God, Childhood and Inherited Belief: Evidence, Probability and the Cost of Certainty
The Blue•Print Research Unit
Research prepared by Grok under John Anthony Brown’s direction. Published at his direction.
Edition and source dates
- V6 source edition and research cutoff
- 3 November 2026
- Publication preparation
- 4 November 2026
- Online sources accessed
- 3 November 2026, unless otherwise noted
- Earlier AI drafts
- V1 generated at 5:01 am AEST on 3 November 2026. V2 generated at approximately 5:30 am AEST that day.
- Executive summary
- Scope, definitions and research method
- Part 1 — Define the claim before assessing it
- Part 2 — Probability without false precision
- Part 3 — Revelation, science and preventable suffering
- Part 4 — Child sexual abuse, moral authority and non-intervention
- Part 5 — Institutional wealth and claims of divine provision
- Part 6 — The geography of childhood belief
- Part 7 — What children risk when adults do not know the truth
- Table of major claims, evidence, strongest counterargument and remaining uncertainty
- What evidence could change the assessment
- Conclusions
- Limitations and unresolved questions
- Sources
1. Executive summary
This paper examines claims that an omnipotent, omniscient and perfectly loving god exists and intervenes in human affairs, particularly through revelation, moral instruction, protection of the vulnerable and authorised earthly representatives. The central challenge remains: if a god knew that children were being sexually abused by people presented as its representatives, had the power to stop it and loved those children perfectly, what explains its failure to intervene in a clear and reliably protective way? A theological possibility is not automatically an evidenced explanation.
What can responsibly be said is that no repeatable empirical sample exists from which a numerical frequency of god-existence can be measured. There is no objectively established empirical percentage for the existence of such a god. This is distinct from the claim that no one can responsibly express any probabilistic degree of belief. Bayesian reasoning shows how prior assumptions and the likelihood of observations shape posterior beliefs, but does not license arbitrary numerical priors presented as objective findings. Unknown does not equal 50/50. The absence of conclusive disproof is not positive evidence for the claim.
The geographic and familial patterning of childhood religious identity challenges claims of independent access to a uniquely revealed truth or of universal intuitive recognition of one deity. It does not by itself prove that any particular belief is false. The lack of specific, usable, pre-discovery scientific content in examined texts that meets tests of specificity and prior correct interpretation, and the documented scale of non-intervention in institutional child sexual abuse, are more readily accounted for by ordinary human cultural transmission and institutional dynamics than by reliable protective intervention by a perfectly loving, all-knowing, all-powerful god who authorises earthly representatives. Cultural transmission and institutional self-preservation can operate whether or not a god exists; they are not automatically mutually exclusive alternatives to every form of theism. The protective and revelatory claims examined here, however, generate expectations that the observations do not meet.
Evidence against one tightly specified claim does not automatically refute every conceivable supernatural hypothesis. Responses that quietly drop omnipotence, perfect love, intervention, or the reliability of claimed representatives change the original claim. The duty to protect children and to investigate documented institutional wrongdoing is not suspended by metaphysical uncertainty.
2. Scope, definitions and research method
This investigation prioritises official inquiries, primary texts, peer-reviewed research and transparent demographic data. It distinguishes findings, allegations, substantiated claims, convictions, estimates, doctrines and philosophical arguments. Secondary reporting is used for context and traced to primary sources where possible. Australian English and DD Month YYYY dates are used. Specialist terms are defined at first use.
Classical theism is the claim of a god that is omnipotent (able to do anything logically possible), omniscient (knowing all truths) and perfectly good or loving. An interventionist personal god is one that makes revelations, answers prayer and acts in human affairs. Particular religious claims add that a specified institution or group speaks for that god; institutional claims to represent a god are themselves claims, not established authorisation. Deism posits a non-interventionist creator. Limited gods lack one or more classical attributes. Non-literal interpretations treat religious language as symbolic. Evidence against one claim does not refute every alternative; equally, a claim cannot evade scrutiny by redefinition after contrary evidence appears.
The expectation that such a god would reliably prevent identifiable assaults by its claimed representatives is an additional premise, not automatic from classical theism alone. It is defended here on the grounds that perfect love, knowledge of the specific harm, power to intervene without eliminating free will generally, and claimed authorisation of representatives jointly generate a strong expectation of clear protective action in those cases. Not every classical theist or religious institution promises observable protection from every harm.
