Trauma Bonding: Why Leaving Someone Who Hurt You Is So Hard

You know what happened. You can list it. You have explained it to a friend and heard how it sounded out loud. And you still find yourself reaching for your phone, or rereading old messages looking for the version of them you first met, or thinking it was not all bad and maybe you overstated it.

People call that pull a failure of will. It is better described as a mechanism, and there is real research behind parts of it.

What does trauma bonding actually mean?

An attachment to the person causing the harm, not a bond formed by going through something hard together.

The term comes from psychologists Donald Dutton and Susan Painter, who in 1981 described traumatic bonding as the strong emotional attachment that can develop when one person in a relationship intermittently abuses the other. They identified two conditions that produce it: a marked power imbalance in which the person being harmed increasingly feels powerless, and abuse that is intermittent, alternating with positive or neutral treatment.

When they tested the theory on 75 women who had recently left abusive relationships, attachment to the former partner persisted even after the women were out. And the strongest predictors of that lingering attachment were not the severity of the violence. They were the intermittency of the abuse and shifts in the power balance, which together accounted for 55 percent of the variance in attachment at six months.

That is the most useful thing here. The reason it is hard to leave has more to do with the pattern than with how bad the worst night was. Which is also why people whose relationships looked comparatively mild from the outside can find leaving just as hard.

Two caveats. Trauma bonding is a descriptive term, not a diagnosis in either manual, and the evidence base is thinner than the phrase's popularity suggests: a 2022 scoping review found no single agreed definition, and none of its fifteen studies addressed how such a bond is broken. The mechanisms are better evidenced than the label. Stockholm syndrome, often mentioned alongside it, is shakier still: a systematic review found only 12 relevant papers, no validated criteria, and no recognition in the international classifications.

Why hope keeps coming back

Because intermittent reward is the most persistent kind.

In learning research this is the partial reinforcement extinction effect: behaviour that produces only periodic reinforcement tends to be more resistant to extinction than behaviour reinforced continuously. It has been documented across species and decades of experiments. A reward every time is easy to give up on when it stops. A reward that arrives unpredictably keeps the behaviour going long after it has mostly stopped paying out. Researchers applying operant learning principles to abusive relationships have argued that the dynamics increase the reinforcing value of staying, reduce the punishing effect of the abuse, and punish the alternative of leaving at the same time.

So when you think maybe this time, that is not naivety. It is a trained system, running on the schedule known to be hardest to unlearn.

What about the cycle of abuse?

It is a useful description with real limits, and one of those limits is a safety matter.

The cycle model, proposed by psychologist Lenore Walker in 1979, describes tension building, an acute incident, then a calm and contrite period in which the person who caused harm seeks forgiveness and promises it will not happen again. Many people recognise themselves in it immediately. But it was built on an unrepresentative single state sample, and Walker acknowledged it could not be considered a legitimate data base from which to make specific generalizations. Many survivors report no honeymoon phase at all. And the idea that some phases are safer times to leave is described by specialists as inaccurate and unsafe, because leaving at any point can escalate the danger.

How many times do people leave before it sticks?

You have probably seen the figure of seven times. It does not hold up.

The National Domestic Violence Hotline states on more than one page that survivors return to abusive partners an average of seven times before leaving for good. No citation is given on either page, and the figure does not appear on the Hotline's own compiled statistics page, where the other numbers do carry sources. We could not trace it to a published study. What the peer reviewed literature supports is the shape rather than the number: an empirical review found that people typically undergo several shifts in their thinking about the abuse before leaving permanently, and gives no count.

The practical meaning is unchanged. Going back is not a relapse. It is a normal part of a process that usually takes more than one attempt.

Leaving is the most dangerous point, and that is a reason for planning

This is the part the inspirational version of this conversation leaves out.

In a multisite case control study published in the American Journal of Public Health, women who had separated from an abusive partner after living together faced more than three and a half times the risk of being killed by that partner, rising roughly nine fold where the partner was highly controlling. Recent separation is treated as a lethality risk factor in risk assessment precisely because a partner who senses a loss of control may retaliate. The National Institute of Justice has also noted that leaving does not reliably end the abuse.

None of that is an argument for staying. It is an argument for not doing this alone and without a plan. A safety plan reduces risk both while you are still in a relationship and while getting ready to leave, covering the things people do not think of in advance: children, pets, phones and location sharing, documents, and safety after it ends. Advocates do this for a living, and working with one is the recommended way to build a plan.

