ONE IN THREE WOMEN. EVERY HOUR,
ELEVEN ARE KILLED BY FAMILY.
The WHO estimates ~1 in 3 women globally have experienced physical or sexual intimate-partner violence. UNODC counts ~89,000 women killed intentionally per year, more than half by a partner or family member, about eleven every hour. UNICEF: ~640 million currently-alive women were married as children. The justice systems that are supposed to respond to this leak almost everything they receive: in the UK, only ~1 in 100 estimated rapes ends in conviction. This is the baseline, not the exception.
Six figures behind the composite.
One 0-100 score.
Heuristic seed snapshot. Violence stress sits firmly in the 'Severe' band. WHO estimates ~1 in 3 women globally have experienced physical or sexual intimate-partner violence in their lifetime. UNODC counts ~89,000 women killed intentionally per year, more than half by intimate partners or family. UNICEF: ~640 million women alive today were married as children. The reporting-to-conviction funnel collapses by roughly 95% in most jurisdictions; only ~5 in 100 reported sexual assaults end in conviction in the UK; the US figure is similar. Online gender-based violence, coercive control as a distinct offence, and dedicated shelter capacity are areas where the policy infrastructure has only begun to catch up to the prevalence.
Over time.
Three series. Two falling slowly. One rising.
- IPV lifetime prevalence (%)WHO. Women aged 15+, ever-partnered.
- Femicides per year (thousands)UNODC. 70K in 2015 → 89K in 2023.
- Child marriage prevalence (%)UNICEF. 33% in 1995 → 19% in 2023.
Women killed per 100,000 women, year on year.
UNODC harmonised rates. South Africa anchors the high end at 9.6; Nordic and East Asian averages sit below 0.5.
- 🇿🇦South AfricaSub-Saharan Africa9.6
- 🇭🇳HondurasAmericas6
- 🇧🇴BoliviaAmericas4.4
- 🇷🇺RussiaRussia & C. Asia3.3
- 🇲🇽MexicoAmericas3.1
- 🇧🇷BrazilAmericas2.6
- 🇺🇸United StatesAmericas2.3
- 🇹🇷TürkiyeMENA2.2
- 🇫🇷FranceEurope0.9
- 🇩🇪GermanyEurope0.5
- 🇬🇧United KingdomEurope0.5
- 🇮🇹ItalyEurope0.4
- 🇪🇸SpainEurope0.4
- 🇯🇵JapanAsia0.3
What the score is measuring.
Several traditions reading the same data.
Why is gendered violence so pervasive, and what frame should we read it in?
Treats IPV as a measurable, preventable public-health problem with quantifiable risk factors and intervention pathways. The WHO Global Burden of Disease analyses put IPV among the top causes of disability-adjusted life-years lost among women aged 15-49. The public-health frame produces interventions but can elide structural-power questions.
“Violence against women is everywhere a problem of magnitude that warrants the responses we give to any other major public-health epidemic.”
Frames IPV and femicide as outcomes of structural patriarchy: economic dependence, legal asymmetries, normalisation of coercive control. Argues for legal reforms (criminalising coercive control, Istanbul Convention compliance, removing marital rape exemptions) and broader political-economic redistribution.
From those who've been inside the systems: what makes a difference is safe housing first, economic independence second, perpetrator accountability third, in that order. Generic legal reform without funded service infrastructure produces glossy laws with no implementation.
In communities where family honour is structurally tied to women's perceived sexual conduct, both the violence and the silence around it are reproduced inside families and tolerated by local authorities. Western-style legal reforms run into reporting suppression that legal codes alone cannot fix.
Religious communities are simultaneously protective (mutual-aid networks, marriage-counsel infrastructure, men's-accountability practices in some Pentecostal and Sufi traditions) and complicit (theology of female submission, refusal to recognise marital rape, sheltering perpetrators inside congregational structures). Both readings are honest.
The Missing & Murdered Indigenous Women & Girls movement frames the violence as both gendered and colonial: the product of dispossession, residential-school trauma, and state-tolerated impunity, not just patriarchy. The Canadian National Inquiry concluded that what is happening to Indigenous women constitutes 'a Canadian genocide'.
Most IPV is committed by men. Treating men as un-intervenable subjects (vs as people who can be re-socialised through accountability programmes) is empirically wrong. Programmes like Caledonian, ManKind Project, Caring Dads show measurable re-offence reduction: small effects but real.
Which interventions actually move the indicators?