Research cutoff and access date: 3 November 2026. Full texts not independently re-checked in every case are marked. No personal biographical details of identifiable survivors are included. Sources actually opened (official web pages, published abstracts, PDF executive summaries where linked) are distinguished from those known through secondary description.
3. Part 1 — Define the claim before assessing it
The claim under primary scrutiny is that an omnipotent, omniscient and perfectly loving god exists, intervenes in human affairs on Earth, provides moral instruction and revelation, responds to people, and authorises earthly representatives who speak and act on its behalf. This is distinct from deism, limited gods, or purely symbolic readings. The additional premise that this god would reliably prevent identifiable assaults by those representatives is justified by the conjunction of perfect love (which cares about the child’s welfare), omniscience (which includes knowledge of the specific ongoing or imminent harm), omnipotence (which includes the power to intervene in limited ways, such as removing unsupervised access, without determining every human choice), and the claim of authorisation (which makes the representatives’ conduct relevant to the god’s moral standing).
Genesis 1:26–27 states that humans are created “in our image, after our likeness” and given dominion. Traditional interpretations do not claim that humans share the god’s omnipotence or omniscience. Functional or vocational views (supported by many contemporary biblical scholars such as J. Richard Middleton) emphasise the human role as representing divine rule (S6). Catholic teaching, as stated in the Catechism of the Catholic Church, paragraphs 355–364, affirms the whole person as the image, linking it to dignity and moral orientation, not to power or knowledge equal to the divine (S18). The mismatch is not a logical contradiction within the classical claim; it is a gap between popular teaching that may imply shared moral reliability or protection and the more careful theological reading that does not transfer divine attributes.
When evidence of non-intervention or of representatives harming children appears, responses that abandon perfect love, intervention, or the reliability of the claimed representatives have changed the original claim. Unfalsifiable appeals to unknown reasons move an initially observable expectation (protection of children by an all-loving god via its representatives) into a form that cannot be tested by any observation. Not every theological response necessarily abandons classical attributes; some attempt to preserve them while arguing that the goods of free will or soul-making outweigh the specific harms. Those arguments must be assessed on their evidence and internal coherence, not dismissed solely for being unpersuasive to the assessor.
4. Part 2 — Probability without false precision
Logical possibility is not empirical probability. A person’s degree of belief is not an established frequency. There is no obvious repeatable sample of universes or of divine interventions from which a numerical frequency of god-existence can be measured. There is therefore no objectively established empirical percentage. This does not entail that no one can responsibly express any probabilistic degree of belief; it does entail that manufactured numerical precision should not be presented as a research finding. Bayesian reasoning updates beliefs by combining a prior probability with the likelihood of the evidence under competing hypotheses, yielding a posterior. The result depends entirely on the prior and the likelihoods. Assigning an arbitrary prior and calling it objective is not warranted. Unknown does not mean 50/50; the number of conceivable explanations does not make them equally plausible.
Statements that observations are “more expected under” or “more consistent with” one hypothesis than another are conditional philosophical assessments unless a fully specified comparative model with quantified likelihoods is supplied. No such quantified model is offered here. Cultural transmission and institutional self-preservation can operate whether or not a god exists; they are not automatically mutually exclusive alternatives to theism. Dependent observations should not be double-counted as independent evidence. The protective-interventionist claim generates specific expectations (clear protective action against identifiable assaults by claimed representatives; specific usable scientific revelation that meets tests of specificity and prior correct interpretation). The observations examined do not meet those expectations as strongly as ordinary human explanations do. That is an assessment of relative fit, not a numerical posterior.
“No conclusive disproof” is not positive evidence. The observations of familial and geographic inheritance of mutually incompatible religious claims, the absence of specific advance scientific knowledge meeting clear tests in the examined corpus, and documented non-protective outcomes in institutional settings are more readily accounted for under humanly constructed traditions and ordinary cultural transmission than under a protective interventionist deity that reliably authorises and oversees earthly representatives.
Plain answer: This review does not establish an objective numerical probability for the existence of such a god. People may express subjective Bayesian degrees of belief, but those depend on stated assumptions and are not measured findings of this paper. Uncertainty does not imply equal probabilities. The evidence examined fits the protective and revelatory claims less well than rival accounts based on cultural transmission and institutional behaviour; this is a conditional philosophical assessment, not a measured likelihood ratio. Metaphysical existence remains undecidable by these data alone; the protective and revelatory claims as examined are not supported by the observations.
