# Document 005: International coercive-control evidence review

Atlas Men’s Advocacy | 19 September 2026 | Public version 1

**Status: Targeted research review. Not a systematic review or pooled prevalence estimate.**

# Coercive control: international evidence and the case for recognising male victims

Atlas research dossier | 19 September 2026 | Targeted research assessment; not a legal complaint

## 1. Finding

**Women can perpetrate coercive control against men, including sustained and serious abuse. Multiple comparative studies report similar levels of particular controlling behaviours, and some report higher female perpetration. The evidence does not establish that every form, frequency and severity of coercive control is distributed equally between women and men.**

“Gender neutral” needs a precise meaning:

| Proposition | Assessment |
|---|---|
| Either sex can perpetrate coercive control or suffer it | Supported, including evidence of sustained female abuse of men |
| Several studies find comparable controlling behaviour in men and women | Supported within the populations and measures studied |
| All coercive control is equally prevalent and equally severe across sexes | Not established; some population studies find greater repeated or extensive victimisation among women |
| Male victims deserve recognisable representation, prevention and support | Atlas's advocacy position, supported by the documented existence and substantial scale of male victimisation |
| Research establishes an exact 50:50 advertising budget or a compensation entitlement | No. Those are proposed remedies requiring an additional policy or legal argument |

The most defensible foundation is **documented male victimisation and female perpetration**, strengthened by comparative findings. Exact population parity is unnecessary to establish that men belong in public education about victims and women belong in public education about perpetrators.

## 2. What was reviewed

This is a targeted international evidence review, not a registered systematic review or meta-analysis. Searches covered coercive control, controlling behaviour, intimate terrorism, female perpetration, male victimisation, gender comparisons, help-seeking and recognition. Psychology research, population surveys, longitudinal research, police records and qualitative studies were considered.

The ledger distinguishes primary papers from reviews and records whether this assessment checked relevant full-text passages or the primary abstract. Abstract-only papers support only the findings actually reported in their abstracts. A full-text designation does not mean independent reanalysis of the underlying data. No pooled prevalence estimate has been manufactured from incompatible studies.

## 3. Comparative evidence and the maths

### 1. CDC: substantial male victimisation in a national US survey

**Source:** Leemis and colleagues, *The National Intimate Partner and Sexual Violence Survey: 2016/2017 Report on Intimate Partner Violence* (2022), Tables 5 and 6, printed pages 24–25. [Official report](https://www.cdc.gov/nisvs/documentation/nisvsreportonipv_2022.pdf).

Lifetime coercive-control victimisation: women **46.2%**, men **42.8%**. The respective 95% confidence intervals are 44.8–47.6 and 41.3–44.3. The survey used 15,152 women and 12,419 men and weighted the results.

**Maths:** 46.2 − 42.8 = **3.4 percentage points**. The male estimate is 42.8 ÷ 46.2 = **92.6% of the female estimate**. This is a comparison of within-sex rates, not men's share of all victims.

This supports substantial male victimisation under the survey's behavioural definition. It does not establish identical prevalence, perpetrator sex, or equal sustained entrapment and harm. The low response rate is a survey-quality consideration. These are 2016/17 lifetime estimates, not 2026 annual incidence. **Checked:** report text and both table images.

### 2. Bates, Graham-Kevan and Archer: high control in a UK student sample

**Source:** *Testing predictions from the male control theory of men's partner violence* (2014), DOI 10.1002/ab.21499. [Primary abstract](https://pubmed.ncbi.nlm.nih.gov/23878077/); [full paper](https://www.mankind.org.uk/wp-content/uploads/2015/05/Testing-Predictions.pdf).

The sample comprised **398 men and 706 women**, recruited at the University of Central Lancashire. Table V classified **62 men (15.6%) and 144 women (20.4%)** as high-control perpetrators. High-control victimisation was **18.8% of men and 18.0% of women**.

**Maths:** female minus male high-control perpetration = **4.8 percentage points**, using rounded table rates. Victimisation differs by **0.8 points**, in the male direction.