The cost of staying, and why it is not weakness

The toll is well documented. A meta-analysis of women with a history of intimate partner violence found weighted mean prevalences of 63.8 percent for PTSD and 47.6 percent for depression. The World Health Organization also lists sleep difficulties, alcohol use disorders and suicide attempts among the consequences.

And the framing of staying as a personal failure does not survive contact with the research. The obstacles are structural: money, housing, children, immigration status, disability, faith, culture, fear of retaliation, and the risk of losing custody. One widely used list runs to fifty distinct barriers. Researchers working on coercive control argue that whether someone stays cannot be understood as an individual choice at all, because a survivor's options are shaped by their partner's behaviour and by community and institutional responses. The VA's National Center for PTSD says the short version directly: it is not your fault.

There is also a reason the pull is physical rather than sentimental. Attachment researchers describe what happens when the person you would turn to for reassurance is the person frightening you as fright without solution: the urge to approach and the urge to get away cannot both be satisfied. That work describes children and caregivers, so treat it as a parallel. But the logic holds. The attachment system intensifies the pull toward someone when danger is perceived, and does not check whether they are the danger.

What helps

• Talk to an advocate before you make a move. Advocacy means education about abuse and your rights, help accessing services, risk assessment and safety planning. A Cochrane review found it needs to run long enough and match your own goals to help, so build a relationship with one rather than make a single call.
• Expect therapy to help with some things more than others. A Cochrane review found psychological therapies probably reduce depression at 6 to 12 months and may reduce anxiety, while the evidence on PTSD was less clear. It sits alongside advocacy, not instead of it.

Where to start in Arizona

For safety planning, the hotline numbers below are the place to start. For the longer work, The Healing Effect sees clients across Arizona, in person and by telehealth for clients located in the state. Read more on our trauma and PTSD page, or reach us through our contact form. You do not have to have left, or be certain you are going to, before you talk to someone.

What is trauma bonding?

It describes an emotional attachment that forms to the person causing harm, rather than a bond formed by shared hardship. Psychologists Donald Dutton and Susan Painter identified two conditions behind it: a marked power imbalance and abuse that is intermittent, alternating with positive or neutral treatment. It is a descriptive term, not a diagnosis in either diagnostic manual.

Why is it so hard to leave someone who hurt me?

Partly because of the pattern rather than the severity. Dutton and Painter found that intermittency of abuse and shifts in the power balance predicted continuing attachment better than how severe the violence was. Intermittent reward is also the schedule most resistant to extinction, which is part of why hope keeps regenerating.

Do survivors really leave seven times on average?

That figure appears on the National Domestic Violence Hotline's website without a citation and does not appear on its own sourced statistics page. We could not trace it to a published study. What the research does support is that leaving is a process involving several shifts in thinking, without putting a number on the attempts.

Is trauma bonding the same as Stockholm syndrome?

No. Stockholm syndrome also has a weak evidence base: a systematic review found only 12 relevant papers, mostly single case reports, no validated diagnostic criteria, and no recognition in the international classification systems.

Is it dangerous to leave?

The period around separation carries elevated risk. A multisite case control study found that women who had separated from an abusive partner after living together faced more than three and a half times the risk of being killed by that partner, rising roughly nine fold where the partner was highly controlling. That is a reason to plan with an advocate rather than a reason to stay.

The Healing Effect is not a crisis service. If you are in crisis or thinking about harming yourself, call or text 988 to reach the Suicide and Crisis Lifeline, available 24 hours a day, 7 days a week.

If you are being hurt or controlled by a partner, the National Domestic Violence Hotline is free and confidential, 24 hours a day: call 1-800-799-7233, or text START to 88788. In Arizona, the Sexual and Domestic Violence Helpline run by the Arizona Coalition to End Sexual and Domestic Violence can be reached toll free at 1-800-782-6400 during business hours.

Sources:

Dutton DG, Painter S. Traumatic bonding: the development of emotional attachments in battered women and other relationships of intermittent abuse. Victimology. 1981;6:139 to 155, as described in the International Encyclopedia of the Social Sciences (power imbalance and intermittent abuse as the two conditions).

Dutton DG, Painter S. Emotional Attachments in Abusive Relationships: A Test of Traumatic Bonding Theory. Violence and Victims. 1993;8(2):105 to 120 (75 women; attachment persisted after leaving; intermittency of abuse and changes in the power balance were the strongest predictors, accounting for 55 percent of the variance at six months).