5. Part 3 — Revelation, science and preventable suffering
The corpus examined is the Hebrew Bible and New Testament as commonly translated in English (primarily the New Revised Standard Version and traditional renderings of key verses). Representative claims of scientific foreknowledge are tested against criteria of specificity, correct interpretation before discovery, independent predictive value, translation stability and capacity to fit multiple incompatible outcomes. No systematic review of every religious text was performed; the assessment is bounded to widely cited examples from this corpus.
Isaiah 40:22 (NRSVUE) states that God “sits above the circle of the earth.” Job 26:7 (NRSVUE) says God “hangs the earth upon nothing.” Leviticus 13–15 concerns examination and isolation of certain skin conditions and associated purification, and bodily discharges. Numbers 19:11–16 addresses ritual impurity after contact with a human corpse. These passages establish what the selected texts say in the NRSVUE translation. They do not, by themselves, establish claims about all historical interpretations, the first recognition of germ theory, historical disease outcomes, or specialist conclusions about Hebrew meaning beyond the translation used. Representative foreknowledge claims sometimes advanced for these passages (advance knowledge of Earth’s sphericity or of germ theory) are not supported by the texts as specific, pre-discovery predictive statements meeting the tests of specificity and prior correct interpretation stated in the methods. Selected passages do not amount to a systematic review of a whole corpus (S11, S12).
Within the selected passages examined, clear, usable, predictive statements of evolution by natural selection, the structure of the atom, or nuclear physics are not present. Later interpretations that map modern discoveries onto ancient verses are retrospective. Absence of a topic in the examined passages is not proof that a deity lacked the knowledge; given the claim that the texts are reliable moral and practical revelation from an omniscient communicator who cares about childhood suffering, the omission of clear, accessible information capable of preventing widespread preventable childhood death and disability carries evidential weight against that particular claim of revelation within the bounded corpus. Historical religious involvement in scholarship, hospitals and education demonstrates human capacity for learning and institution-building; it is not evidence of supernatural revelation of the relevant scientific content in the passages examined. Theological responses that the texts were never intended as scientific textbooks or that revelation is progressive narrow the original claim of reliable, protective revelation for the subjects tested.
6. Part 4 — Child sexual abuse, moral authority and non-intervention
The Royal Commission into Institutional Responses to Child Sexual Abuse (Australia) presented its final report on 15 December 2017. Of 6,875 survivors heard in private sessions to 31 May 2017, 4,029 (58.6 per cent) described abuse in religious institutions; 2,489 of those (61.8 per cent of the religious-institution survivors) described abuse in Catholic institutions. The survey of 75 Catholic Church authorities with priest members identified 9,025 priests who ministered in the period 1950–2010, of whom 507 were alleged perpetrators (unweighted proportion 5.6 per cent; weighted proportion 7.0 per cent accounting for duration of ministry). These figures appear in Final Report, Volume 16, Religious institutions, Book 2, unredacted edition, section 13.3.6, printed page 85. Claims were sought for 1 January 1980 to 28 February 2015 with no limit on the date of alleged incidents (section 13.3.1, printed page 75). There were 4,444 claimants and 4,756 claims (section 13.3.2, printed page 77). Some religious orders showed markedly higher proportions. The Commission concluded there were catastrophic failures of leadership of Catholic Church authorities over many decades, particularly before the 1990s: leaders knew of allegations, failed to report to police, transferred perpetrators, and prioritised institutional reputation. Similar patterns of knowledge without effective action appeared across multiple religious institutions. These are allegation and claim counts from surveyed authorities and private sessions, not population prevalence estimates or conviction counts. Underreporting is widely acknowledged (Volume 16 Book 2, S1).