This challenges claims that high control is exclusively male. The classification is a research typology in a nonrepresentative student sample, not a national estimate or a finding that each participant committed a criminal offence. **Checked:** full-text methods and Table V. The recruitment location is UK, not USA.

### 3. Bates and Graham-Kevan: Canadian national-survey analysis

**Source:** *Is the presence of control related to help-seeking behavior?* (2016), DOI 10.1891/1946-6560.7.1.3. [University record](https://research.cumbria.ac.uk/en/publications/is-the-presence-of-control-related-to-help-seeking-behavior-a-tes/); [accepted manuscript mirror](https://scispace.com/pdf/is-the-presence-of-control-related-to-help-seeking-behavior-329x8cys1l.pdf).

Secondary analysis of the **1999 Canadian General Social Survey**. The parent survey included 25,876 respondents, but that is not the denominator for every relationship analysis. Eight items assessed current-partner control over five years. The authors found no significant sex difference in high/low-control victim classification (**p = .454**).

This is useful population-survey evidence against assuming control is necessarily male-perpetrated. However, current-partner analysis can miss abuse in relationships already ended; victim reports are not self-reports of perpetration. A nonsignificant difference is not a statistical demonstration of equivalence. **Checked:** primary manuscript methods and relevant results. Do not advertise this as 25,876 coercive-control victims.

### 4. Hilton and Radatz: US longitudinal self-report study

**Source:** *Coercive control in a national U.S. self-report survey: Prediction of repeated intimate partner violence* (2025; online 2024), DOI 10.1037/ser0000881. [Primary abstract](https://pubmed.ncbi.nlm.nih.gov/38884951/).

Among **1,039 women and 509 men**, **26% of women and 14% of men** reported perpetration under the study's definition. That definition included any reported threat of physical harm, threat to use information to control, or putting down/disrespecting a partner.

**Maths:** 26 − 14 = **12 percentage points**; 26 ÷ 14 = **1.86**.

The broad threshold prevents treating these figures as prevalence of sustained coercive domination. Control predicted subsequent physical IPV independently of baseline violence for men, but not women. Both results belong together. **Checked:** primary abstract; full methods not independently inspected.

### 5. Fanslow and colleagues: comparable “any control”, unequal repeated exposure

**Source:** *Evidence of Gender Asymmetry in Intimate Partner Violence Experience at the Population-Level* (2023), DOI 10.1177/08862605231163646. [Full paper](https://journals.sagepub.com/doi/pdf/10.1177/08862605231163646).

New Zealand population study: **1,431 ever-partnered women and 1,355 ever-partnered men**. Table 1 reports any controlling behaviour for **20.22% of men and 21.57% of women**, **p = .4512**. Item denominators vary with available responses.

**Maths:** 21.57 − 20.22 = **1.35 percentage points**.

However, examining breadth and repetition revealed greater female exposure. This is a particularly useful stress test: similar “ever experienced anything” rates can coexist with unequal frequency or extent. Cite both findings. It supports recognition of male victims, but challenges an unqualified equal-severity claim. **Checked:** primary full-text table and analysis.

### 6. Graham-Kevan and Archer: control tactics in heterosexual relationships

**Source:** *Control tactics and partner violence in heterosexual relationships* (2009), DOI 10.1016/j.evolhumbehav.2009.06.007. [Publisher](https://www.sciencedirect.com/science/article/abs/pii/S1090513809000609).

The study of **1,350 respondents** found substantial similarities in men's and women's control tactics, with control associated with physical aggression in both sexes. It supplies a psychological account extending beyond a male-only control model.

The convenience sample cannot establish national prevalence. Before counting related papers by these authors as independent confirmations, check recruitment and dataset overlap. **Checked:** primary publisher abstract and reported findings; complete recruitment and weighting details not independently audited.

### 7. Robertson and Murachver: control and violence in New Zealand

**Source:** *Women and men's use of coercive control in intimate partner violence* (2011), DOI 10.1891/0886-6708.26.2.208. [Primary abstract](https://pubmed.ncbi.nlm.nih.gov/21780535/).

The study involved **172 participants: 85 men and 87 women**, across three samples. Control was reciprocal, and violence status was more informative than sex in explaining control.

