchers, % of researchers · CSV

• patent_applications_by_origin — Patent applications by origin, worldwide, applications per year · CSV
• patent_applications_abroad — Patent applications abroad by origin, applications per year · CSV
• pct_applications — PCT international patent applications, applications per year · CSV

• nasa_outlays — NASA outlays, billion USD, current prices · CSV
• nasa_share_federal_outlays — NASA share of US federal outlays, % of federal outlays · CSV

• esa_contribution — Contribution to ESA programmes, million euro · CSV

• ev_car_sales_share — Electric cars: share of new car sales, % of new car sales · CSV
• ev_car_sales — Electric cars: new car sales, cars a year · CSV
• bev_car_sales — Battery-electric cars (BEV): sales, cars a year · CSV
• phev_car_sales — Plug-in hybrid cars (PHEV): sales, cars a year · CSV
• ev_car_stock — Electric cars on the road (stock), cars · CSV
• ev_car_stock_share — Electric cars: share of all cars on the road, % of the car stock · CSV
• ev_bus_sales_share — Electric buses: share of new bus sales, % of new bus sales · CSV
• ev_truck_sales_share — Electric trucks: share of new truck sales, % of new truck sales · CSV
• ev_public_chargers — Public EV charging points, charging points · CSV
• ev_public_chargers_fast — Fast public charging points (over 22 kW), charging points · CSV
• ev_public_chargers_slow — Slow public charging points (up to 22 kW), charging points · CSV
• evs_per_public_charger — Electric cars per public charging point, vehicles per point · CSV

• new_cars_bev_share — New cars: battery-electric (BEV), share, % of new car registrations · CSV
• new_cars_phev_share — New cars: plug-in hybrids (PHEV), share, % of new car registrations · CSV
• new_cars_hev_share — New cars: hybrids without a plug, share, % of new car registrations · CSV
• new_cars_petrol_share — New cars: petrol, excluding hybrids, share, % of new car registrations · CSV
• new_cars_diesel_share — New cars: diesel, excluding hybrids, share, % of new car registrations · CSV
• new_cars_bev_registrations — New battery-electric cars (BEV): registrations, cars a year · CSV
• car_stock_bev_share — All cars on the road: battery-electric (BEV), share, % of all passenger cars · CSV
• car_stock_phev_share — All cars on the road: plug-in hybrids (PHEV), share, % of all passenger cars · CSV
• car_stock_hev_share — All cars on the road: hybrids without a plug, share, % of all passenger cars · CSV

• bev_range_median_epa — Electric cars: median range of new models (EPA), km · CSV
• bev_range_max_epa — Electric cars: longest range (EPA), km · CSV
• bev_models_epa — Electric cars: configurations tested by EPA, configurations · CSV
• bev_consumption_median_epa — Electric cars: median electricity use (EPA), kWh per 100 km · CSV

• us_ev_charging_ports_public — US: public EV charging ports, ports · CSV
• us_ev_charging_stations_public — US: public EV charging stations, stations · CSV
• us_ev_charging_ports — US: EV charging ports (public and private), ports · CSV
• us_ev_charging_ports_dc_fast — US: DC fast charging ports, ports · CSV

• orbital_launches — Orbital launches, launches a year · CSV
• orbital_launch_failures — Failed orbital launches, launches a year · CSV
• orbital_launches_spacex — SpaceX orbital launches, launches a year · CSV
• payloads_to_orbit — Satellites and spacecraft put into orbit, payloads a year · CSV
• starlink_satellites_launched — Starlink satellites launched, satellites a year · CSV

AFDC Station Locator snapshot at the end of December, public and private non-residential chargers together (home chargers excluded). ev_charging_stations are locations, ev_charging_ports are ports (one car each): level1 and level2 are slow AC charging, dc_fast is DC fast charging. From 2022 temporarily unavailable ports are included. In a few AFDC rows the levels do not add up to the port total (e.g. Georgia 2016); the numbers are kept as published and flagged in quality.

US government data (DOE Alternative Fuels Data Center), public domain in the US (17 U.S.C. 105)

CSV 1

Metadata

One record is one EPA-tested configuration (make, model, model year, drive, motor), not sales. range_epa is the EPA label range (combined cycle) in km and miles; consumption_kwh_per_100km is electricity use including charging losses. Battery-electric cars only (BEV); plug-in hybrids are not included. EPA ≠ WLTP ≠ CLTC: Europe's and China's tests give the same car a longer range. Model year is not the calendar year.

US government data (EPA/DOE fueleconomy.gov), public domain in the US (17 U.S.C. 105)

CSV 1

Metadata

Numbers typed from the yearly "ESA budget <year>" infographics on esa.int (page link and image SHA256 in every row). budget_total is the whole ESA budget, including work for the EU, EUMETSAT and other partners; budget_esa_programmes covers ESA's own programmes: contributions of member states, associate members and Canada (contribution_esa_programmes) plus other income (other_income_esa_programmes). 2016-2019: partner money in million euro (budget_from_partners); 2020-2025: funding shares of the whole budget in percent (budget_funding_share). The 2026 total is not on the image and is not added. budget_status is the image's label (budget = adopted in January, final, adopted). A budget, not actual spending.

Numbers typed from ESA infographics published under the ESA Standard Licence (credit: ESA); the images are not redistributed

CSV 1

Metadata

Jonathan McDowell's General Catalog (GCAT). orbital_launches by country use the launch vehicle state (LVState); dimensions.outcome is success, failure or unknown; entity_type launch_operator is the GCAT agency code (SPX — SpaceX, CALT — China's CALT, etc.), named in dimensions.name. payloads_to_orbit are payloads put into orbit by successful launches, by owner state (SatState), which is not the rocket's country. GCAT codes outside the UN country list stay as they are: SU — the Soviet Union, I-ESA — ESA, I-ELDO — ELDO, etc. An orbital launch has code O or D (deep space); suborbital flights are excluded. The current, unfinished year is not included.

CC BY 4.0 (cite: McDowell, J., General Catalog of Artificial Space Objects, planet4589.org/space/gcat)

CSV 1

Metadata

Historical years only, no IEA scenarios. measure is the IEA parameter (EV sales — sales in the year, EV stock — stock at year end, EV sales share and EV stock share in %, EV charging points — public chargers, Electric LDVs per charging point). dimensions.mode is the vehicle type (Cars, Buses, Trucks, Vans, 2 and 3 wheelers, EVSE for chargers); powertrain is EV (BEV + PHEV), BEV, PHEV, FCEV (fuel cell) or the charger class: slow up to 22 kW, fast over 22 and below 150 kW, ultra-fast 150 kW and more (fast and ultra-fast do not overlap). Regions (European Union, Europe, Asia Pacific...) are entity_type region. IEA rounds values to two significant figures.

CC BY 4.0 (IEA Global EV Data Explorer)

CSV 1

Metadata

• pisa_reading — PISA: reading, mean score of 15-year-olds, PISA points · CSV
• pisa_math — PISA: mathematics, mean score of 15-year-olds, PISA points · CSV
• pisa_science — PISA: science, mean score of 15-year-olds, PISA points · CSV
• pisa_reading_low — PISA: reading, share of 15-year-olds below baseline, % of students · CSV
• pisa_math_low — PISA: mathematics, share of 15-year-olds below baseline, % of students · CSV
• pisa_science_low — PISA: science, share of 15-year-olds below baseline, % of students · CSV
• pisa_math_top — PISA: mathematics, share of top performers (Levels 5-6), % of students · CSV

• timss_math_4 — TIMSS: mathematics, grade 4, TIMSS score · CSV
• timss_math_8 — TIMSS: mathematics, grade 8, TIMSS score · CSV
• timss_math_8_low — TIMSS: grade 8 students below the Low benchmark in mathematics, % of students · CSV
• timss_science_4 — TIMSS: science, grade 4, TIMSS score · CSV
• timss_science_8 — TIMSS: science, grade 8, TIMSS score · CSV

• pirls_reading — PIRLS: reading, grade 4, PIRLS score · CSV
• pirls_reading_low — PIRLS: grade 4 students below the Low reading benchmark, % of students · CSV