The John Jay College study commissioned by the United States Conference of Catholic Bishops found that 4,392 priests and deacons (approximately 4 per cent of those in ministry 1950–2002) had allegations, with 10,667 individuals making allegations. The dataset is allegation-based and self-reported by dioceses; it is not a set of independently adjudicated criminal convictions (S2). Hunt et al. (2024), using data from the Australian Child Maltreatment Study (nationally representative survey of 8,503 people aged 16 and over), estimated that 0.4 per cent (95 per cent CI 0.3–0.6 per cent) of the population reported childhood sexual abuse by a leader or other adult in a religious organisation. Men reported higher rates (0.8 per cent) than women (0.1 per cent); 71.9 per cent of such reports occurred in Catholic organisations. Prevalence was higher in older cohorts. Limitations include reliance on retrospective self-report, the survey design of the parent ACMS study, and the fact that the estimate covers religious organisations rather than all institutional settings (S9). Not all male religious leaders, not all religious people and not all institutions are abusers or concealers. The relevant questions are conduct, authority, incentives and accountability.
Spiritual authority, clericalism, secrecy, obedience, institutional reputation and threats of supernatural consequences operated, where evidence supports them, to increase access to children and to impede disclosure and external reporting. Some religious people and institutions have supported reform, reporting and protective practices; such evidence neither erases nor cancels the documented harm.
The evidential problem is that an all-knowing witness who does not visibly defend the child or provide dependable testimony against the offender fails the expectation generated by the classical theist claim of perfect love plus intervention via authorised representatives. Free-will defences (e.g. Plantinga’s in The Nature of Necessity) are logical defences showing compatibility of moral evil with classical theism by arguing that significant free will is a great good that God cannot determine without eliminating the freedom. They are not, by themselves, accounts claimed to explain actual historical events. They do not explain why preventing a particular assault by a known abuser in a position of claimed divine authority would eliminate human freedom generally, or why an intervention that removes one person from unsupervised access to children would control every adult choice. Soul-making and afterlife-compensation responses do not show that later goods require the specific preventable harm of childhood sexual assault. Appeals to unknown divine reasons do not independently establish why these particular assaults were permitted. Where no testable implications or independent evidence are supplied, this paper cannot verify that explanation. That limitation does not itself demonstrate its logical impossibility. Human witnesses who can intervene without eliminating free will are expected to do so; the same expectation applies more strongly to an omnipotent, perfectly loving witness. The Stanford Encyclopedia of Philosophy entries on evil and theodicies map the debate and note the difficulties these responses face with horrendous evils (S7, S15, S16).
Institutional wrongdoing, theological explanation and the question of a deity’s existence are distinct. The connection is that the observed pattern is more readily accounted for under ordinary human institutional dynamics than under the protective-interventionist classical claim as examined. This is a conditional assessment of relative fit, not a numerical likelihood ratio.
7. Part 5 — Institutional wealth and claims of divine provision
Representative religious institutions obtain resources through donations, fees, investments, property, commercial activity, public funding and tax arrangements. In Australia, churches that register as charities with the Australian Charities and Not-for-profits Commission (ACNC) under the subtype of advancing religion, and obtain ATO endorsement, are income-tax exempt and may receive other concessions. Basic Religious Charities that meet six specific criteria are exempt from answering financial questions in the Annual Information Statement, from submitting annual financial reports, and from ACNC Governance Standards, but retain other obligations such as submitting an Annual Information Statement and notifying changes (S5, S17). These are ordinary legal and organisational mechanisms available to qualifying entities; the ACNC page establishes the rules, not a complete empirical account of any institution’s finances. Broader tax claims require verification against ATO guidance.
Dependence on human funding does not logically disprove a god. Appeals to supernatural provision or miraculous wealth are not supported by the observed financial behaviour, which is adequately explained by the ordinary operations of large human organisations that solicit donations, hold property and benefit from tax concessions. Charitable work, institutional assets, compensation obligations and the moral use of resources are distinct questions. The pattern is consistent with human institutions that claim divine backing while relying on human economic activity. No specific institutional income totals are asserted here beyond the existence of the legal framework.
8. Part 6 — The geography of childhood belief
Pew Research Center analyses distinguish two related but non-identical datasets. The June 2025 short read, synthesising surveys from 117 countries and territories, estimates that among adults under 55 raised in a religion, approximately 10 per cent have switched. Hindus and Muslims show approximately 99 per cent retention of broad affiliation category. The March 2025 report covers a 36-country survey of nearly 80,000 people and finds higher switching in East Asia, Western Europe and North America, with very high retention in India, Israel, Nigeria and Thailand. Both measure broad affiliation categories, not certainty, doctrinal agreement, sect retention, coercion or the truth of any religion. Most people who currently identify with a religion were raised in it (S3, S4).