This supports investigating behaviours and relationship dynamics rather than presuming the controller's sex. Reciprocity does not demonstrate equal power, fear, harm or responsibility. The samples are not a national prevalence survey. **Checked:** primary abstract.

### 8. Gou, Duerksen and Woodin: transition to parenthood

**Source:** *Coercive control during the transition to parenthood: An overlooked factor in intimate partner violence and family wellbeing?* (2019; online 2018), DOI 10.1002/ab.21803. [Publisher](https://onlinelibrary.wiley.com/doi/10.1002/ab.21803).

Canadian research followed **98 heterosexual couples** during pregnancy and after childbirth. Control was highly bidirectional; women were more likely to perpetrate it before parenthood.

Longitudinal couple data are valuable because both partners and change over time can be considered. This small, specific life-stage sample cannot establish an all-adult population ratio. **Checked:** primary abstract.

### 9. Jolliffe Simpson, Taaka and Polaschek: police-recorded control

**Source:** *Identifying Controlling Behaviors in Police Reports for Intimate Partner Violence: Implications for Risk Assessment* (2025), DOI 10.1007/s10896-025-00946-6. [Full paper](https://link.springer.com/article/10.1007/s10896-025-00946-6).

Researchers assessed **1,158 New Zealand police reports**, involving **948 male and 210 female aggressors**. Types of controlling behaviour occurred at similar rates within those groups.

**Maths:** 948 ÷ 210 = **4.51 male-aggressor reports per female-aggressor report**. Similar within-group percentages therefore do not mean equal numbers of male and female perpetrators in police records.

This documents female controlling behaviour in police-attended incidents. Single reports cannot establish the complete relationship pattern or community prevalence. **Checked:** relevant full-text methods and findings.

## 4. Sustained abuse, recognition and harm

### 10. Hines and Douglas: intimate terrorism against men

**Source:** *Intimate Terrorism by Women Towards Men: Does it Exist?* (2010), DOI 10.5042/jacpr.2010.0335. [Primary paper](https://pmc.ncbi.nlm.nih.gov/articles/PMC3002073/).

The research compared **302 male help-seekers and 520 community men**. It identified female-perpetrated patterns consistent with intimate terrorism among help-seekers, rather than merely occasional relationship disagreement.

This is central to answering the objection that female control never amounts to sustained serious abuse. Male-only and help-seeking samples cannot establish a male/female prevalence ratio. **Checked:** primary abstract and accessible source material; no independent data reanalysis.

### 11. Hines and Douglas: closer examination of the same research

**Source:** *A Closer Look at Men Who Sustain Intimate Terrorism by Women* (2010), DOI 10.1891/1946-6560.1.3.286. [Primary paper](https://pmc.ncbi.nlm.nih.gov/articles/PMC2913504/).

Useful supporting analysis of men's experiences. **This is related work from the same 302/520 research project, not another independent population confirmation.** Count papers and datasets separately. **Checked:** primary abstract and sample description.

### 12. Bates: men's descriptions of sustained control

**Source:** *“Walking on egg shells”: A qualitative examination of men's experiences of intimate partner violence* (2020; online 2019), DOI 10.1037/men0000203. [University record](https://research.cumbria.ac.uk/en/publications/walking-on-egg-shells-a-qualitative-examination-of-mens-experienc/).

Men described isolation, restrictions, fear and psychological abuse by female partners. This qualitative work helps explain how male victims can lose autonomy even when an outside observer assumes men retain control.

Use it to establish experiences and mechanisms, not percentages of all men or comparisons with all female victims. **Checked:** primary research abstract.

### 13. Frankland and Brown: same-sex relationships

**Source:** *Coercive Control in Same-Sex Intimate Partner Violence* (2014), DOI 10.1007/s10896-013-9558-1. [University record](https://researchers.mq.edu.au/en/publications/coercive-control-in-same-sex-intimate-partner-violence/).

Australian research with **184 gay men and lesbians** identified controlling relationship patterns beyond heterosexual male-to-female abuse. It tests the adequacy of an exclusively heterosexual explanatory model.

The abstract does not provide sufficient sex-separated prevalence to establish equal male and female perpetration. **Checked:** primary abstract.

