Resistance training, and choosing a class

A self-defence class promises something a page can't: that you'll practise, with other people, what to do if someone tries to hurt you. Whether that promise is kept is a question the evidence can partly answer. This lesson sets out the best-studied randomised trial of a programme that teaches women to resist, what happened to its effect over two years, the trials that followed it outside Canada, and the question underneath: whether this kind of teaching puts the job of prevention on the people at risk. Then it turns to what a class can and can't do, whom the evidence leaves out, and how to judge a class before you sign up.

Before any of the evidence, one thing this lesson will not do. It will not suggest that anyone who was assaulted should have taken a class, or that a class they took should have worked. The people who built the best-studied programme say the same in their own two-year paper: "The program makes clear that perpetrators are entirely responsible for the crimes they commit."² Lesson 11, "Sexual assault: the freeze, resistance and the law", showed that the body can stop a person resisting at all, and that this is involuntary. A trial's result is an average over hundreds of people. It is never a measure of what any one person could have done. That holds for someone who froze, and for someone who fought, or had trained, and was hurt anyway.

One trial, read closely

The programme is called Enhanced Assess, Acknowledge, Act, or EAAA. Senn and colleagues tested it in a randomised trial at three Canadian universities, published in the New England Journal of Medicine in 2015 (the abstract), and followed the same women for two years in a second paper.¹²

Each first-year student in it was randomly assigned either to the programme or to a session giving access to brochures about sexual assault, "as was common university practice".¹ The programme was "four 3-hour units ... with the goal of being able to assess risk from acquaintances, overcome emotional barriers in acknowledging danger, and engage in effective verbal and physical self-defense".¹ The word to notice is acquaintances. The abstract's own background says young women at university are at substantial risk of being sexually assaulted, "primarily by male acquaintances", and the programme was built for that, not for a stranger in a car park.¹ In all, 451 women were assigned to the programme and 442 to the brochures, and 91% of the programme group came to at least three of the four sessions.¹ The main outcome was completed rape over the following year, from the women's answers to a standard survey.¹

In the abstract's own words: "The 1-year risk of completed rape was significantly lower in the resistance group than in the control group (5.2% vs. 9.8%; relative risk reduction, 46.3% [95% confidence interval, 6.8 to 69.1]; P=0.02). The 1-year risk of attempted rape was also significantly lower in the resistance group (3.4% vs. 9.3%, P<0.001)."¹

The same result, three ways

Logic and Argument lesson 9, "Bad reasoning in the news and everyday life", taught you to turn a relative figure into an absolute one. The working below is this course's arithmetic from the published figures.

Relative. 46.3% is the share of the brochure group's risk that the programme group didn't have. Divide 4.6 by 9.8 yourself and you get about 47%; the published percentages are rounded.

Absolute. 9.8 minus 5.2 is 4.6 percentage points: out of every hundred women, about four or five fewer were raped in the year.

One in how many. 100 divided by 4.6 is about 22. So about one completed rape was prevented in a year for every 22 women who took the programme. A university deciding whether to run the programme would want this number, and "46%" hides it.

The interval. The 95% confidence interval runs from 6.8% to 69.1%: the data fit a programme that did a little good and one that did a great deal. The two-year paper gives the interval for the absolute difference at twelve months as 0.6 to 8.5 percentage points,² so "one in 22" could be anything from about one in 12 to about one in 170. The trial shows the programme lowered the risk. It pins down the size much less tightly than the headline suggests.

What it measured, and on whom

Lesson 1 gave this course's question for every piece of advice: measured on whom, and measured as what? Here the answer is narrow. That's the trial's scope, not a flaw in it.

• On whom. First-year women at three universities in one country, who volunteered for a study about sexual assault. The two-year paper notes that volunteers came in with somewhat higher rates of past victimisation than students in general, and that the authors don't know whether the programme would work as well for "less privileged young women who do not attend university".²
• As what. Completed rape, reported by the women themselves in a survey, over one year.¹
• Against what. Brochures and the chance to talk to someone about local services, not nothing and not another class.¹²

The trial was funded by the Canadian Institutes of Health Research and the University of Windsor.¹ In this course's labels, the programme is tested, for first-year university women and for the risk it was built around. For anyone else, the result is an inference.