From service providers: the single highest-leverage intervention is access to safe emergency accommodation. Without it, leaving is impossible; with it, everything else becomes available. UK Women's Aid trials show housing-first dramatically reduces post-separation IPV exposure.
Cash transfers to women, micro-finance to women, equal-pay enforcement, parental-leave reform. The IMF-WB review of 30+ studies: ~10-15% reductions in IPV prevalence in trials where women's independent income materially increases.
Criminalising coercive control, removing marital-rape exemptions, mandatory prosecution policies, Istanbul Convention compliance. UK 2015 coercive control law: ~30,000 convictions to date, normative effect possibly larger than direct prosecution count.
Court-mandated batterer-intervention programmes with regular supervision. Effect sizes modest (~5-15% reduction in re-offending) but consistent. Best-evidenced versions integrate substance-abuse and mental-health support.
Long-running community-level programmes that shift gender-norm attitudes among men and women alike. SASA! in Uganda is the best-evidenced; reduced past-year IPV by ~52% over 4 years. Norm change is slow; without it, legal reform under-performs.
Where the cases go.
100 → 16 reported → 3 charged → 1.3 convicted.
Of every 100 estimated rapes in England and Wales, about 16 are reported, 3 are charged, and 1.3 lead to conviction. The steepest drop is between report and charge.
- Estimated occurrencesCrime Survey of England and Wales extrapolation.100of 100
- Reported to policeMost are never reported; under-reporting especially severe for intimate-partner cases.↓ DROP-OFF: 84% from previous stage16of 100
- Charge brought↓ DROP-OFF: 81% from previous stage3.0of 100
- Conviction↓ DROP-OFF: 57% from previous stage1.3of 100
100 → 31 reported → 10 referred → 6 prosecuted → 3 convicted → 2.5 incarcerated.
Reporting rate is higher than the UK (~31% vs ~16%); the prosecution funnel is similarly steep. Plea-bargain practice masks some of the variance.
- Estimated occurrencesBJS NCVS extrapolation for sexual-assault incidents in the US population.100of 100
- Reported to policeNCVS estimate; the remaining ~69 are never reported.↓ DROP-OFF: 69% from previous stage31of 100
- Arrest made↓ DROP-OFF: 68% from previous stage10of 100
- Prosecution opened↓ DROP-OFF: 40% from previous stage6.0of 100
- Felony conviction↓ DROP-OFF: 50% from previous stage3.0of 100
- Incarceration↓ DROP-OFF: 17% from previous stage2.5of 100
Sources, weights, and code are open.
Where every number comes from
The composite index is computed from the signals listed on this page, each backed by one or more named sources. Where the source publishes a public dataset or feed it is linked below; where a signal involves qualitative judgement, the LLM-assisted pass is explicitly marked on the signal card.
- ·Amnesty International Toxic Twitter
- ·CDC BRFSS; WHO ACE-IQ international studies
- ·HBVA database; UN OHCHR
- ·NNEDV (US); Women's Aid (UK); WAVE Europe
- ·Plan International State of the World's Girls
- ·Rape crisis network UK; Bureau of Justice Statistics (US); EUR Eurostat
- ·UN Women
- ·UN Women legal database; Council of Europe GREVIO
- ·UN Women legal database; WomanStats
- ·UNFPA
- ·UNICEF Child Marriage Data Hub
- ·UNODC; UN Women Femicide Watch
- ·WHO Global Database on the Prevalence of Violence Against Women
Everything is versioned
- → Every hourly snapshot is committed to git with a message naming the signals that moved.
- → A daily snapshot is archived to
data/history-current/for the calibration log. - → Raw scraped article lists are written to
data/raw/so a score is reproducible from its input bundle. - → Signal definitions, weights, and seeded scores all live in plain JSON or TypeScript; anyone can open a PR challenging a value and explain why.
How this pillar is scored.
Methodology & limits
Ten signals, weighted into a 0-100 score. Four prevalence measures (IPV, femicide, child marriage, ACE exposure), three structural (coercive-control law, shelter access, policy recognition), and three on the justice funnel (reporting-to-charge-to-conviction attrition).
Online gender-based violence is weighted on a par with offline. The Glitch / Plan International / Amnesty body of work on cross-over harm has gotten too consistent to discount: large online harassment campaigns measurably raise real-world threat exposure and silencing effects, especially for women in public life.
Three other pillars feed this one in the data. War: women in conflict zones face systematically higher IPV and conflict-related sexual violence. Inequality: economic dependence is one of the strongest structural risk factors in the literature. Religion: religious-nationalist movements often run restrictive gender policy as a flagship issue.