Bengtson, Putney and Harris, drawing on the Longitudinal Study of Generations (more than 350 multigenerational families followed over decades), found that a child is more likely to remain within the parental religious fold than to leave it, and that emotional closeness, particularly with the father, is a stronger predictor than parental piety alone (S8). This establishes family transmission patterns in the studied sample. Children do not independently converge on one deity or doctrine regardless of birthplace.
A socially inherited belief may be true or false. Geographical variation challenges claims of independent access to a uniquely revealed truth, claims of universal intuitive recognition of one deity, and the reliability of particular belief-forming processes that would produce convergence independent of teaching. Broad religious retention does not establish why an individual believes, that everyone believes uniquely, or that switching is absent. Reported spontaneous religious experiences exist; they do not reliably identify one tradition over others, and contradictory experiences across traditions are evaluated by ordinary criteria of consistency with evidence and background knowledge. The observed distribution is explained by cultural transmission and differential access to alternatives. “Mythology” is used here in its academic sense of traditional narratives that explain origins and values within a culture; the term does not by itself adjudicate truth claims. Evidential standards are applied consistently across traditions.
9. Part 7 — What children risk when adults do not know the truth
Teaching about a family’s belief, sharing a tradition, and requiring a child to accept a proposition as unquestionable fact are distinct. Certainty claims can affect intellectual autonomy and the permission to ask questions; generate fear of punishment, damnation, exclusion or rejection; shape shame, bodily boundaries, consent and trust in authority; impede disclosure of abuse and access to independent help; and influence understanding of illness and psychological distress. These are proposed mechanisms and observed associations, not population estimates of harm, and not diagnoses of individuals.
Cole (2023) developed the Christian Teachings about Sin and Hell Scale and surveyed 683 former or current self-identified Christians. Individuals who had disaffiliated, were LGBTQ+, female, nonbinary, or had deconstructed their beliefs reported significantly increased overall distress related to teachings about sin and hell. This adult-sample study does not establish causal effects in children; it documents associations between exposure to certain teachings and self-reported distress in the sample (S13). Hansen (1998) examined long-term effects of religious rigidity and fear in a sample of 129 adults and found that high histories of fear predicted guilt and that rigidity predicted family strife. This adult-sample study does not establish causal effects in children (S14). Not every religious upbringing is coercive or harmful; not every non-religious upbringing is safe. Claims of benefits under non-coercive religious participation require suitable evidence that is not supplied in this paper. Correlation is not causation; testimony is not a population estimate. Hunt et al. is a prevalence study of abuse in religious organisations and cannot support the full range of psychological effects of religious upbringing.
Adults can responsibly say “this is what we believe, but we may be mistaken”, teach how claims are tested, permit age-appropriate disagreement, and keep protection of the child independent of obedience to spiritual authority. Concrete, trauma-informed principles include: separate the child’s safety and bodily autonomy from any spiritual claim; never threaten supernatural punishment for questioning or disclosure; model epistemic humility; provide access to independent information and support; and treat institutional loyalty as secondary to the child’s welfare. These are ethical recommendations grounded in the documented risks of certainty claims and the priority of child protection; they are not empirical findings about every upbringing.