Two years on

The second paper, in Psychology of Women Quarterly in 2017, says that over two years the programme significantly reduced completed and attempted rape, attempted coercion and non-consensual sexual contact, "yielding reductions between 30% and 64% at 2 years".² Its table shows how completed rape changed, point by point.

At twelve months the programme group was 4.6 points lower; at eighteen months and two years, 3.7. So, on this course's arithmetic, the programme group did not pull further ahead in the second year. The relative reduction fell to 34.0% at eighteen months and 31.3% at two years, and the table marks both as no longer statistically significant (p = .10 and .13, against .02 at twelve months). At two years the interval for the difference runs from 8.5 points fewer to 1.1 more.² Attempted rape held, at about 63% or 64%, significant at each point.² And only the women who joined in the first year of recruitment, 370 in all, were surveyed at two years, because, the authors write, funding didn't allow following everyone.²

The programme went on working for at least two years, they write, but its effect on completed rape diminished and was no longer statistically significant, and a booster may be needed in the second year of university.² They offer one possible reason, which the trial did not test: as students move through university, boyfriends may become the more likely perpetrators, bringing in forms of partner violence a twelve-hour course may not cover well enough.² A real effect that fades is a pattern this lesson comes back to.

The critics' reply, and the authors' answer

A trial like this has a reply waiting, and the authors print it themselves: "Some feminist scholars (e.g., Basile, 2008, as well as one of the reviewers for this article) have suggested that resistance education for women is misguided, in that perpetrators may simply move on to another woman."² Prevented for the woman who resists, that is, but not necessarily for the community. Lesson 6, "Out and away", met it briefly. This course hasn't read the 2008 work they cite, so the critics' case here is as the trial's authors report it.

Their answer: "We, along with other feminist scholars (e.g., Hollander, 2016), suspect that perpetrators may learn important lessons that then have an impact on their subsequent behavior".² And they grant the point that matters most: "which explanation is correct is an empirical question".² The trial can't answer it. It measured what happened to the women in it, not to other women the same men met afterwards.

Later designs on campus

Two further studies of the same programme are weaker in design than the trial; a public body's view follows them.

Out in the world. Once the programme was offered through a non-profit, the team studied it at five Canadian universities between 2017 and 2021, comparing women who signed up but didn't attend. On the abstract, "Completed rape was significantly reduced by 57.3% at 6-months."³ Nobody was randomly assigned, and on this course's reading, women who sign up and don't come may differ in ways that bear on their risk.

Partner violence. A 2025 substudy followed 153 women from the trial into new relationships. On its abstract, the programme "significantly reduced the 1-year risk of IPV by 54.4% (p = .037, 95% CI [2.9%, 79.8%])", IPV being intimate partner violence.⁴ The interval runs from almost nothing to a great deal, and the sample is small.

A public body's view, dated. In 2016 the US Centers for Disease Control and Prevention (CDC) listed "Empowerment-based training for women to reduce risk for victimization" in a package of approaches to preventing sexual violence, with this programme as its example. It said intensive skills training of this kind has "greater research and theoretical support than approaches focused primarily on physical self-defense training".⁵ It added that these approaches had mostly been evaluated with college students.⁵ The CDC has since withdrawn the package, so this is its 2016 document, read from an archived copy, not current CDC guidance.

Beyond the campus

The other studies this course read come from Kenya, Malawi and South Sudan, of programmes teaching girls and young women empowerment and self-defence skills, and they are more mixed.

Nairobi, before and after. A 2014 cohort study near Nairobi, the weakest design here, found sexual assault fell among adolescents who took the programme, with no significant change among those given a life-skills class.⁶

Nairobi, randomised by school. The cluster-randomised trial that followed, in 28 schools, estimated "a 3.7 % decrease, p = 0.03 ... (initially 7.3 % at baseline)".⁷ The abstract doesn't say whether the 3.7 is points or a share of the 7.3; this course reads it as points. Boys in the same schools took a parallel programme ("The boys' program promotes healthy gender norms"), so the result is for the pair, not for self-defence alone.⁷ The conflict-of-interest statement says the instructors, "and thus the survey interviewers, were employees of Ujamaa-Africa", the organisation that ran the programme and funded the data collection there, and that the Stanford evaluators had no financial ties to it.⁷ The authors stated it themselves, and it isn't an accusation, but the people asking girls about assault were the people who had taught them.