• piaac_literacy — Adult literacy (PIAAC), PIAAC score points · CSV
• piaac_numeracy — Adult numeracy (PIAAC), PIAAC score points · CSV
• piaac_literacy_16_24 — Literacy of young adults aged 16-24 (PIAAC), PIAAC score points · CSV
• piaac_literacy_55_65 — Literacy of adults aged 55-65 (PIAAC), PIAAC score points · CSV
• piaac_numeracy_16_24 — Numeracy of young adults aged 16-24 (PIAAC), PIAAC score points · CSV
• piaac_numeracy_55_65 — Numeracy of adults aged 55-65 (PIAAC), PIAAC score points · CSV
• piaac_literacy_low — Adults with low literacy (PIAAC), % of adults 16-65 · CSV
• piaac_problem_solving — Adaptive problem solving of adults (PIAAC), PIAAC score points · CSV

• naep_ltt_reading_9 — Reading at age 9 in the US (NAEP Long-Term Trend), NAEP score points · CSV
• naep_ltt_reading_13 — Reading at age 13 in the US (NAEP Long-Term Trend), NAEP score points · CSV
• naep_ltt_reading_17 — Reading at age 17 in the US (NAEP Long-Term Trend), NAEP score points · CSV
• naep_ltt_math_9 — Mathematics at age 9 in the US (NAEP Long-Term Trend), NAEP score points · CSV
• naep_ltt_math_13 — Mathematics at age 13 in the US (NAEP Long-Term Trend), NAEP score points · CSV
• naep_ltt_math_17 — Mathematics at age 17 in the US (NAEP Long-Term Trend), NAEP score points · CSV
• naep_reading_4 — Reading in grade 4 in the US (NAEP), NAEP score points · CSV
• naep_reading_8 — Reading in grade 8 in the US (NAEP), NAEP score points · CSV
• naep_reading_12 — Reading in grade 12 in the US (NAEP), NAEP score points · CSV
• naep_math_4 — Mathematics in grade 4 in the US (NAEP), NAEP score points · CSV
• naep_math_8 — Mathematics in grade 8 in the US (NAEP), NAEP score points · CSV
• naep_math_12 — Mathematics in grade 12 in the US (NAEP), NAEP score points (0-300) · CSV

• hbsc_problem_social_media_15 — HBSC: problematic social media use, 15-year-olds, % of pupils · CSV
• hbsc_online_contact_15 — HBSC: continuous online contact with friends and others, 15-year-olds, % of pupils · CSV
• hbsc_sleep_difficulty_15 — HBSC: difficulty getting to sleep more than once a week, 15-year-olds, % of pupils · CSV
• hbsc_school_pressure_15 — HBSC: pressured by schoolwork, 15-year-olds, % of pupils · CSV
• hbsc_like_school_15 — HBSC: like school a lot, 15-year-olds, % of pupils · CSV

One row = one number of the HBSC Data Browser: country or HBSC member, wave (end year = period), age (11, 13, 15; 11-15 = all ages together in the affluence breakdown), sex. Girls and boys (value_basis=published) are whole percentages as in the charts; sig_gender: significant gender difference (P < 0.05), empty where HBSC does not show it. sex=all: for countries the prevalence for girls and boys combined that the browser sorts countries by (value_basis=combined_sort_value, not printed in charts); for the HBSC average (entity HBSC_AVG) the published average. affluence: low_20pct / high_20pct = the 20% least and most affluent by the HBSC Family Affluence Scale within country, sex and age; sig_affluence: significant difference. No confidence intervals or sample sizes in the browser; about 1500 pupils per age group. trend_comparable: first comparable wave (social media and cyberbullying from 2017/18; WHO-5 well-being and loneliness 2021/22 only). England, Scotland, Wales and the Flemish and French communities of Belgium are separate members (entity_type subnational, parent), not GBR or BEL; Greenland separately (GRL). Definitions, HBSC codes and HBSC notes on missing data by wave: measure_definitions. Cross-check with the text of the WHO 2021/22 reports (volumes 1, 6, 7) is in the loader.

HBSC Data Browser: no separate licence stated; required citation "Health Behaviour in School-aged Children study (2023), Data browser (findings from the 2021/22 international HBSC survey): https://data-browser.hbsc.org". The same figures are published in the Annex "Key data" of the WHO Regional Office for Europe HBSC international reports 2021/22 (volumes 1-7, 2023-2024), licence CC BY-NC-SA 3.0 IGO (https://creativecommons.org/licenses/by-nc-sa/3.0/igo/).

CSV 1

Metadata

One row = one NCES number. measure scale_score, stat: mean or p10, p25, p50, p75, p90 (the score below which 10, 25, ... 90% of students fall). study: NAEP LTT (Long-Term Trend, age 9, 13, 17) or NAEP (main, grade 4, 8, 12). Scale 0-500 except grade 12 mathematics from 2005 (framework 2005, 0-300); grade 12 mathematics 1990-2000 (framework 1990) is not comparable with it (trend_comparable=false). sample: R1 = LTT original format without accommodations (1971-2004), R2 = main NAEP without accommodations (1990-2000), R3 = accommodations permitted (LTT from 2004, main NAEP from 1996/1998). catalog_chain=true marks the catalog value; bridge=true marks years with both samples (LTT 2004; main NAEP 1996/1998 and 2000). se = standard error of the average from the Digest tables (LTT to 2023, main NAEP to 2022; empty for 2024, for the transition samples without accommodations and for percentiles, which NCES does not publish in open tables). sig_vs_previous and sig_vs_first are the NCES significance tests (NAEP Data Service, sigacrossyear, p < .05) against previous_year and first_year: higher, lower or equal (not significant); significant = the same as yes/no against the previous year. mode: paper / digital (main NAEP grades 4 and 8 from 2017, grade 12 from 2019; LTT always paper). LTT mathematics 1973 is an NCES extrapolation (extrapolated=true), average only. States (entity_type subnational, USA-XX) and national public schools (group USA_PUBLIC): grade 4 and 8 averages 2003-2024, no standard errors.

Public domain: NAEP results and the Digest of Education Statistics are works of the U.S. federal government (17 U.S.C. § 105), no permission needed. NCES asks for a citation: "U.S. Department of Education, Institute of Education Sciences, National Center for Education Statistics, National Assessment of Educational Progress (NAEP), <years> Reading and Mathematics Assessments" (NAEP Data Service API; Digest of Education Statistics tables 221.10, 222.10 (2022), 221.85, 222.85 (2023))

CSV 1

Metadata

Rows: mean_score (se = standard error), level_share, score_change and level_share_change (cycle 1 -> 2; se includes the linking error; significant = |d/se| > 1.96, matching the OECD markings), age_gap (25-34 minus 55-65 or 16-24 minus 55-65), age_gap_change. variant: original = as published for cycle 1; headline = 2023 headline values (including doorstep-only adults); trend = 2023 trend tables (excluding them). Cycle 1 problem solving (problem_solving_tre) and cycle 2 adaptive problem solving are not comparable. The two OECD averages cover different countries and are not a trend. Flanders, England, Northern Ireland are entity_type subnational. Russia: 2012 only, without Moscow. Year = data collection year (2012/2015/2017/2023).

2023 report and its StatLinks: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/); Skills Matter 2019 tables: © OECD 2019, reuse with acknowledgement of the OECD as source under the OECD Terms and Conditions (https://www.oecd.org/en/about/terms-conditions.html)

CSV 1

Metadata

Average scores (mean_score) with standard errors (dimensions.se), percentages reaching the 400/475/550/625 international benchmarks (pct_reaching_*, with se where IEA publishes it) and below 400 (pct_below_low = 100 − percent reaching 400). source_report is PIRLS 2021 (trend and results exhibits) or PIRLS 2016 (participants and cycles absent from the 2021 report). diff_vs: this cycle minus the given cycle with IEA significance (higher/lower/none). trend_comparable: the row belongs to the country's comparable series. as_published: the value printed in an earlier report when it differs (e.g. Azerbaijan 2016: 470 in the 2016 report, 472 in the 2021 trend table). timing_2021: when the 2021 test was taken - end of grade 4 as planned, delayed to the beginning of grade 5 (students 4+ months older; IEA reports these in a separate trend table) or one year later; mode: digitalPIRLS, paper, or paper bridge sample (United States); data collection period and average age are in sample_note. notes are IEA footnotes: 1/2/3 coverage, †/‡/≡ school participation below guidelines, ψ reservations (15-25% too low to estimate), Ж more than 25%, + / ± took the easier PIRLS Literacy version in 2016 (± both versions). Benchmarking participants (Moscow City, Canadian provinces, UAE emirates, ...), England, Northern Ireland and the Belgian communities are entity_type subnational. Russia: every cycle 2001-2021. Use under IEA terms (non-commercial, with attribution), not an open licence.

