Trauma Bonding: What the Term Actually Rests On, and Where the Seven Stages Came From
The seven stages appear in no study. Four measures of trauma bonding do exist, and no page on the first results page mentions any of them.
Before the article. If you are reading this because of what is happening to you now rather than out of curiosity about a term, that is the more urgent question and it has a different answer. Monitoring, having to account for ordinary hours, being discouraged from seeing friends and family, control of money or transport, threats, intimidation or any physical harm are warning signs of abuse, and the second section of this page lists them in full. Find A Helpline, in the sources at the bottom, lists free and confidential services by country. In the United States the National Domestic Violence Hotline is 1-800-799-7233, or text START to 88788, 24 hours a day.
With that said. Search for trauma bonding and you get a remarkably consistent answer: a definition, a cycle, seven stages you can locate yourself in, and a list of steps for breaking free. The pages giving it are serious ones — a hospital system, two health publishers, an encyclopaedia, a therapy directory. We checked what they rest on, and the results were not what we expected in either direction.
The seven stages appear in no indexed study. On 29 October 2026 a PubMed search pairing this construct with any word beginning “stage” returned zero records. The page that ranks for them cites five works, none about stages, and misattributes the theory to a 1997 trade book rather than to Dutton and Painter, who published it sixteen years earlier. The hostage analogy underneath the whole genre comes from a literature whose best dataset — over 1,200 FBI incidents — found the reaction in fewer than one case in ten, and whose own account says that abuse makes it extremely unlikely and tends to end it.
And then the other direction, which is the part no page on that first results page carries. There are four measures of trauma bonding, two of them with published psychometrics across more than a thousand people in two countries. There is one study of 354 people in currently abusive relationships finding that childhood maltreatment and attachment insecurity predict it over and above romantic love. And the newest work, published in September 2026, argues the attachment is not something that develops in a victim but something a perpetrator manufactures.
What follows is twenty-one sections: where the term came from, what the one empirical test actually reported, what the research literature contains when you count it, which measures exist, who has been studied, what the FBI's own figures say, and what nobody has studied at all — including, according to the only systematic review, how to break one. Our editorial policy explains why every claim here carries a source you can open.
Where the term came from, and it was not 1997
Almost every page on this subject opens by telling you who invented the idea, and a good share of them name the wrong person.
The theory is Donald Dutton and Susan Painter's. They published it under the title Traumatic Bonding: the development of emotional attachments in battered women and other relationships of intermittent abuse, in a journal called Victimology: An International Journal, volume 6, issues 1–4, pages 139–155. Patrick Carnes wrote The Betrayal Bond sixteen years later, in 1997. It is a trade book from a self-help publisher, it is where the phrase betrayal bond comes from, and it is not where the construct comes from.
You can watch the attribution slip in public. The page that ranks for the 7 stages of trauma bonding states that trauma bonding “as a theory was devised by Dr. Patrick Carnes in 1997” — and then lists Dutton and Painter 1993 in its own reference section, four lines further down. Another page high in the same results says the term was “coined by Patrick Carnes, building on earlier research by Donald Dutton and Susan Painter,” which gets the order right and the authorship backwards.
Now the part that an honest version of this page has to include. That founding paper is not indexed in Crossref or in PubMed, it has no DOI, and we could not read it. Sources that cite it do not even agree on the year: some give 1981, some 1983, and at least one encyclopaedia entry gives both in different footnotes of the same article. So the origin document of a construct that currently fills the first page of results for millions of searches is a forty-five-year-old article in an unindexed journal.
What can be read is the test they ran twelve years later, and it is more interesting than the summaries suggest. The pattern — a term with a confident origin story that nobody has checked — is the same one behind the three-day rule in dating, which turns out to come from a 1996 comedy. Our editorial policy explains why every claim on this site carries a source you can open.
Before the rest of this page: if this is your situation
This article is about where a word came from and what has been measured. If you are reading it because of what is happening to you right now, that question is more urgent than any of it, and it has a different answer.
The US National Domestic Violence Hotline publishes a list of warning signs, and it is specific rather than atmospheric: extreme jealousy of your friends or of time spent away; being discouraged or prevented from seeing friends and family; having to account for where you were and who you spoke to; your phone or location being checked; decisions about money, work or transport being made for you; being told that your reaction is the real problem; threats, intimidation, or any physical harm. The Hotline notes that even one or two of these is a warning sign, and that many abusive partners seem ideal early on, with the signs emerging and intensifying as the relationship goes.