Nairobi, independent. A later trial in the same setting followed 3,263 girls aged 10 to 14 for two years. Its abstract says it "improved on previous ESD studies in this setting with longitudinal follow-up of individuals and independent data collection", ESD being empowerment self-defence, and it found no effect: "Odds of reporting rape were not significantly different in the intervention versus SOC group (OR: 1.21; 95% CI (0.40, 5.21), p = 0.63)", SOC being the standard-of-care comparison.⁸ An odds ratio of 1 means no difference, and this interval runs from well below 1 to well above it. It too paired the girls' programme with one for boys, and the authors ask for the findings to be read cautiously because of the study's limitations.⁸

That trial is itself disputed. The journal that had published its protocol, its plan, later issued an Expression of Concern: the surveys were planned for groups of two to four girls with one interviewer, the ratio in the study was much higher, which the protocol didn't reflect though it was published after the study ended, and this "may have caused underreporting of sexual assault incidents". "The actual ratio is under dispute among the authors", and some of them disagree with the Expression of Concern.⁹

Malawi and South Sudan. A trial across 141 schools in Malawi found past-year sexual assault fell among students in programme schools (a risk ratio of 0.68, 95% interval 0.56 to 0.82), but not in the others.¹⁰ A study in South Sudan and the Kakuma refugee camp found annual rape rates fell "from 10.7% to 5.5%" in programme schools, and "there was no change in the control schools (10.0%-9.0%)".¹¹ The Malawi paper calls itself an implementation trial, and authors of both papers list Ujamaa organisations, the programme's provider, among their affiliations.¹⁰¹¹

Put together, on this course's reading: five studies, four finding a fall, and the one that describes its data collection as independent finding none and carrying a dispute about its method. The evidence is contested, and none of the results this course read is about men or boys.

The meta-analysis argument

A meta-analysis pools many studies to find an average effect. In 2023 Kettrey and colleagues pooled 80 studies of sexual assault prevention programmes for American college students, published between 1991 and 2021. On the abstract, "Effects on sexual assault victimization were significant but small (g = 0.15)", g being a standard measure of an effect's size across studies, and "effects on sexual assault perpetration were nonsignificant".[12] They found that "programs that use a risk reduction framework are associated with less favorable outcomes than programs that do not", and recommended that prevention move beyond a focus on individuals to an "ecological perspective", aimed at relationships, communities and society as well.[12]

The abstract does not define a risk reduction framework, and its studies were of American students, so it doesn't say whether EAAA, a Canadian trial, was among the 80, or in the group that did less well. Researchers in empowerment self-defence, among them Hollander, Ullman and Senn, published a short reply, and Kettrey's team answered it; this course read neither.[13]

Hollander's own work sits on the other side. In a non-randomised study of a ten-week university self-defence class, women who took it were "less likely to experience sexual assault" over a year than similar women who didn't.[14] In a 2018 review she argues that empowerment-based women's self-defence is "so far the only approach that has produced substantively significant decreases in victimization rates".[16]

On this course's reading, a pooled average across 80 studies and one randomised trial of one programme answer different questions, how campus programmes do on average and whether this one worked, so both results can stand at once. That reading depends on which programmes the average covers, which the abstract does not say.

Where the effort should go

Underneath the evidence is a question evidence can't settle: should prevention effort go into teaching people at risk to resist, or into changing the people who commit sexual violence and the places where it happens? This is a value question, and the course describes it without choosing.