© IEA (International Association for the Evaluation of Educational Achievement). Not an open licence: TIMSS & PIRLS publications may be used for non-commercial, educational and research purposes with attribution ("SOURCE: IEA's Progress in International Reading Literacy Study – PIRLS 2021", publisher TIMSS & PIRLS International Study Center, Boston College); reproducing or redistributing the publications themselves needs written permission of IEA. Only aggregate published figures are used here, with attribution. https://timssandpirls.bc.edu/Copyright/index.html

CSV 1

Metadata

One row = one number of a published OECD table. For each country the trend tables of the latest report listing it are used (PISA 2022 → 2018 → 2015 → 2012 → 2009 Vol. V; report, source_table). Where that report marks a cycle m, the older value is kept with trend_comparable=no (e.g. Argentina, Kazakhstan, Malaysia 2015; Viet Nam 2018; Kazakhstan reading 2018). earlier_value is the same cycle in the earliest parsed report when it differs. trend_comparable: since_2000 (reading), since_2003 (mathematics), since_2006 (science); reading shares from 2009. 2015: switch to computer (mode) and a new scaling model. mean_change is the change from the report table to the cycle in period, with an SE that includes the link error; significant = |change| > 1.96·SE. sample_note: other collection year (PISA 2000+, 2009+), PISA for Development, OECD flags for missed sampling standards, 2022 coverage and response. OECD averages (entity_type group) have their own country set in each report (label). Subnational participants: Chinese provinces, Shanghai, Buenos Aires, Baku, Dubai, Ukrainian regions 2022.

PISA 2022 and 2015 Results: CC BY-NC-SA 3.0 IGO (https://creativecommons.org/licenses/by-nc-sa/3.0/igo/); PISA 2018, 2012, 2009: OECD Terms and Conditions (copying and excerpts allowed with acknowledgement of the OECD as source and copyright owner) — https://www.oecd.org/en/about/terms-conditions.html

CSV 1

Metadata

One row = one number of a published OECD table (report, source_table); no microdata. 2022 (PISA 2022 Results Vol. II): hours per day on digital devices for learning and leisure at school, before/after school and on weekends (ST326; mean from interval midpoints, "more than 7 hours" = 7.5), shares by interval, weekly totals; mean mathematics score by leisure and learning time on devices at school; score change per hour and between adjacent intervals before and after accounting for socio-economic profile (adjusted: before_escs / after_escs); distraction by devices in mathematics lessons (ST273); school policies (phone bans etc., principals' reports); habits (notifications, anxiety without a device, ST322); computers per student 2012-2022. 2012-2018: hours per week on the Internet outside school, at school and in total (21st-Century Readers; interval midpoints, ×5 weekdays and ×2 weekend days), home computer and Internet 2003-2018; 2012 and 2015: minutes per day on the Internet outside school (PISA 2015 Vol. III; lower bound of the interval) and scores of low to extreme users (more than 6 hours a day). The 2022 device-time questions are not comparable with the 2012-2018 Internet questions (trend_comparable). significant = |value| > 1.96·SE (the tables' bold is not stored as data); coverage = share of the population covered (†/‡). Score differences are associations, not effects: students with different screen time differ in other ways too. The ICT questionnaire is optional, so many countries are missing. Licence per report in the license field.

PISA 2022 Results Vol. II and PISA 2015 Results Vol. III: CC BY-NC-SA 3.0 IGO (https://creativecommons.org/licenses/by-nc-sa/3.0/igo/); 21st-Century Readers (OECD 2021): OECD Terms and Conditions (excerpts allowed with acknowledgement of the OECD as source) — https://www.oecd.org/en/about/terms-conditions.html

CSV 1

Metadata

Days from 16 February 2020 to 30 April 2022 with the national status "Closed due to COVID-19" (days_fully_closed) or "Partially open" (days_partially_open: some regions or grades closed), and their total. Academic breaks are not counted. One status per country and day, so "partially" ranges from one closed region to closed upper grades; it is not the share of pupils at home. Values: UNESCO "Duration of school closures by country" (2022); ISO3 codes from the daily UNESCO database (to 31 March 2022), which is also the cross-check: its days to 31 March never exceed the totals to 30 April and fall short by at most the 30 days of April.

UNESCO Institute for Statistics, open data with attribution (UIS terms: CC BY-SA 4.0, https://databrowser.uis.unesco.org/terms-and-conditions). Citation: UNESCO Institute for Statistics based on UNESCO map on school closures (https://en.unesco.org/covid19/educationresponse)

CSV 1

Metadata

Average scores (mean_score) with standard errors (dimensions.se) and standard deviation (sd, 2023 only), percentages reaching the 400/475/550/625 international benchmarks (pct_reaching_*, with se where IEA publishes it) and below 400 (pct_below_low = 100 − percent reaching 400). source_report is TIMSS 2023 (trend and results exhibits) or TIMSS 2019 (countries and cycles absent from the 2023 report: Russia, Croatia, Turkey grade 8 and others). diff_vs: this cycle minus the given cycle with IEA significance (higher/lower/none). trend_comparable: the row belongs to the country's comparable series (the latest report trend table containing it). as_published: the value printed in an earlier report when it differs. notes are IEA footnotes: 1/2/3 population coverage, †/‡/≡ school participation below guidelines (≡ not met), ψ reliability reservations (15-25% of students too low to estimate), Ж average not reliably measured (>25%). Benchmarking participants (Moscow, Canadian provinces, UAE emirates, Madrid, ...) are entity_type subnational; so are England, Northern Ireland and the Belgian communities. mode: test mode (eTIMSS/paperTIMSS in 2019, almost fully digital in 2023). Russia: TIMSS 1995-2019, not in 2023; Moscow City a separate participant in 2019. Use under IEA terms (non-commercial, with attribution), not an open licence.

© IEA (International Association for the Evaluation of Educational Achievement). Not an open licence: TIMSS & PIRLS publications may be used for non-commercial, educational and research purposes with attribution ("SOURCE: IEA's Trends in International Mathematics and Science Study – TIMSS 2023", publisher TIMSS & PIRLS International Study Center, Boston College); reproducing or redistributing the publications themselves needs written permission of IEA. Only aggregate published figures are used here, with attribution. https://timssandpirls.bc.edu/Copyright/index.html

CSV 1

Metadata

• nyfed_unemployment_recent_graduates — Unemployment: recent college graduates aged 22-27 (NY Fed), % of labour force · CSV
• nyfed_unemployment_young_no_degree — Unemployment: young workers aged 22-27 without a bachelor's degree (NY Fed), % of labour force · CSV
• nyfed_unemployment_college_graduates — Unemployment: all college graduates aged 22-65 (NY Fed), % of labour force · CSV
• nyfed_unemployment_all_workers — Unemployment: all workers aged 16-65 (NY Fed), % of labour force · CSV
• nyfed_underemployment_recent_graduates — Underemployment: recent college graduates aged 22-27 (NY Fed), % of employed graduates · CSV
• nyfed_underemployment_college_graduates — Underemployment: all college graduates aged 22-65 (NY Fed), % of employed graduates · CSV
• nyfed_wage_median_bachelor_only — Median wage of bachelor's graduates aged 22-27 (NY Fed), USD, 2025 prices · CSV
• nyfed_wage_median_high_school_only — Median wage of high-school graduates aged 22-27 (NY Fed), USD, 2025 prices · CSV

• bls_unemployment_25plus_less_than_high_school — Unemployment, aged 25 and over: less than a high school diploma (US, BLS), % of labour force · CSV
• bls_unemployment_25plus_high_school — Unemployment, aged 25 and over: high school graduates, no college (US, BLS), % of labour force · CSV
• bls_unemployment_25plus_some_college — Unemployment, aged 25 and over: some college or associate degree (US, BLS), % of labour force · CSV
• bls_unemployment_25plus_bachelor_or_higher — Unemployment, aged 25 and over: bachelor's degree and higher (US, BLS), % of labour force · CSV