### 14. Jordan and colleagues: recognition changes with victim/perpetrator gender

**Source:** *Attitudes of U.S. Adults Towards Coercive Control as a Function of Victim and Perpetrator Gender and Respondent Sexual Orientation* (2025), DOI 10.1007/s11199-025-01616-z. [Full paper](https://link.springer.com/article/10.1007/s11199-025-01616-z).

An experiment with **1,872 US adults** varied victim and perpetrator gender in scenarios. Judgments of fear, harm and responses differed by gender configuration, including greater recognition in female-victim/male-perpetrator scenarios.

This is directly relevant to Atlas's concern about recognising male victims. It does not show that a particular WA advertisement caused those attitudes or quantify WA campaign harm. **Checked:** relevant full-text design and findings; effect sizes not extracted into this dossier.

### 15. Huntley and colleagues: barriers to men's help-seeking

**Source:** *Help-seeking by male victims of domestic violence and abuse: a systematic review and qualitative evidence synthesis* (2019), DOI 10.1136/bmjopen-2018-021960. [Review](https://pmc.ncbi.nlm.nih.gov/articles/PMC6585830/).

Synthesis of **12 studies** identified barriers including fear of disclosure, expectations of masculinity and perceived invisibility of services.

This supports the importance of recognisable male-victim messages and accessible help. It is a synthesis of existing studies, not twelve new national surveys, and does not prove a particular campaign caused non-reporting. **Checked:** primary review abstract and reported themes.

## 5. Contrary findings and measurement checks

### 16. Myhill: population evidence of gendered coercive control

**Source:** *Measuring Coercive Control: What Can We Learn From National Population Surveys?* (2015), DOI 10.1177/1077801214568032. [Primary abstract](https://pubmed.ncbi.nlm.nih.gov/25680801/).

Analysis of England and Wales survey data distinguished coercive control from situational violence and found coercive control more strongly concentrated among female victims.

This directly challenges saying “there is no evidence men perpetrate coercive control more than women.” The appropriate task is to examine its definitions, thresholds and population alongside other studies. Rejecting the finding solely because it conflicts with Atlas's hypothesis would weaken the dossier. **Checked:** primary abstract; full operationalisation requires further audit before detailed rebuttal.

### 17. Policastro and Finn: differences between control tactics

**Source:** *Coercive Control in Intimate Relationships: Differences Across Age and Sex* (2021; online 2017), DOI 10.1177/0886260517743548. [Primary abstract](https://pubmed.ncbi.nlm.nih.gov/29294994/).

A US NISVS analysis found differing patterns by age and sex, including more surveillance reported by men and more intimidation among younger women.

This cautions against compressing all tactics into one headline. Confirm survey-wave overlap before treating it and other NISVS analyses as independent datasets. **Checked:** primary abstract.

### 18. AIFS: measurement and Australian applicability

**Source:** *Coercive control literature review* (2023). [Australian Institute of Family Studies](https://aifs.gov.au/all-research/research-reports/coercive-control-literature-review).

The review highlights inconsistent measurement and limitations in Australian prevalence evidence available at its publication date. Emotional abuse, individual controlling acts and sustained coercive control cannot automatically be substituted for one another.

Its 2023 assessment is not proof that no newer Australian evidence exists. International estimates must retain their country, dates and definitions when used in an Australian case. **Checked:** institutional review.