The case that it puts the burden in the wrong place. This course hasn't read a critic making the case at length, so the case below is put together from what it did read. Hollander, who disagrees with the critics, lists the objections that have kept self-defence out of prevention work as "concerns that it is ineffective, encourages victim blaming, neglects acquaintance assault, and does not target the underlying factors that facilitate sexual violence".[15] The critics the trial's authors report argue that resistance may only move an offender on to someone else.² And Kettrey and colleagues, from their pooled evidence, write: "Considering the limited effect of campus sexual assault prevention programs on violence, we recommend programming efforts move beyond a focus on individuals".[12] At its strongest, in this course's words: when the people at risk are taught to fend off the people who harm them, the work of prevention lands on those at risk, and even a programme that works for them may leave the offenders, and the settings that allow them, untouched.

The case that it is the best tool available now. Hollander argues that those critiques are "misguided", and that empowerment-based self-defence "helps to change the root conditions that allow violence against women to flourish".[15] Senn and colleagues call their programme "one critical piece of the solution", to be offered while wider change comes through bystander education and "the few effective or promising interventions for boys". Their reason is that "few interventions with men have been rigorously evaluated", and those that work usually change attitudes for a few months "with no effect on the occurrence of rape or other forms of sexual assault", with two recent exceptions that combine bystander and social-norms approaches.² They add that in most sexual assaults there is "no bystander present".² Kettrey's pooled evidence bears on that too: effects on perpetration were nonsignificant.[12] The CDC's 2016 package calls empowerment training "a useful and effective complement" to preventing perpetration, which it names as its own emphasis.⁵ At its strongest, in this course's words: on its proponents' reading, this approach has the strongest trial evidence of any campus programme, and holding it back until the larger work succeeds leaves women without the one tool shown to help them now.

What both sides share. Both sides, as this course read them, put the responsibility on the offender, and two sources say so outright. The trial's authors write that interventions for women "must not inadvertently suggest that women are responsible for sexual assault".² The CDC's package says any programme aimed at reducing the risk of being victimised must avoid placing any responsibility on its participants, stated or implied.⁵ The critics' worry is that such programmes may do it anyway.

The course holds that one shared commitment and takes no side on the rest. Lesson 6 found the two sides closer on one point than they look, and Vera-Gray and Kelly show why. Their critique of the "safety work" mandated for women and girls in public is that it functions "to limit women's space for action and responsiblise them for preventing violence" (their spelling), and their abstract ends by exploring feminist self-defence as a way of making that work visible.[17] Their target is advice telling women to restrict themselves, which is not what the trial tested. Whether training carries the same cost is the value question above, and the course leaves it there.

What a page, a class and practice can each do

A page can teach you to recognise trouble, the order of options from lesson 1 (avoid, leave, talk, comply or resist, get away), what sudden fear does to the body, and how to choose a class. It can't teach a strike, a release or an escape. Lesson 8, "Talking it down", quoted Project BETA's caution about de-escalation: knowledge can come from a book, but skills come only with practice.

A class is where the practice happens. The US workplace safety regulator, the Occupational Safety and Health Administration (OSHA), says in its guidance for health and social care workers that "Both de-escalation and self-defense training should include a hands-on component."[19] Carrying that beyond those workplaces is this course's step. The University of Connecticut's police department runs a twelve-hour women's course that "progresses from awareness, prevention, risk reduction, and avoidance to basic self-defense participation", and says it "cannot offer self-defense classes of fewer hours than listed."[20] That's one department's standard, practitioner opinion. EAAA ran for twelve hours too, and its authors write that in their early pilots, shorter units with less practice time gave effects that didn't last.²

Practice under pressure is what lesson 10, "What fear does to you, and the robbery decision", took from two police studies: after officers trained against an opponent who shot back, their "shot accuracy ... no longer deteriorated under anxiety", while mistakes in split-second decisions about whether to shoot "persisted" after three weeks of practice.[21][22] Those were police officers with firearms. The 2011 abstract says the gain was "potentially related to changes in visual attention on task-relevant information"; the 2015 one offers no mechanism, and this course adds none.[21] What it takes from them, as its own inference, is that a class where you practise against a partner who pushes back, with your consent and in steps you choose, may prepare you better than one where you only watch: sensible, untested, in this course's labels.