• eu_unemployment_25_29_below_upper_secondary — Unemployment aged 25-29: below upper secondary (ISCED 0-2) (Eurostat), % of labour force · CSV
• eu_unemployment_25_29_upper_secondary — Unemployment aged 25-29: upper secondary and post-secondary non-tertiary (ISCED 3-4) (Eurostat), % of labour force · CSV
• eu_unemployment_25_29_tertiary — Unemployment aged 25-29: tertiary (ISCED 5-8) (Eurostat), % of labour force · CSV
• eu_unemployment_25_64_below_upper_secondary — Unemployment aged 25-64: below upper secondary (ISCED 0-2) (Eurostat), % of labour force · CSV
• eu_unemployment_25_64_upper_secondary — Unemployment aged 25-64: upper secondary and post-secondary non-tertiary (ISCED 3-4) (Eurostat), % of labour force · CSV
• eu_unemployment_25_64_tertiary — Unemployment aged 25-64: tertiary (ISCED 5-8) (Eurostat), % of labour force · CSV
• eu_employment_25_29_below_upper_secondary — Employment aged 25-29: below upper secondary (ISCED 0-2) (Eurostat), % of population · CSV
• eu_employment_25_29_upper_secondary — Employment aged 25-29: upper secondary and post-secondary non-tertiary (ISCED 3-4) (Eurostat), % of population · CSV
• eu_employment_25_29_tertiary — Employment aged 25-29: tertiary (ISCED 5-8) (Eurostat), % of population · CSV
• eu_employment_25_64_below_upper_secondary — Employment aged 25-64: below upper secondary (ISCED 0-2) (Eurostat), % of population · CSV
• eu_employment_25_64_upper_secondary — Employment aged 25-64: upper secondary and post-secondary non-tertiary (ISCED 3-4) (Eurostat), % of population · CSV
• eu_employment_25_64_tertiary — Employment aged 25-64: tertiary (ISCED 5-8) (Eurostat), % of population · CSV

• oecd_unemployment_25_34_below_upper_secondary — Unemployment aged 25-34: below upper secondary (ISCED 0-2) (OECD), % of labour force · CSV
• oecd_unemployment_25_34_upper_secondary — Unemployment aged 25-34: upper secondary or post-secondary non-tertiary (ISCED 3-4) (OECD), % of labour force · CSV
• oecd_unemployment_25_34_tertiary — Unemployment aged 25-34: tertiary (ISCED 5-8) (OECD), % of labour force · CSV
• oecd_employment_25_34_below_upper_secondary — Employment aged 25-34: below upper secondary (ISCED 0-2) (OECD), % of population · CSV
• oecd_employment_25_34_upper_secondary — Employment aged 25-34: upper secondary or post-secondary non-tertiary (ISCED 3-4) (OECD), % of population · CSV
• oecd_employment_25_34_tertiary — Employment aged 25-34: tertiary (ISCED 5-8) (OECD), % of population · CSV
• oecd_relative_earnings_25_64_tertiary — Earnings aged 25-64, tertiary (ISCED 5-8) relative to upper secondary (OECD), % (upper secondary = 100) · CSV
• oecd_relative_earnings_25_64_below_upper_secondary — Earnings aged 25-64, below upper secondary (ISCED 0-2) relative to upper secondary (OECD), % (upper secondary = 100) · CSV
• oecd_relative_earnings_25_34_tertiary — Earnings aged 25-34, tertiary (ISCED 5-8) relative to upper secondary (OECD), % (upper secondary = 100) · CSV

• ilo_unemployment_25plus_basic — Unemployment aged 25 and over: basic education (ILO), % of labour force · CSV
• ilo_unemployment_25plus_intermediate — Unemployment aged 25 and over: intermediate education (ILO), % of labour force · CSV
• ilo_unemployment_25plus_advanced — Unemployment aged 25 and over: advanced (tertiary) education (ILO), % of labour force · CSV
• ilo_unemployment_15_24_basic — Unemployment aged 15-24: basic education (ILO), % of labour force · CSV
• ilo_unemployment_15_24_advanced — Unemployment aged 15-24: advanced (tertiary) education (ILO), % of labour force · CSV
• ilo_employment_25plus_basic — Employment aged 25 and over: basic education (ILO), % of population · CSV
• ilo_employment_25plus_advanced — Employment aged 25 and over: advanced (tertiary) education (ILO), % of population · CSV

One row = the monthly unemployment rate (% of labour force) of people aged 25 and over at one level of education, seasonally adjusted, as published by BLS (series_id). attainment: less_than_high_school, high_school (no college), some_college (or associate degree), bachelor_or_higher. October 2025 is missing: there was no survey.

Public domain: BLS data are works of the U.S. federal government (17 U.S.C. § 105). BLS asks to cite "U.S. Bureau of Labor Statistics, Current Population Survey" and the series id.

CSV 1

Metadata

One row = one Eurostat number: unemployment_rate (% of the group's labour force) or employment_rate (% of the group's population), for a country or aggregate (EU27_2020, EA21 — entity_type group), age (Y15-24, Y20-24, Y25-29, Y25-64), sex and ISCED 2011 level (ED0-2, ED3_4, ED5-8, TOTAL). quality = Eurostat flag: b break in series, u low reliability (small sample), d definition differs. United Kingdom to 2019.

Eurostat: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) — reuse allowed with acknowledgement of the source; Commission Decision 2011/833/EU. Source: Eurostat, EU Labour Force Survey (lfsa_urgaed, lfsa_ergaed).

CSV 1 · CSV 2 · CSV 3 · CSV 4 · CSV 5

Metadata

One row = one ILO number from a national survey: unemployment_rate (% of the group's labour force) by sex, or employment_rate (% of the group's population, both sexes). age: 15-24, 25+, 25-34. attainment = ILO aggregate: less_than_basic, basic, intermediate, advanced, all. source_type: labour_force_survey, other_household_survey, population_census etc.; source_code = ILO survey code. quality: u unreliable (small sample), b break in series. Not ILO modelled estimates: countries appear only in survey years.

ILO: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) — ILOSTAT data may be reused with attribution, ILO Terms of Use (https://www.ilo.org/about-ilo/terms-use); cite "ILO, ILOSTAT database" and the indicator id.

CSV 1 · CSV 2 · CSV 3 · CSV 4 · CSV 5 · CSV 6 · CSV 7

Metadata

One row = one NY Fed number. group: recent_graduates = aged 22-27 with a bachelor's degree or higher; young_no_degree = aged 22-27 without a bachelor's degree (same age group, the difference is the degree); college_graduates = aged 22-65 with a degree; all_workers = everyone aged 16-65. Students excluded. unemployment_rate and underemployment_rate are monthly, seasonally adjusted, three-month moving average (time_basis month_3m_moving_average), from January 1990; estimated = October 2025, the bank's estimate. recession_nber = NBER recession month. Underemployment = share of employed graduates in jobs that typically do not require a degree (O*NET, 50% threshold). annual_wage_full_time = annual wage of full-time workers aged 22-27, stat p25/median/p75, bachelor's only or high school only, 2025 dollars. group by_major = outcomes by major, ACS 2024: unemployment and underemployment of graduates aged 22-27, median wage early (22-27) and mid-career (35-45), share with a graduate degree (25-65). These are national series: the bank publishes no New York City or State figures here.

© 2026 Federal Reserve Bank of New York. Content from the New York Fed subject to the Terms of Use at newyorkfed.org (https://www.newyorkfed.org/privacy/termsofuse): use, copying and redistribution allowed with this attribution; modified content must be labelled — calendar-year means of the monthly rates are computed by Macroscope, not published by the bank. Underlying microdata: U.S. Census Bureau and BLS, Current Population Survey; American Community Survey (IPUMS); O*NET (U.S. Department of Labor).

CSV 1

Metadata

One row = one OECD number: unemployment_rate (% of the group's labour force), employment_rate (% of the group's population) or relative_earnings (earnings as % of upper-secondary earnings = 100; earners: all, or full_time_full_year). age 25-34 or 25-64, sex, isced/attainment (0T2 below upper secondary, 3_4 upper secondary and post-secondary, 5T8 tertiary, 6T8 bachelor's or higher). se = OECD standard error where published: a difference is significant roughly when |a-b| > 2*sqrt(se_a^2+se_b^2). quality: b break in series, u low reliability, w includes another category. OECD (average), EU25, G20 are entity_type group.

OECD: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) — OECD data published from 1 July 2024 are open under the OECD Terms and Conditions (https://www.oecd.org/en/about/terms-conditions.html); cite "OECD, Education at a Glance Database, labour market outcomes (DSD_EAG_LSO_EA)".