10. Table of major claims, evidence, strongest counterargument and remaining uncertainty
| Claim | Evidence | Strongest counterargument | Remaining uncertainty |
|---|---|---|---|
| Protective interventionist classical theism via authorised representatives | Royal Commission (S1) and John Jay (S2) allegation/claim data on scale of non-intervention and institutional failure; Hunt et al. (S9) prevalence estimate; geographic inheritance of belief (S3, S4); absence of specific scientific revelation meeting tests in examined corpus (S11, S12) | A theist may argue that significant freedom or morally sufficient reasons make the classical attributes compatible with the existence of evil. A logical compatibility defence does not claim to explain each particular assault (S7, S15, S16). | Whether any specific reason could be independently evidenced; the figures are allegation and claim counts from surveyed authorities, not prevalence or convictions |
| Reliable advance scientific knowledge in core revelations (bounded corpus) | The selected NRSVUE passages do not explicitly state the modern scientific mechanisms discussed (S11, S12). No systematic historical interpretation review was conducted. | Texts were never intended as science textbooks; progressive revelation | Translation, historical interpretation and coverage limitations remain; the assessment is bounded to the examined passages and does not exhaust the corpus |
| Childhood religious identity tracks family and geography | Pew 2025 retention and switching data (S3, S4); Bengtson et al. longitudinal family study (S8) | Some conversion and spontaneous experience occur; inherited beliefs may still be true | Exact contribution of social learning versus other factors in each case |
| Institutional resources are humanly generated | Donations, fees, investments, ACNC/ATO tax concessions and Basic Religious Charity reporting rules (S5, S17) | Human means can be used by a divine agent | Whether any particular financial outcome is miraculous (none identified); complete institutional balance sheets not examined |
| Certainty claims can produce fear, shame and barriers to disclosure | Cole 2023 associations with distress from sin/hell teachings (S13); Hansen 1998 on fear and rigidity (S14); proposed mechanisms around authority and autonomy | Religious participation may also offer community and meaning; evidence for those benefits is not reviewed here. The cited studies do not establish harm in every religious upbringing. | Causation versus correlation; generalisability beyond studied samples; individual outcomes not diagnosed |
11. What evidence could change the assessment
Clear, contemporaneous, independently verified intervention that stops abuse by claimed representatives in multiple documented cases, or specific predictive scientific content in primary texts that was correctly interpreted and applied before discovery and that meets the specificity tests, would raise the likelihood of the protective-revelatory claim. Reliable independent convergence of children’s beliefs on one deity or doctrine across isolated cultures would challenge the cultural-transmission account. Transparent, falsifiable mechanisms linking claimed divine action to observed outcomes, rather than post-hoc unfalsifiable appeals, would allow stronger comparison. Absence of such evidence leaves the current assessment unchanged.
12. Conclusions
Empirical findings: childhood religious identity predominantly follows family, community and geography (Pew Research Center 2025; Bengtson et al.). Official inquiries document substantial numbers of allegations and claims of child sexual abuse in religious institutions, with systemic failures to protect, report and remove perpetrators (Royal Commission 2017; John Jay 2004; Hunt et al. 2024). This paper examines selected NRSVUE passages. It does not constitute an exhaustive review of the Hebrew Bible and New Testament. The selected passages do not explicitly state the modern scientific mechanisms discussed. Institutional resources are generated by ordinary human economic and legal mechanisms.
Philosophical reasoning: no numerical probability of god-existence can be derived from available data. The observations fit the protective and revelatory claims less well than accounts based on cultural transmission and institutional self-preservation. Logical defences and proposed theodicies can preserve the classical attributes in principle. They do not, by themselves, establish why a particular assault was permitted. This paper judges them insufficient to resolve its additional expectation of reliable protection of children; that is a philosophical assessment, not a demonstration that all attribute-preserving explanations are logically impossible. Cultural transmission is compatible with theism in principle but does not support the specific protective and revelatory claims under scrutiny.
Ethical judgements: the duty to protect children, to investigate documented institutional wrongdoing, and to avoid imposing untestable certainty on children is independent of metaphysical conclusions. Uncertainty about the existence of a god is not uncertainty about these duties.
Supported: the pattern of inherited belief and the scale of non-protective outcomes in institutions claiming divine authority. Weak: explanations that preserve all classical attributes while accounting for the non-intervention. Undecidable by these data: the bare existence of some form of deity that does not make the protective and revelatory claims examined.
13. Limitations and unresolved questions
Full primary texts of every volume of the Royal Commission and the complete John Jay datasets were not re-transcribed; key published findings were verified against official website pages, executive summaries and linked PDFs where available. The 9,025/507 figures (5.6% unweighted, 7% weighted) are located in Volume 16 Book 2, section 13.3.6, printed page 85. Underreporting rates remain estimates. Theological responses that appeal to unknown reasons cannot be fully ruled out by observation; they also cannot be confirmed. Cross-cultural developmental data are stronger for some regions than others. No global historical total of victims is asserted. Individual mental-health outcomes are not diagnosed. The science assessment is bounded to the examined corpus and representative claims; it is not an exhaustive survey of all religious texts. Future primary research on transmission mechanisms or new documentary evidence of intervention could alter weightings. Sources known only through abstracts or secondary description are marked as such in the accompanying sources.csv.
14. Sources
Sources not independently re-checked in full text are marked. No invented quotations or statistics are used. Advocacy sources were assessed against underlying official data where available. Complete source records with limitations appear in sources.csv.