Why it fades

The EAAA effect on completed rape shrank in the second year, and its authors suggested a booster.² You've seen the same shape before: lesson 8's Indianapolis officers, whose improved attitudes mostly faded by the follow-up, and First Aid and CPR lesson 16's CPR skills, which fell within three months. The EAAA authors offer one possible reason for their own fade, above; none of the sources tests why, and none says how often a self-defence refresher is needed. The Malawi programme added refreshers to its twelve hours.¹⁰

So "take a refresher" is sensible, untested, in this course's labels: it follows from the fading seen in all three places, and no source this course read has measured the right interval for this skill. Habits and Self-Discipline lesson 3, "What the cue is doing", gave the tool for making a refresher happen: tie it to a cue you meet reliably and notice easily.

The gap, said out loud

Everything controlled in this lesson is about women and girls, and about sexual assault. The research behind this course found no controlled evidence on self-defence classes for men, or for general assault or robbery. That gap includes men who have been sexually assaulted, whom lesson 11 names.

That matters because of where men's risk lies. Lesson 3 showed that in England and Wales men were more likely than women to be victims of violence, with or without injury, by a stranger: 1% of men against 0.4% of women in the survey year ending March 2024.[23] Stranger violence, where men's risk runs above women's, is exactly the kind of harm for which this course found no trial of a class. UConn's police, whose women's course appears above, also run a twelve-hour course for men, "to provide realistic self-defense options to men regardless of their physical conditioning"; like every class for men, no trial this course found has tested it.[20]

Two wrong conclusions are easy to draw here. One is that the EAAA result carries over: it doesn't, because it measured different people facing a different harm, with a programme built around men they knew. The other is that classes for men or for street violence don't work. Logic and Argument lesson 6, "How much should the evidence move you", showed that absence of evidence is evidence of absence only when the evidence would have been expected. This course's research found no such trials, and a search that turns up nothing is weak evidence either way. The honest label for a class aimed at street violence is practitioner opinion: experienced people teach it and no controlled study this course found stands behind it.

Choosing a class

This course names no martial art, style, product or school, and ranks none. It gives two checklists, each labelled.

A practitioners' guide

The National Women's Martial Arts Federation, a non-profit, publishes a short guide, "How to choose a self-defense class", by Amy Eckert, 2025. This course read it in full.[18] It opens: "In a good empowerment self-defense program, students learn awareness, assertiveness, and verbal and physical skills." It then lists what to look for in an instructor or programme:

• One that knows the facts about violence and teaches to them: "a good class will address situations involving acquaintances and romantic partners as well as attacks by strangers".[18]
• One that "does not blame or judge survivors".[18]
• One that offers strategies "to help students make their own decisions ... instead of telling students what they should or should not do".[18]
• One where students "determine their own levels of participation", and where "no one should feel pressured into doing specific exercises".[18]

The guide sets no length and names no style. It credits part of its material to the National Coalition Against Sexual Assault.[18] In this course's labels, the guide is practitioner opinion: guidance from experienced teachers, not tested in a study. It was written with women, transgender and non-binary people and others targeted for violence in mind: those are the groups whose risks it asks an instructor to know.

This course's checklist

The second list is this course's own inference from the sources in this lesson and the ones before it, not a tested tool: sensible, untested, in this course's labels.

1. Covers avoidance, talking and getting away, not only fighting. Lesson 1's order of options; UConn's course starts from awareness and avoidance.[20]
2. Covers people you know, not only strangers. EAAA was built on acquaintance risk.¹[18]
3. Includes practice with a partner, under some pressure, consented and in steps. Lesson 10's police studies; OSHA's "hands-on component".[19][21]
4. Counts getting away and calling for help as success. Leaving is allowed at every rung.
5. Covers the law where you live, as general information, with escape first. Lesson 13, "The law on using force", sets it out.
6. Never blames survivors. The trial's authors, the CDC's 2016 package and the federation's guide.²⁵[18]
7. Lets you opt out of any drill.[18]
8. Offers a refresher. EAAA's second year.²
9. Is candid about what a short course can't do, including, for most classes, that no trial this course found has tested it.

What not to count as evidence. A black belt, a long lineage, a certificate from the school's own association, years in business, or a promise that you'll never be a victim. The commercial "how to choose" pages the research behind this course found lean on lineage, certificates and years teaching, and they're written by schools selling classes. None of it tells you whether the teaching prevents harm; the one controlled finding here was judged by what happened to the women afterwards.