CSV 1 · CSV 2 · CSV 3 · CSV 4

Metadata

Six disclosed quarters, Q2 2024–Q3 2025. Monetary values: million USD; employees: quarter-end count. Paid units follow Amazon’s definition and exclusions and are not orders; only growth and seller share are disclosed. AWS and service revenue are separate from online-store sales.

Attributed factual numeric extracts; linked disclosure, no republication

CSV 1

Metadata

Young Center estimates using its definition of Amish, adults and children. Totals and subgroups can overlap. Differences between estimates are not birth counts or forecasts.

Attributed factual numeric extracts; source methodology retained

CSV 1

Metadata

NASA/JPL CNEOS, U.S. Government sensors, API v1.2. UTC time of peak brightness. Total impact energy is an empirical estimate, not energy of a ground strike or damage. Optical energy converted from 10¹⁰ J into J. Altitude refers to peak brightness. Missing values are not zero; many fireballs are unreported, so catalogue record counts do not estimate real incidence or its trend. CNEOS does not independently verify parameters. Event-level uncertainty/provenance is limited; speeds are not used to infer new orbits or interstellar origins.

Public NASA/JPL scientific numerical data; attribution retained. SSD API fair-use policy: cached server-side, one API request at a time, no browser API embedding.

CSV 1

Metadata

Published mean hours per day, including people who did not do the activity: US civilian population aged 15+. Hierarchical categories overlap and must not all be added. S is excluded; Z denotes rounding near zero. This is not loneliness, and gaming includes more than computer games.

US government public-use statistics, attribution to BLS

CSV 1

Metadata

BJS December 31 observations, 2013–2023. Counts cover state/federal prison jurisdiction irrespective of where people are held; local jails are excluded. Rates count sentences longer than one year per 100,000 residents in the same group, using Census populations on January 1 of the following year. White/Black/AIAN/Asian exclude Hispanic; Asian includes NHOPI. Some demographic values are estimates. These are not crime probabilities or causal discrimination estimates; ages/composition are not standardised.

US Department of Justice public domain; https://bjs.ojp.gov/reuse-policy

CSV 1

Metadata

CPS ASEC estimates, not a census of every household. For duplicated years, the revised r row is selected. Early MS-2 points 1890–1940 are separate decennial censuses. Same-sex marriages are included from 2019. A Vintage population update affects 2025 annual comparisons. Household size differs from family size; living alone does not establish loneliness. HH-4 uses householder person weights, not HVS housing inventory weights.

US government public-domain statistics, attribution to US Census Bureau

CSV 1

Metadata

Archived 2020–2022 data. Monthly means require every day of the month; annual indicators require every day of the year. Policy indices do not measure compliance, policy quality or causal effects. Average combines vaccination-status groups following the Oxford methodology.

CC BY 4.0, archived policy dataset

CSV 1 · CSV 2 · CSV 3 · CSV 4

Metadata

Statcounter GlobalStats, CC BY-SA 3.0. CSVs reshaped to long format. 14 markets: worldwide and 13 selected countries, not complete country coverage. Referrals to network websites, not people, queries, downloads or app time. All provider columns retained, YANDEX and YANDEX RU separate. Complete months through September 2026; AI from April 2025. All-zero periods are excluded as no sample. Annual and monthly records are alternative aggregations and must not be added.

CC BY-SA 3.0 (https://creativecommons.org/licenses/by-sa/3.0/); attribution: Statcounter GlobalStats; CSV reshaped, source annual aggregates retained

CSV 1

Metadata

Cetic.br / NIC.br, ICT Households: Brazilian internet users aged 10+ (internet use in the last three months). National totals and separate groups, not all residents. Published integer percentages retained; sampling uncertainty is not imported here, so comparisons are descriptive and small differences are not claims of significance. Activities overlap. Education levels use Brazilian definitions, not years of schooling. Generative AI only in 2025; missing is not zero.

CC BY 4.0; attribution: NIC.br / Cetic.br, TIC Domicílios; selected table cells reshaped

CSV 1

Metadata

Mainland China excluding Hong Kong, Macao and Taiwan. Telephone survey, residents aged 6+; internet users used the internet in the previous six months. Activity percentages use internet users as denominator; generative AI penetration uses total population under CNNIC’s separate definition. Categories overlap. Hours per week are self-reported internet time, not social-media time or device measurements. Short video is not TikTok or Douyin audience. Online-video definition includes micro-dramas from December 2024. June snapshots remain separate; only December snapshots enter annual charts. Published rounding retained. Report 57 states a 25.2pp AI increase in prose, while its chart levels 17.7 → 42.8 imply 25.1pp; chart levels retained.

Attributed factual numeric excerpts; CNNIC retains report copyright. Reports are linked, not republished.

CSV 1

Metadata

Eurostat survey of individuals aged 16–74; percent of everyone in the stated group, not only internet users. Survey year. DSI 2.0 is a proxy based on reported activities in five areas, not an examination; at least basic overall skills require at least basic skills in all five areas. Education: ISCED 0–2, 3–4, 5–8. Other group characteristics are not controlled. AI use refers to the last three months; purposes can overlap. Source flags and missing values retained; groups and components must not be added.

Eurostat reuse policy; attribution: Eurostat; selected cells reshaped

CSV 1

Metadata

Eurostat survey of individuals aged 16–74; percent of everyone in the stated group, not only internet users. Survey year. DSI 2.0 is a proxy based on reported activities in five areas, not an examination; at least basic overall skills require at least basic skills in all five areas. Education: ISCED 0–2, 3–4, 5–8. Other group characteristics are not controlled. AI use refers to the last three months; purposes can overlap. Source flags and missing values retained; groups and components must not be added.

Eurostat reuse policy; attribution: Eurostat; selected cells reshaped

CSV 1 · CSV 2

Metadata

Snap DSA: average monthly active recipients in January–June 2026, provider estimates. Not daily users, population penetration or advertising reach. Source unit: recipients; displayed in millions. The annual CSV rounds to four decimals; the typed dataset retains the original count. The EU total is separately reported; country values are not added. Snap warns in its H1 2025 report that calculations change and figures are not intended for comparison between periods. This is one snapshot, with no growth rate or comparison to other services’ audiences.

Attributed factual DSA disclosure; Snap terms: https://www.snap.com/terms

CSV 1

Metadata

TikTok DSA report for January–June 2025 (15 April 2026 version), Annex H, pp.26–27. Average monthly active recipients in each of the 27 EU member states, rounded to 100,000 (expressed here in millions). A recipient may access from several countries: country values are not summed as unique people. Provider DSA estimate, not a survey, daily audience, population percentage or count of all residents. The annual chart point represents only H1 2025; other services and periods have their own definitions.

Attributed numerical excerpt from TikTok DSA Transparency Report January–June 2025, Annex H, pp.26–27; report copyright retained by TikTok.

CSV 1

Metadata

Denominator: all individuals aged 16–74, IND_TOTAL/PC_IND. Survey year, activities in the last three months. EU/euro-area aggregates remain separate from countries. Categories overlap and must not be added as people. Source flags b/e/u and others are retained. Missing is not zero. Old music/news definitions are not spliced into the newer codes. Finding goods and services does not measure all search queries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Denominator: all individuals aged 16–74, IND_TOTAL/PC_IND. Survey year, activities in the last three months. EU/euro-area aggregates remain separate from countries. Categories overlap and must not be added as people. Source flags b/e/u and others are retained. Missing is not zero. Old music/news definitions are not spliced into the newer codes. Finding goods and services does not measure all search queries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Denominator: all individuals aged 16–74, IND_TOTAL/PC_IND. Survey year, activities in the last three months. EU/euro-area aggregates remain separate from countries. Categories overlap and must not be added as people. Source flags b/e/u and others are retained. Missing is not zero. Old music/news definitions are not spliced into the newer codes. Finding goods and services does not measure all search queries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Statcounter GlobalStats, CC BY-SA 3.0. CSVs reshaped to long format. 14 markets: worldwide and 13 selected countries, not complete country coverage. Referrals to network websites, not people, queries, downloads or app time. All provider columns retained, YANDEX and YANDEX RU separate. Complete months through September 2026; AI from April 2025. All-zero periods are excluded as no sample. Annual and monthly records are alternative aggregations and must not be added.

