Documentary record Bound by Fear

Data bias

What this archive cannot see

This is a record of abuse that produced a court case, a conviction, or an official finding. That is a deliberately narrow standard, and it has a consequence we would rather state plainly than let you discover: the shape of this archive is determined as much by where abuse gets documented as by where abuse happens.

Nothing on this page is a hedge about the records themselves. Every record here is sourced, and the sourcing is published. The question this page answers is a different one — what is missing, and why.

The distribution

What the counts do not mean

Tradition Records Share
Christian institutions & contexts 794 60.2%
Islamic institutions and contexts 115 8.7%
Spiritual healing, divination, and mediumship 77 5.8%
New religious movements 69 5.2%
Buddhist institutions and contexts 42 3.2%
Jewish institutions and contexts 36 2.7%
Hindu and Hindu-derived contexts 31 2.4%
Nigerian juju 23 1.7%
Indigenous spiritual authority 16 1.2%
Cuban Santería 13 1.0%
Western occult and esoteric movements 13 1.0%
Afro-Brazilian religious settings 12 0.9%
Witchcraft- and sorcery-accusation violence 11 0.8%
False ritual-abuse accusation and miscarriage of justice 11 0.8%
Divination and fortune-telling services 10 0.8%
Southern African muti 9 0.7%
New Age and human-potential movements 9 0.7%
Sikh institutions and contexts 6 0.5%
Left-hand-path / Satanist / occult 6 0.5%
Haitian Vodou 5 0.4%
Cuban Palo Mayombe 4 0.3%
Fraternal orders 3 0.2%
Ásatrú, Heathen, and Norse-traditional settings 2 0.2%
Jain institutions and contexts 1 0.1%

Christian-context records make up 60% of this archive. This is not a finding that Christianity is more abusive than other traditions. It is a finding about inquiries.

Since the 1990s, majority-Christian countries have run large public investigations into religious abuse — the Australian Royal Commission, IICSA in England and Wales, the Ryan and Murphy reports in Ireland, the Scottish Child Abuse Inquiry, the Pennsylvania grand jury, Canada's Truth and Reconciliation Commission, France's CIASE. Each produced sworn testimony, named institutions and published findings. Each one manufactured documentation that would not otherwise exist.

No comparable inquiry has ever been held into religious abuse in most of the world. Where an inquiry happened, we have records. Where none did, we have silence — and silence in this archive never means safety.

The clearest example

Coverage tracks press freedom, not population

Islamic-context records are 115 of 1318 (8.7%). That coverage is not distributed by population.

Country Muslim pop. (m) Records Per 10m
Senegal 17 6 3.53
Malaysia 20 7 3.50
Tunisia 12 3 2.50
Morocco 37 9 2.43
Nigeria 110 17 1.55
Indonesia 242 35 1.45
Turkey 85 7 0.82
Bangladesh 150 9 0.60
Somalia 18 1 0.56
Egypt 102 3 0.29
Saudi Arabia 35 1 0.29
Pakistan 240 6 0.25
Afghanistan 41 1 0.24
Iran 87 2 0.23
Yemen 32 0 0.00
Syria 21 0 0.00

Population figures approximate, Pew Research Center. Record counts computed from this corpus at build time.

Yemen, Syria currently have no records at all — together roughly 53 million Muslims.

The countries at the top of that table are not the most religious. They are the ones with courts that publish and newspapers that can report on a cleric. Our most-used Islamic-context sources come from countries with a semi-free press. About 20% of our Islamic-context records come from Britain, North America, Western Europe and Australia, far out of proportion to where Muslims actually live. In those countries a madrasa abuse case reaches an open court and a published judgment. We are substantially documenting Muslims in countries with open courts, not Islam.

There is a doctrinal layer too, and we state it carefully because it is easy to abuse. In jurisdictions that prosecute sexual violence through the classical zina framework, a complainant who cannot meet the evidentiary standard can find herself facing prosecution. The UN Committee on the Rights of the Child put this to Afghanistan directly: it was “deeply concerned” that “the crime of rape has not been clearly defined and separated from the offence of zina” and that “shame and stigma are attached to the child victim rather than the perpetrator.” That is a mechanism that suppresses reporting. It is not a claim about what Islam teaches — Islamic jurisprudence also contains strongly victim-protective positions, including a minority tradition classifying rape as hirabah rather than zina, and rules accepting a single witness or physical injury as proof.

Editorial decisions

Two choices we made, and why

Some of what shapes this archive is not a limitation we suffer but a decision we took. Two of those decisions changed what the record can see, and both are worth stating.