Reid JA, Haskell RA, Dillahunt-Aspillaga C, Thor JA. Contemporary Review of Empirical and Clinical Studies of Trauma Bonding in Violent or Exploitative Relationships. International Journal of Psychology Research. 2013;8(1):37 to 73 (description of the two conditions).

Casassa K, Knight L, Mengo C. Trauma Bonding Perspectives From Service Providers and Survivors of Sex Trafficking: A Scoping Review. Trauma, Violence, and Abuse. 2022;23(3):969 to 984 (no single agreed definition; no included article addressed how trauma bonds could be severed and replaced with healthy attachments).

Namnyak M, Tufton N, Szekely R, Toal M, Worboys S, Sampson EL. Stockholm syndrome: psychiatric diagnosis or urban myth? Acta Psychiatrica Scandinavica. 2008;117(1) (12 relevant papers, mostly case reports; no validated diagnostic criteria; not recognised in international classification systems).

Williams WA. Partial Reinforcement Extinction Effect. In: Encyclopedia of Animal Cognition and Behavior. Springer, 2019 (periodic reinforcement produces behaviour more resistant to extinction than continuous reinforcement).

Miller SA, Lund C, Weatherly J. Applying Operant Learning to the Stay-Leave Decision in Domestic Violence. Behavior and Social Issues. 2012;21(1):135 to 151.

Walker LE. The Battered Woman. 1979, and Promising Futures, Futures Without Violence: The Cycle of Violence and Its Use to Understand Domestic Violence (three phase model; unrepresentative single state sample; Walker's own caution about generalizing; many survivors report no honeymoon phase; no phase is a safer time to leave).

Anderson DK, Saunders DG. Leaving an Abusive Partner: An Empirical Review of Predictors, the Process of Leaving, and Psychological Well-Being. Trauma, Violence, and Abuse. 2003;4(2):163 to 191 (survivors typically undergo several shifts in thinking about the abuse before leaving permanently; no number of attempts given).

The National Domestic Violence Hotline. Supporting Someone Who Keeps Returning to an Abusive Relationship, and 50 Obstacles to Leaving (the seven times figure appears without a citation; obstacles list credited to Sarah M. Buel).

Campbell JC, Webster D, Koziol-McLain J, et al. Risk Factors for Femicide in Abusive Relationships: Results From a Multisite Case Control Study. American Journal of Public Health. 2003;93(7):1089 to 1097 (separation after living together, adjusted odds ratio 3.64; combined with a highly controlling partner, adjusted odds ratio 8.98).

National Institute of Justice. Practical Implications of Current Domestic Violence Research. 2009 (those separated or divorced experienced more non fatal domestic violence than those together).

DV RISC, Center for Justice Innovation. Lethality Risk Factors Series: Recent Separation.

Golding JM. Intimate Partner Violence as a Risk Factor for Mental Disorders: A Meta-Analysis. Journal of Family Violence. 1999;14(2):99 to 132 (weighted mean prevalences of 63.8 percent for PTSD and 47.6 percent for depression).

World Health Organization. Violence against women fact sheet (consequences including depression, post traumatic stress and anxiety disorders, sleep difficulties, alcohol use disorders and suicide attempts).

Tolmie J, Smith R, Wilson D. Understanding Intimate Partner Violence: Why Coercive Control Requires a Social and Systemic Entrapment Framework. Violence Against Women. 2024;30(1).

US Department of Veterans Affairs, National Center for PTSD. Intimate Partner Violence (it is not your fault).

Zagaria A, Baggio T, Rodella M, Leto K. Toward a definition of Attachment Trauma: integrating attachment and trauma studies. European Journal of Trauma and Dissociation. 2024;8(3) (fright without solution; the attachment system monitors availability when danger is perceived). This work concerns caregiver attachment rather than adult partners.

Rivas C, Vigurs C, Cameron J, Yeo L. A realist review of which advocacy interventions work for which abused women under what circumstances. Cochrane Database of Systematic Reviews. 2019 (components of advocacy; needs sufficient duration and alignment with the woman's own goals).

Hameed M, O'Doherty L, Gilchrist G, et al. Psychological therapies for women who experience intimate partner violence. Cochrane Database of Systematic Reviews. 2020 (probably reduce depression at 6 to 12 months; may reduce anxiety; less clear for PTSD; therapy alone may not address ongoing safety and support needs).

New York State Office for the Prevention of Domestic Violence. Safety Planning (what a safety plan covers; working with an advocate is the best way to develop one).