CC BY-SA 3.0 (https://creativecommons.org/licenses/by-sa/3.0/); attribution: Statcounter GlobalStats; CSV reshaped, source annual aggregates retained

CSV 1 · CSV 2 · CSV 3 · CSV 4 · CSV 5 · CSV 6 · CSV 7 · CSV 8

Metadata

Tencent disclosures: combined MAU of Weixin and WeChat / mobile-device MAU of QQ. Millions of active accounts in the final month of the period under company definitions. Whole services, not China residents or unique people. Do not add services or divide by China’s population to calculate penetration. Platform entities are retained; no country/world indicators are created from these figures.

Attributed factual numeric excerpts from Tencent disclosures; no republication of the reports.

CSV 1

Metadata

Ofcom ©, Adults’ Media Use and Attitudes 2026. Survey of 7,533 UK adults aged 16+, September 29–November 28, 2025. Each estimate has its own denominator: online adults, social media users, search engine users, AI users or AI-aware adults; modules use subsamples. Self-reports except responses to stated tasks. Rounded estimates; intervals not imported and small differences are not significance claims. Three surveys merged in 2025, changing question order and routing: older waves are not spliced into a trend. Associations do not show causation. Limited numerical excerpt; the full report is not republished.

Ofcom copyright; limited attributed factual numerical excerpt; https://www.ofcom.org.uk/about-ofcom/our-website/copyright

CSV 1

Metadata

Ofcom ©, News Consumption in the UK 2026: continuous survey November 2025–March 2026, 5,397 UK adults aged 16+. News encountered in the last month, including incidental exposure. The 2026 annual label identifies the survey edition, not a full calendar year. Questionnaire, news definition and online/face-to-face mix changed: no trend spliced to older editions. Reach uses all respondents; trust uses users of news on each channel, with different samples. AI includes apps, summaries and overviews, not just chatbots. Channels overlap; rounded estimates; intervals not imported.

Ofcom copyright; limited attributed factual numerical excerpt; https://www.ofcom.org.uk/about-ofcom/our-website/copyright

CSV 1

Metadata

Ofcom, Online Nation 2025, Ipsos iris. UK internet users aged 18+; May of the stated year, not an annual average. Passive measurement on smartphones, tablets and computers, primarily personal devices; TV and most work devices excluded. Time is per internet user per day. Service reach is the share of online adults in a month or average day (separate measures). Hours and minutes converted to minutes. Small checked numerical excerpt, not a redistribution of the commercial panel. Messenger covers the app and messenger.com, excluding messages inside Facebook. Differences from CNNIC surveys prevent a cross-country ranking of these hours.

Limited attributed numerical excerpt: Ofcom copyright, Online Nation 2025, underlying estimates Ipsos iris. https://www.ofcom.org.uk/about-ofcom/our-website/copyright . Not an open licence to the Ipsos panel or its underlying records.

CSV 1

Metadata

NTIA/Census CPS: US internet users aged 15+, not 18+. Selected 2015–2023 waves; adultInternetUser denominator, not the whole population. NTIA harmonized the variables. Proportions and standard errors converted from 0–1 to percent and percentage points; groups and original descriptions retained. Missing stays missing; activities overlap. Calls include voice; publishing excludes personal photos/videos shared on social media.

US federal government statistical data; attribution: NTIA / U.S. Census Bureau

CSV 1

Metadata

Pew Research Center, Washington, D.C., Social Media Fact Sheet, November 20, 2025. Limited numerical excerpt: five platforms, four survey waves (2021/2023/2024/2025). US adults aged 18+ reporting ever using the platform; not MAU, daily users or advertising reach. Mode break: phone through 2021, web/mail in 2023, web/mail/phone in 2024–2025. Changes cannot all be attributed to behavior. Point dates are published survey labels, not single-day measurements. Pew Research Center has published the original content in English but has not reviewed or approved this translation.

Limited attributed factual excerpt, Pew Research Center terms (https://www.pewresearch.org/about/terms-and-conditions/); not an open-data license. Pew Research Center has published the original content in English but has not reviewed or approved this translation.

CSV 1

Metadata

Pew Research Center, Washington, D.C., How Teens Use and View AI, February 24, 2026. Survey September 25–October 9, 2025: 1,458 US teens aged 13–17 living with parents, and their parents. Denominators are all teens or all parents as stated, never silently restricted to AI-aware people or chatbot users. Rounded self-reports and opinions, not measured learning or AI effectiveness. Overall margin of error ±3.3pp for each sample; cell-specific uncertainty not imported. Purposes overlap. Pew Research Center has published the original content in English but has not reviewed or approved this translation.

Limited attributed factual excerpts; Pew Research Center terms: https://www.pewresearch.org/about/terms-and-conditions/ . Pew Research Center has published the original content in English but has not reviewed or approved this translation.

CSV 1

Metadata

England only, not the whole UK. April–March financial years. Local-authority reports, not a complete estimate of illegal waste. There is a methodology break from 2019/20; a continuous trend needs that qualification. Incidents and actions are different events; category sums overlap with totals. Each column unit is identified by dimensions.column and the table header.

Open Government Licence 3.0

CSV 1

Metadata

DCMS-sponsored institutions in England, not every museum or unique visitors. Monthly data from 2004 to June 2026; groups and sites overlap. Closures, refurbishment, exhibitions, the pandemic and coverage changes affect the series. Some estimated counts have decimals; they are preserved. Partial 2026 is not annualized; missing cells are not zeros.

Open Government Licence 3.0; credit DCMS

CSV 1 · CSV 2

Metadata

Epoch AI and dataset authors: estimates with confidence information, not complete corporate disclosures. The selected dataset is not a census of all models. Training-data size uses different units across domains and cannot be compared mechanically. Organisation country is not training location. Events use model publication dates; individual authors and annotations are not republished.

Creative Commons Attribution; credit Epoch AI and dataset authors

CSV 1 · CSV 2

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

Exact source filters and quality flags accompany every value. Regions and aggregates are not recorded as countries.

Eurostat reuse policy: reuse with attribution

CSV 1

Metadata

NASA Exoplanet Archive, NExScI, Caltech/IPAC under contract with NASA. PSCompPars composite values may be derived rather than measured; missing is not zero. Earth-radius/mass units and parsecs retained. Discovery-year counts reflect the current confirmed archive, not historical publication vintages or new planets forming.

Public scientific data; acknowledge NASA Exoplanet Archive, Caltech/IPAC under contract with NASA; https://exoplanetarchive.ipac.caltech.edu/docs/acknowledge.html

CSV 1 · CSV 2 · CSV 3 · CSV 4

Metadata

Estimated minimum cost and affordability of a healthy diet following FAO/World Bank methodology, not actual household spending. PPP international dollars differ from exchange-rate US dollars. Release version and indicator codes are retained; national currencies are not mixed with PPP values.

CC BY 4.0, FAO statistical database terms; credit FAO and original providers

CSV 1

Metadata

FAO using NASA GISS data: land excluding the ocean, with a 1951–1980 baseline. The source label Meteorological year is retained rather than silently changed to calendar year. Baseline standard deviation is not a confidence interval for each annual anomaly.

CC BY 4.0, FAO statistical database terms; credit FAO and original providers

CSV 1 · CSV 2 · CSV 3 · CSV 4

Metadata

Food balances measure food supply available for human consumption, not actual food eaten or its distribution within households. Meat and milk may use FAO primary-product equivalents. Aggregates overlap and must not all be added. Methods and nutrient conversion factors have been revised.

CC BY 4.0, FAO statistical database terms; credit FAO and original providers

CSV 1 · CSV 2 · CSV 3 · CSV 4 · CSV 5 · CSV 6 · CSV 7 · CSV 8 · CSV 9 · CSV 10 · CSV 11 · CSV 12 · CSV 13 · CSV 14 · CSV 15 · CSV 16 · CSV 17 · CSV 18

Metadata

Source periods are retained: multi-year windows are not inserted into annual series. Inequalities < and > are bounds rather than exact values; confidence bounds remain distinct from central estimates. FAO/WHO/UNICEF providers follow the FAOSTAT metadata.

CC BY 4.0, FAO statistical database terms; credit FAO and original providers

CSV 1 · CSV 2 · CSV 3 · CSV 4 · CSV 5 · CSV 6 · CSV 7 · CSV 8

Metadata

Area, not the market value of land. Thousand hectares are not hectares. Equipped for irrigation and actually irrigated are different measures. Organic certification depends on national coverage.