If you've been assaulted. You don't have to take a class, and not taking one isn't a choice anyone can hold against you. If you'd like to, you can take your time. The EAAA authors record that a few survivors in the trial found their experience too recent and left, and that some took the programme a semester or a year later.² And if you'd taken a class and were assaulted anyway, the class didn't fail because of you, and nor did you. Even in the best trial, trained women were still raped, and the programme itself teaches that "only the presence of a man willing to commit sexual assault generates any 'risk' of sexual assault".²

A class for the street

Dominic is 34 and works night shifts at a parcel depot on the edge of Exeter, and walks across a big car park to his car at about three in the morning. He'd like "a class for the street" and has found three near him. Which, if any, should he pick?

What the evidence covers. Not him: nothing controlled that this course found covers a man and violence from a stranger. Lesson 4 gave the opportunity tools for a place like his car park, the route he takes, the time, and staying near other people. A class is one option beside those, and any class he picks is practitioner opinion.

The three descriptions (invented for this lesson):

• Class A. "Street Survival. One two-hour seminar. Learn the five moves that end any fight. Taught by a seventh-degree master with thirty years' experience."
• Class B. "Personal safety for shift workers. Four three-hour sessions: noticing trouble, talking it down, getting away and calling for help, and practice with padded partners in scenarios you build up to. A session on the law in England and Wales. You choose which drills you join. A refresher evening six months later."
• Class C. "Combat fitness. Get fit, get strong, feel confident. Pads, bags and conditioning, three evenings a week."

Class B meets most of this course's list, though it says nothing about people you know, which both lists ask for. The federation's guide was written for a different audience, so for Dominic it's a reference rather than a standard.

What people get wrong

"Self-defence means fighting off a stranger." The best-studied programme was built around men the women knew, and the federation's guide asks a good class to cover acquaintances and partners as well as strangers.¹[18]

"The trial settled whether resistance training puts the burden on women." It didn't, and couldn't: the trial measured rape among the women in it. Critics argue that training women puts the work of prevention on them and leaves the "underlying factors" untouched; the authors answer that their programme "makes clear that perpetrators are entirely responsible for the crimes they commit". That part is a value question, and this course doesn't settle it.²[15]

"You can learn a technique from a page or a video." Every source on training in this lesson points to practice with other people, which is why this course teaches no technique.

Practice

Connections

Back. Lesson 1 gave the order of options and the advice labels; lesson 10, the police studies of practice under pressure; lesson 11, the freezing that no class makes anyone's fault.

Forward. Lesson 13, "The law on using force", sets out the law a good class should teach as general information. Lesson 16, "Afterwards", covers reporting and support, whether or not you report.

Go deeper

• Senn and colleagues, "Secondary and 2-Year Outcomes of a Sexual Assault Resistance Program for University Women", 2017, free. Read in full for this course. What the programme taught, what changed and what faded, with the critics' reply and the authors' answer.
• National Women's Martial Arts Federation, "How to choose a self-defense class". Read in full. Short, and the plainest statement this course found of a good class from the teachers' side.