## 6. Evidence held back from headline use

1. **Walker, Sleath and Tramontano**, *The Prevalence and Typologies of Controlling Behaviors in a General Population Sample*, DOI 10.1177/0886260517731785. [Accepted manuscript](https://pureportal.coventry.ac.uk/files/19103555/Manuscript_JIV_Submitted_REVISED2_SubmittedV1.1_1_.pdf). Community sample of 427 is relevant. Widely repeated sex-specific percentages have not been reconciled against the original table in this assessment. Do not publish those numbers yet.
2. **Felson and Outlaw**, *The control motive and marital violence* (2007), DOI 10.1891/088667007781553964. [Primary abstract](https://pubmed.ncbi.nlm.nih.gov/17691548/). A correction concerning sex labels has been flagged in secondary bibliographic material. Obtain the original erratum before using figure-specific claims; do not infer misconduct.
3. **Carney and Barner**, *Prevalence of Partner Abuse: Rates of Emotional Abuse and Control* (2012), DOI 10.1891/1946-6560.3.3.286. [Review tables](https://domesticviolenceresearch.org/pdf/PASK.Tables5.Revised.pdf). Trace any aggregate figure to its underlying studies before representing it as pooled prevalence. A review is not independent of the papers it includes.
4. **Schemmel, Maciey and Goede**, *Intimate Partner Violence Against Men in Germany*, DOI 10.1177/08862605251321003. [Publisher](https://journals.sagepub.com/doi/10.1177/08862605251321003). Promising recent male-victim research. Full sampling and denominator checks remain necessary before numerical use.

## 7. What survives a hostile review

1. **Existence is established.** Research documents women controlling men, including sustained serious abuse. “Men cannot be victims” is untenable.
2. **Comparable rates occur in the literature.** Specify the measure, sample and period. Do not translate “no statistically significant difference” into proven equivalence.
3. **Equal severity is a different proposition.** Counting one act ever cannot settle repetition, fear, entrapment, injury or autonomy loss.
4. **Victim sex is not automatically perpetrator sex.** Use sex-specific perpetrator information when claiming female-to-male abuse; do not infer it from a mixed-orientation victimisation table.
5. **Police and help-seeking samples answer different questions.** Reporting and service access shape who appears, while community surveys also face nonresponse and disclosure problems. Neither source should be dismissed automatically.
6. **The bibliography is not a vote.** Shared datasets, repeated authors, reviews of the same studies and different measurement thresholds cannot be counted as equivalent independent confirmations.
7. **International evidence is relevant, not a WA population estimate.** Do not multiply a US or Canadian percentage by WA's population and label the result measured WA victims.
8. **The campaign claim requires its own evidence.** A four-film review can describe those four films. It cannot establish the content of every placement, expenditure totals, audience reach or causally attributable harm across the campaign.

## 8. Recommended foundation for the Atlas case

**Coercive control can be perpetrated by women against men. International research documents sustained abuse of male partners, and several comparative studies find similar levels of particular controlling behaviours across sexes. Male victims are part of the evidence. They should be recognisable in public education, and female perpetrators should be recognisable too.**

**Atlas seeks equally visible reverse portrayals, with dollar-for-dollar investment in reaching the relevant audiences. The government should disclose what it spent, who its advertising reached, and what evidence informed its decisions about whose experiences to depict.**

This formulation supports the requested remedy without falsely presenting a 50:50 budget as a result calculated by the studies. A campaign can serve different audience needs while still being accountable for whether men can recognise their own victimisation.

For the proposed human-rights approach, separate four evidential questions:

1. **Recognition:** What male victimisation and female perpetration does the research document?
2. **Government conduct:** What exactly was commissioned, approved, funded, broadcast and evaluated, and by whom?
3. **Impact:** What direct evidence shows exclusion, distorted recognition, missed help or other harm?
4. **Remedy:** How do corrective advertising, an apology, additional funding and oversight address that conduct and impact?

Atlas’s proposed outcomes remain: an apology to men, dollar-for-dollar corrective advertising, additional funding for men's initiatives and independent oversight. This dossier strengthens the first evidential question. It does not yet prove the whole human-rights case or a legally enforceable compensation amount.

## 9. Next research priorities

1. Retrieve and audit the remaining full texts, especially Myhill's operational definition, the Walker tables and the Felson correction.
2. Map independent datasets and assess design quality, nonresponse, current/former-partner coverage and sustained-pattern thresholds.
3. Seek an independent methods review from a researcher experienced in both male victimisation and population measurement.
4. Build the campaign evidence schedule: dated assets, placement logs, actual expenditure, AFL placements, audience measures, decision documents and evaluations. Distinguish commitments from spending and impressions from people.
5. Assemble evidence of practical recognition and support failures affecting male victims. Research about attitudes is relevant background; direct WA evidence would strengthen attribution.

**Publication status:** This research assessment is published for examination. It is not a complaint or submission to an organisation.