CC BY 4.0, FAO statistical database terms; credit FAO and original providers

CSV 1 · CSV 2 · CSV 3 · CSV 4

Metadata

Current index counts records, not animals, people, species or biodiversity trends. Duplicate events, observation effort, citizen-science coverage and digitized collections affect geography. Records lacking countries remain in the global total. No personal details or precise occurrence coordinates are republished.

Attributed aggregate numeric facts; individual occurrence-dataset licenses vary

CSV 1

Metadata

Public activity on GitHub, not all programmers or all code. Accounts, repositories and organisations are quarter-end stocks and include inactive entities. Economy follows predominant location, not citizenship. The source suppresses small groups: absent is not zero. Distinct developers may appear in multiple languages, licences and topics, so category totals are not a count of people. Economy relationships represent code pushes and pull requests, not pairs of developers.

CC0-1.0; credit GitHub Innovation Graph

CSV 1

Metadata

Public activity on GitHub, not all programmers or all code. Accounts, repositories and organisations are quarter-end stocks and include inactive entities. Economy follows predominant location, not citizenship. The source suppresses small groups: absent is not zero. Distinct developers may appear in multiple languages, licences and topics, so category totals are not a count of people. Economy relationships represent code pushes and pull requests, not pairs of developers.

CC0-1.0; credit GitHub Innovation Graph

CSV 1

Metadata

Public activity on GitHub, not all programmers or all code. Accounts, repositories and organisations are quarter-end stocks and include inactive entities. Economy follows predominant location, not citizenship. The source suppresses small groups: absent is not zero. Distinct developers may appear in multiple languages, licences and topics, so category totals are not a count of people. Economy relationships represent code pushes and pull requests, not pairs of developers.

CC0-1.0; credit GitHub Innovation Graph

CSV 1

Metadata

Public activity on GitHub, not all programmers or all code. Accounts, repositories and organisations are quarter-end stocks and include inactive entities. Economy follows predominant location, not citizenship. The source suppresses small groups: absent is not zero. Distinct developers may appear in multiple languages, licences and topics, so category totals are not a count of people. Economy relationships represent code pushes and pull requests, not pairs of developers.

CC0-1.0; credit GitHub Innovation Graph

CSV 1 · CSV 2 · CSV 3 · CSV 4 · CSV 5 · CSV 6 · CSV 7 · CSV 8 · CSV 9 · CSV 10 · CSV 11 · CSV 12

Metadata

Public activity on GitHub, not all programmers or all code. Accounts, repositories and organisations are quarter-end stocks and include inactive entities. Economy follows predominant location, not citizenship. The source suppresses small groups: absent is not zero. Distinct developers may appear in multiple languages, licences and topics, so category totals are not a count of people. Economy relationships represent code pushes and pull requests, not pairs of developers.

CC0-1.0; credit GitHub Innovation Graph

CSV 1

Metadata

Public activity on GitHub, not all programmers or all code. Accounts, repositories and organisations are quarter-end stocks and include inactive entities. Economy follows predominant location, not citizenship. The source suppresses small groups: absent is not zero. Distinct developers may appear in multiple languages, licences and topics, so category totals are not a count of people. Economy relationships represent code pushes and pull requests, not pairs of developers.

CC0-1.0; credit GitHub Innovation Graph

CSV 1 · CSV 2 · CSV 3 · CSV 4 · CSV 5 · CSV 6 · CSV 7 · CSV 8 · CSV 9 · CSV 10 · CSV 11 · CSV 12 · CSV 13 · CSV 14 · CSV 15 · CSV 16 · CSV 17 · CSV 18 · CSV 19 · CSV 20 · CSV 21 · CSV 22 · CSV 23 · CSV 24 · CSV 25 · CSV 26

Metadata

Public activity on GitHub, not all programmers or all code. Accounts, repositories and organisations are quarter-end stocks and include inactive entities. Economy follows predominant location, not citizenship. The source suppresses small groups: absent is not zero. Distinct developers may appear in multiple languages, licences and topics, so category totals are not a count of people. Economy relationships represent code pushes and pull requests, not pairs of developers.

CC0-1.0; credit GitHub Innovation Graph

CSV 1 · CSV 2 · CSV 3 · CSV 4

Metadata

Public activity on GitHub, not all programmers or all code. Accounts, repositories and organisations are quarter-end stocks and include inactive entities. Economy follows predominant location, not citizenship. The source suppresses small groups: absent is not zero. Distinct developers may appear in multiple languages, licences and topics, so category totals are not a count of people. Economy relationships represent code pushes and pull requests, not pairs of developers.

CC0-1.0; credit GitHub Innovation Graph

CSV 1

Metadata

Public activity on GitHub, not all programmers or all code. Accounts, repositories and organisations are quarter-end stocks and include inactive entities. Economy follows predominant location, not citizenship. The source suppresses small groups: absent is not zero. Distinct developers may appear in multiple languages, licences and topics, so category totals are not a count of people. Economy relationships represent code pushes and pull requests, not pairs of developers.

CC0-1.0; credit GitHub Innovation Graph

CSV 1

Metadata

Public activity on GitHub, not all programmers or all code. Accounts, repositories and organisations are quarter-end stocks and include inactive entities. Economy follows predominant location, not citizenship. The source suppresses small groups: absent is not zero. Distinct developers may appear in multiple languages, licences and topics, so category totals are not a count of people. Economy relationships represent code pushes and pull requests, not pairs of developers.

CC0-1.0; credit GitHub Innovation Graph

CSV 1 · CSV 2 · CSV 3 · CSV 4 · CSV 5 · CSV 6 · CSV 7 · CSV 8 · CSV 9 · CSV 10 · CSV 11 · CSV 12

Metadata

NOAA GML; Lan, Thoning and colleagues, with source-file and DOI credit. Means, seasonally adjusted CO2 and uncertainties are retained. ppm and ppb differ by a factor of 1,000; concentrations are not emissions. Source uncertainties are not presented as 95% intervals.

Freely available NOAA GML data; cite Lan, Thoning et al., source files and linked dataset DOIs

CSV 1

Metadata

GISTEMP Team; Lenssen et al. 2024, doi:10.1029/2023JD040179. Anomaly relative to 1951–1980, not absolute temperature or deviation from the preindustrial era. Months and full J-D means come from the source; partial years do not become annual observations. Retrospective revisions are possible.

Public scientific data; cite GISTEMP Team and Lenssen et al. 2024, https://doi.org/10.1029/2023JD040179

CSV 1

Metadata

Compact original indicators for educational research/commentary using published HFEA summaries; source workbooks are not republished. UK licensed clinics only; one clinic excluded. Cycle-start year; 1991 starts in August and is omitted from annual indicators. 2020–21 are preliminary; birth outcomes2024 are preliminary. Patients overlap across treatment types; cycles/patients/embryos/birth occurrences differ. Per-embryo success is not a patient probability. Funding follows clinic location, not residence. Fresh/frozen outcome populations differ; no causal comparisons.

HFEA permits attributed non-commercial educational research/commentary; broader publication/reuse requires permission. Source workbooks are not redistributed.

CSV 1

Metadata

Elhacham et al., Nature 588, 442–444 (2020), DOI 10.1038/s41586-020-3010-5; author files/code pinned to commit 5a0170a51d164c1cc98b232452e86feb1e4ee334. A teratonne is 10¹² tonnes. Human-made material stocks at year-end, not annual production. In-use total excludes separately retained waste. 1900–2015 are historical estimates; 2016–2037 are the 2020 study extrapolation, not 2026 observations or an updated forecast. Biomass is a global model estimate; irregular years are not interpolated. Dry excludes water; wet includes it using the authors’ dry-to-wet conversion assumptions. Biomass after 2017 is author extrapolation. Relative standard deviation retained in %, not a 95% interval. Crossings 2020±6 (dry) and 2037±10 (wet) are published model results sensitive to definitions and uncertainty, not precisely observed transition dates. We preserve source-table years, without reproducing the figure’s shifts or smoothing.

Author data/code repository MIT, Copyright (c) 2020 emilyelhacham; full notice /data/indicators/life/human_made_mass_LICENSE.txt; attribution Elhacham et al., Nature 588, 442–444 (2020), doi:10.1038/s41586-020-3010-5; upstream literature sources retained in author repository.