Sources

1. Senn and colleagues, "Efficacy of a Sexual Assault Resistance Program for University Women", New England Journal of Medicine 372(24), 2015, pp. 2326 to 2335. Read: abstract read in full (PubMed).
2. Senn, Eliasziw, Hobden and colleagues, "Secondary and 2-Year Outcomes of a Sexual Assault Resistance Program for University Women", Psychology of Women Quarterly 41(2), 2017. Read: in full.
3. Senn and colleagues, "Testing the effectiveness of a sexual assault resistance programme in 'real-world' implementation", European Journal of Psychotraumatology 14(2), 2023. Read: abstract.
4. Barata and colleagues, "A Successful Sexual Assault Resistance Program Also Reduced Intimate Partner Violence", Journal of Interpersonal Violence 40(9-10), 2025. Read: abstract.
5. Basile and colleagues, STOP SV: A Technical Package to Prevent Sexual Violence, Centers for Disease Control and Prevention, 2016, as archived June 2021. Read: from an Internet Archive capture of 2021, capture 20210608134353. Withdrawn from the CDC's site; not current CDC guidance.
6. Sarnquist, Omondi, Sinclair and colleagues, "Rape prevention through empowerment of adolescent girls", Pediatrics 133(5), 2014. Read: abstract (cohort).
7. Baiocchi, Omondi, Langat and colleagues, "A Behavior-Based Intervention That Prevents Sexual Assault: the Results of a Matched-Pairs, Cluster-Randomized Study in Nairobi, Kenya", Prevention Science 18(7), 2017. Read: abstract (cluster-randomised), with its conflict-of-interest statement.
8. Sarnquist and colleagues, "Sexual Assault Among Young Adolescents in Informal Settlements in Nairobi, Kenya: Findings from the IMPower and SOS Cluster-Randomized Controlled Trial", Prevention Science 25(4), 2024. Read: abstract.
9. BMC Public Health, "Expression of Concern to: A protocol for a cluster-randomized controlled trial testing an empowerment intervention to prevent sexual assault in upper primary school adolescents in the informal settlements of Nairobi, Kenya",
   2021. Read: in full. The protocol it concerns carries the same trial registration as source
10. Decker, Wood, Ndinda and colleagues, "Sexual violence among adolescent girls and young women in Malawi: a cluster-randomized controlled implementation trial of empowerment self-defense training", BMC Public Health 18, 2018. Read: abstract.
11. Wheeler, Edwards, Omondi and colleagues, "Empowerment Transformation Training Reduces Rape Among Girls and Young Women in South Sudan and the Kakuma Refugee Camp", Journal of Adolescent Health 74(4), 2024. Read: abstract.
12. Kettrey, Thompson, Marx and Davis, "Effects of Campus Sexual Assault Prevention Programs on Attitudes and Behaviors Among American College Students: A Systematic Review and Meta-Analysis", Journal of Adolescent Health 72(6), 2023, pp. 831 to 844. Read: abstract (meta-analysis).
13. Hollander and colleagues, letter, Journal of Adolescent Health 74(1), 2024, pp. 208 to 209; and Kettrey and colleagues, reply, pp. 209 to 210. Read: titles only (PubMed); neither letter read.
14. Hollander, "Does Self-Defense Training Prevent Sexual Violence Against Women?", Violence Against Women, 2014. Read: abstract.
15. Hollander, "The importance of self-defense training for sexual violence prevention", Feminism & Psychology, 2016. Read: abstract.
16. Hollander, "Women's self-defense and sexual assault resistance: The state of the field", Sociology Compass, 2018. Read: abstract.
17. Vera-Gray and Kelly, "Contested gendered space: public sexual harassment and women's safety work", International Journal of Comparative and Applied Criminal Justice 44(4), 2020. Read: abstract only.
18. Eckert, "How to choose a self-defense class", National Women's Martial Arts Federation, 2025. Read: in full.
19. Occupational Safety and Health Administration, Guidelines for Preventing Workplace Violence for Healthcare and Social Service Workers, OSHA 3148, 2016 revision. Read: the PDF, searched and sections read.
20. UConn Police, "Women's Self Defense Classes: Surviving Violent Encounters". Read: in full.
21. Nieuwenhuys and Oudejans, "Training with anxiety: short- and long-term effects on police officers' shooting behavior under pressure", Cognitive Processing 12(3), 2011. Read: abstract only.
22. Nieuwenhuys, Savelsbergh and Oudejans, "Persistence of threat-induced errors in police officers' shooting decisions", Applied Ergonomics 48, 2015. Read: abstract only.
23. Office for National Statistics, The nature of violent crime in England and Wales: year ending March 2024, 26 September 2024. Read: in full.
24. Support lines in the content note, each read on its service's own page on 25 September 2026: Rape Crisis England & Wales, Victim Support, Live Fear Free, Rape Crisis Scotland, Victim Support Scotland, the Domestic and Sexual Abuse Helpline, Rape Crisis Northern Ireland, Victim Support NI, RAINN (in a browser) and VictimConnect. Read: each page in full, re-read 25 September 2026.