Where religion and the state are the same thing

Our test asks whether someone held spiritual authority over the person they harmed, and whether the harm depended on that authority. That question quietly assumes something: that religious authority is separable from state authority — that you could strip the religion away and still have a functioning judge, police officer or legislator underneath. In much of the world that assumption is simply false. The UN Committee on the Rights of the Child found that judges in Saudi Arabia “rule on the basis of their personal interpretation of religion.” Such a judge is not a secular official who happens to be devout. The religious interpretation is the authority.

Read narrowly, our own criterion would have excluded exactly those cases — the cleric-judge because he is a judge, the religious police because they are police, a religious legal code because it is legislation. The effect was perverse: the more completely a state fused religion and government, the less of it this archive could see. That is the opposite of what the archive is for, and unlike press freedom, it was our doing rather than the world's.

So where a state's legal authority is derived from religious law, we treat officials exercising that authority as exercising religious authority, and we ask whether the coercion survives removing the doctrine rather than removing the office. This applies wherever it applies — Iran and Saudi Arabia, but equally Israel's rabbinical courts, Vatican City, and historically Ireland's Magdalene laundries and the church-run residential schools of Canada and Australia, which are already in this archive. A rule that reached only one religion would be a bias, not a fix. It does not extend to ordinary crime that happens to occur in a religious state, and it never converts persecution into abuse: a theocratic government prosecuting a religious minority remains the inverse category, counted separately.

5 records rely on this rule.

Records without a name

We used to require that a source name the person convicted. That sounds like rigour and was actually a provincialism. Russia, Germany, Poland and much of the civil-law world anonymise published judgments by law — a Russian verdict calls the accused “ФИО1.” Insisting on a name meant rejecting a court's own judgment, the strongest document available, while accepting a weaker English-language news report that happened to print one. The standard is an adjudicated legal record, not a jurisdiction's publishing convention. Those records are identified by case number and court. We never guess at a name, and we never cross-reference other sources to unmask someone a court chose to protect.

Known limits

Other distortions we know about

  1. Anglophone sourcing

    Our researchers work mainly in English. We do read court records in Russian, Spanish, Portuguese, Bahasa, Turkish and other languages, but not evenly, and not with equal depth. Every other bias on this page is compounded by the languages we read well.

  2. Institutional traditions leave more paper

    Traditions with formal hierarchies produce personnel files, transfer records and internal complaint processes. Decentralised traditions — independent congregations, unaffiliated healers, small communities — leave far less behind. Their abuse is systematically harder to document at any evidence tier, which makes hierarchical institutions look worse than diffuse ones regardless of what actually happened.

  3. Conviction as the bar excludes most abuse

    We require a conviction or an official finding. That excludes abuse reported and believed but never charged, cases closed by settlement and non-disclosure agreement, cases lost to statutes of limitation, and every victim who was disbelieved. This is the largest category of missing records, and it is not close.

  4. The record skews recent

    The archive covers 1819 to 2026, but older abuse is under-recorded. Witnesses die, records are destroyed, and limitation periods close cases before they are heard. A thin decade is not a safe decade.

  5. Our own screening has been wrong

    Every record here was screened by researchers applying a written criterion, and that work has erred. We have found and removed duplicate records, corrected institutions attributed to the wrong religious order, and identified screening rules that were wrongly excluding valid institutional findings. Where we correct something, we log it.

The offence types are concentrated too. The largest categories in the corpus are child-abuse (523), sexual-abuse (414), institutional-abuse (139), death (68), ritual-murder (46). Prosecutions and inquiries have focused on sexual abuse of children. Financial exploitation, forced labour, medical neglect and coercive control under religious authority are all under-represented here — not rarer in the world.

Guidance

How to use this archive honestly

  1. Do use it to see documented patterns of how religious authority is used to enable abuse.
  2. Do not use it to rank traditions against each other. The ordering you would get reflects inquiry history and press freedom, not comparative harm.
  3. Do not read a country's absence as its safety.
  4. Do read a large record count as evidence that a society investigated itself — which is usually a mark of accountability, not of exceptional depravity.

In progress

What we are doing about it

We are actively working the gaps: UN treaty-body findings for states with no domestic inquiry, national inquiry reports not yet extracted, and court records in languages other than English. Coverage will improve unevenly, and this page will be updated as it does.

If you have documentation we are missing — particularly court records or official findings from the countries listed above with no coverage — send it to us. Our method and corrections log set out the standards these records are held to and the errors we have found in our own work.