CSV 1

Metadata

Library financial years differ. Sum of nonnegative published/imputed values for the 50 states and DC; valid coverage and missing counts are retained. Negative administrative codes are excluded. Visits and loans are events, not unique people. Definitions of electronic materials and virtual programmes have changed. ATTEND denotes visits in the FY1992 codebook, Appendix A: https://www.imls.gov/sites/default/files/fy1992_pls_database_documentation.pdf

IMLS public-use data; attribution; no person-level files

CSV 1 · CSV 2 · CSV 3 · CSV 4 · CSV 5 · CSV 6 · CSV 7 · CSV 8 · CSV 9 · CSV 10

Metadata

Library financial years differ. Sum of nonnegative published/imputed values for the 50 states and DC; valid coverage and missing counts are retained. Negative administrative codes are excluded. Visits and loans are events, not unique people. Definitions of electronic materials and virtual programmes have changed. ATTEND denotes visits in the FY1992 codebook, Appendix A: https://www.imls.gov/sites/default/files/fy1992_pls_database_documentation.pdf

IMLS public-use data; attribution; no person-level files

CSV 1 · CSV 2

Metadata

USGS ComCat earthquake-type events with published magnitude ≥7, 1900-01-01 ≤ date ≤ 2026-10-01. Preferred catalogue magnitude, with scale type retained. Hypocentre depth, not damage depth; reported errors are preserved. Historical completeness, accuracy and networks vary. The catalogue does not prove an increase in true event incidence or damage; counts refer to the stated filter. The logarithmic 10^(1.5ΔM) relation is an approximate educational energy ratio, not measured energy per event, local shaking or mortality.

USGS-produced public-domain scientific data; contributing catalogue/network names retained; https://www.usgs.gov/faqs/are-usgs-reportspublications-copyrighted

CSV 1

Metadata

Live-birth registrations in Scotland, not the entire UK. Names with counts 1–2 are suppressed; an absent name is not zero. Shares use published names, not all births. Spellings and variants are separate; accents are combined under NRS rules. Inverse HHI=1/Σp² measures spelling diversity, not ethnicity. No individual records.

Crown copyright 2026, Open Government Licence (explicit archive metadata)

CSV 1

Metadata

Census 2021 usual residents, voluntary religious self-identification. England and Wales exclude Scotland and Northern Ireland. Not answered differs from no religion. Country, region and local-authority levels overlap and are not additive. Census disclosure-control perturbation may prevent exact reconciliation.

Open Government Licence 3.0; credit ONS / Nomis

CSV 1

Metadata

Numeric extraction from reviewed pages 12, 13 and 15. 9M is neither a full year nor a separate quarter. Historical buyers/frequency keep the published year labels: their reference window is not completed by assumption. Their product is not converted into order counts. GMV includes services. Pickup points and area were disclosed as >78 thousand and >4.0 million square metres: lower bounds rather than exact numbers. Frequency and buyers may use different reference windows.

Attributed factual numeric extracts; source PDF is linked, not republished

CSV 1

Metadata

Standard XBRL facts from 10-K/10-K/A: annual intervals of 300–400 days and balances at the end of those intervals. Latest filing for each exact tag/start/end; initial vintages are not mixed in. Different revenue tags stay separate and are not added. Financial year need not be calendar year. Amazon revenue includes AWS; it is not orders. Companies are not recorded as countries.

Attributed factual numeric extracts from public filings; no filing republication

CSV 1

Metadata

Fetterer et al. (2025), Sea Ice Index v4, DOI 10.7265/a98x-0f50. Northern/Southern Hemisphere are not countries. Extent counts whole grid cells with ice concentration ≥15%; area weights by concentration. Monthly satellite estimates in million km², not volume or land glaciers. Recent points may be preliminary. The published annual mean is calendar-day weighted and retained only with all 12 months present. Hemispheres are not summed into a universal climate index.

Public government-funded scientific data; required attribution: Fetterer et al. (2025), Sea Ice Index v4, doi:10.7265/a98x-0f50; NSIDC data/citation policies.

CSV 1

Metadata

Database records by notice year, not original article year. Retractions are distinct from corrections and expressions of concern. Country refers to affiliations; a record can belong to several countries and disciplines, which must not be added to a world total. This is not the share of misconduct or a ranking of research quality; database coverage and delays change.

CC0; attribution to Crossref and Retraction Watch

CSV 1 · CSV 2

Metadata

1 April 2026 release, news id 15, unaudited IFRS disclosure. RWB Group in Russia and other countries, rather than a single Russian marketplace. GMV is not revenue; > and ≤ remain bounds. Rounded values and source year labels retained. No absolute order count or platform-penetration measure is disclosed here.

Attributed factual numeric extracts; linked disclosure, no republication

CSV 1

Metadata

Official current online players for eight games, observed at request time. Not DAU/MAU, unique people, sales or the whole gaming audience; offline players are excluded. History starts with our own snapshots; --refresh appends a snapshot. Recurring collection has not been scheduled.

Limited attributed current-count snapshots; Steam Web API terms apply

CSV 1

Metadata

PDSA/YouGov estimates of animals, not owners or households. Two separate 2020 survey waves are retained; 2020 is not inserted into the annual series.

Attributed factual numeric estimates; no report republication

CSV 1

Metadata

Prices: BLS CPI average prices, US city average, not seasonally adjusted. Pay: CES0500000003 mean gross hourly earnings of all private-sector employees, seasonally adjusted. Not median, net or individual wages. Minutes = 60 × price / earnings in the same month. Annual means require all 12 months; partial 2026 and missing months are not filled.

US government public statistics; attribution to BLS

CSV 1

Metadata

Downloaded from ARDA; original collectors ASARB / U.S. Religion Census 2020. Congregations/adherents use participating groups’ definitions and are not all believers or residents. Totals and categories overlap. All numeric cells, including zeros, are retained for seven broad traditions; detailed groups include positive cells only, so absent detailed rows cannot establish zero or absence. No person-level records.

ARDA download terms; acknowledge ARDA and original collectors ASARB / U.S. Religion Census; attributed aggregate factual extracts

CSV 1 · CSV 2 · CSV 3 · CSV 4 · CSV 5 · CSV 6 · CSV 7 · CSV 8 · CSV 9 · CSV 10 · CSV 11 · CSV 12 · CSV 13 · CSV 14 · CSV 15 · CSV 16 · CSV 17 · CSV 18 · CSV 19 · CSV 20 · CSV 21 · CSV 22 · CSV 23 · CSV 24 · CSV 25 · CSV 26 · CSV 27

Metadata

US dollars per acre, not per hectare. Estimated value, not the price of every transaction. US total covers the 48 contiguous states, subject to source exclusions. Farm real estate includes buildings; cropland and pasture differ. D means undisclosed, not zero.

US government public statistics; credit USDA NASS

CSV 1

Metadata

Week by reporting date, not infection or death date. Missing is not zero; negative corrections are retained. Changes in testing and reporting limit comparisons. Cumulative values must not be added. Excluded source codes: PN, XK, XXF, XXG, XXH, XXI, XXJ, XXL

WHO COVID-19 dashboard data, CC BY 4.0

CSV 1 · CSV 2 · CSV 3 · CSV 4 · CSV 5 · CSV 6 · CSV 7 · CSV 8 · CSV 9 · CSV 10 · CSV 11 · CSV 12 · CSV 13 · CSV 14 · CSV 15 · CSV 16 · CSV 17 · CSV 18 · CSV 19 · CSV 20 · CSV 21 · CSV 22 · CSV 23 · CSV 24 · CSV 25 · CSV 26 · CSV 27 · CSV 28 · CSV 29 · CSV 30 · CSV 31 · CSV 32 · CSV 33 · CSV 34 · CSV 35 · CSV 36 · CSV 37

Metadata

Seven-day measures are retained; overlapping 28-day windows are not summed. National definitions and coverage differ; the US switched to a sentinel network in February 2024. Missing is not zero.

WHO COVID-19 dashboard data, CC BY 4.0

CSV 1 · CSV 2 · CSV 3 · CSV 4 · CSV 5

Metadata

Twelve selected English articles from July 2015. Agent=user excludes detected automated requests. Not unique readers, all languages or public opinion. Renames, redirects and related articles may change coverage. Attention forecasts require total-traffic normalization and additional sources.

CC0 1.0; credit Wikimedia Analytics API

CSV 1

Metadata