Find A Helpline, in the sources at the foot of this page, lists free and confidential services by country. In the United States the National Domestic Violence Hotline is 1-800-799-7233, or text START to 88788, around the clock.
One more thing belongs here rather than in a footnote, because the advice genre rarely says it. Leaving is not automatically the safe moment. In the eleven-city study comparing 220 intimate partner femicide cases with 343 abused control women, estrangement from a controlling partner was among the risk factors that survived multivariate analysis. That is not an argument for staying. It is an argument for planning with somebody who does this for a living, rather than acting on a web page — and it is set out in more detail where the line between jealousy and control is drawn.
Nothing below is a diagnosis, a treatment, or a substitute for a person. The rest of this page is an audit of a vocabulary, and a vocabulary is not what gets anybody out.
The condition the popular version quietly drops
Dutton and Painter's theory has three conditions, not two. As Mags Lesiak and Loraine Gelsthorpe restate them in their 2026 study: a power imbalance, intermittent reinforcement of good and bad treatment, and captivity.
Count how many of the three survive the trip to a listicle. Power imbalance, yes. Intermittent reinforcement, always — it is the slot-machine paragraph every page has, and it is also the explanation routinely borrowed for breadcrumbing, where there is no captivity, no power imbalance, and no test of the mechanism at all. Captivity disappears without comment.
That omission is not a detail, because captivity is what made the whole analogy work in the first place. The model was built by reasoning across from hostages, prisoners and abducted children to women in violent homes, and the bridge between those situations was that the person could not simply walk out. Remove it and the theory is being asked to explain something it was not built to explain: why somebody who can leave does not.
The honest objection is the one Lesiak and Gelsthorpe raise directly. Captivity and power imbalance are present in nearly every long-term abusive situation, yet only a small minority of people in those situations show anything resembling the hostage reaction — a figure this page gets to in a moment, and it is much smaller than the genre implies. A condition that is nearly always present cannot explain an outcome that is rare. And the theory has nothing at all to say about the people with no captivity and no power imbalance who nonetheless report an intense attachment to someone who hurt them.
So the version you are being given is the original model with its most restrictive ingredient removed, which is exactly the edit that makes it apply to everybody. That is worth knowing before the stages arrive. The same move — a construct widened until it covers the whole audience — turns up in what has actually been measured about falling in love fast, where the thing being widened is at least an instrument somebody validated.
The one empirical test of the theory, at full size
Twelve years after the theory, Dutton and Painter tested it, and the paper is worth having in full rather than as the single clause everyone quotes.
They assessed 75 women who had recently left abusive relationships, 50 of them relationships in which physical violence had occurred. The assessment was in-depth — interviews plus questionnaires — and it was run twice: immediately after separation, and again six months later.
The findings, in their own terms. The extremity of intermittent maltreatment and the power differential between the partners predicted how strongly a woman remained attached to her former partner, how many trauma symptoms she reported, and her self-esteem. Those three measures were significantly intercorrelated at each time point, and each one at the first assessment correlated with its own counterpart at the second. And the headline number: the relationship variables — total abuse, intermittency of abuse, and power differentials — accounted for 55% of the variance in the attachment measure at six months.
Fifty-five per cent is a large figure, and it is the strongest result the construct has. It is also the point at which the limits have to be said out loud, because the paper is cited everywhere and read almost nowhere.
There is no comparison group. All 75 women came from abusive relationships. Nothing in this design can tell you whether abuse produces a stronger attachment than its absence, which is the claim the entire popular literature makes. It can tell you that, among women leaving abusive relationships, those who reported more extreme and more intermittent maltreatment reported more attachment afterwards. That is a different and much narrower sentence.
Everything else is self-reported and retrospective, the sample is small and entirely female, and the design is correlational at two points rather than experimental. This remains the best direct evidence the theory has, thirty-three years later. The brief version of it, in the context of advice that recommends running that pattern on someone deliberately, is in what the research says about the 72-hour rule.
The number from that study that nobody quotes
Here is a result sitting in the same abstract as the famous one, and it points the other way from almost every page on the subject.
After six months, attachment had decreased by about 27%.
Read that against the register the genre uses. Trauma bonds are routinely described as near-unbreakable, as chemical, as addiction that requires a specific protocol to escape. The single study that actually measured the thing over time found that, among women who had left, felt attachment to the former partner dropped by roughly a quarter in half a year — without any intervention being tested, studied or reported.
Two cautions, both of which matter. A 27% drop is not disappearance; attachment was still there at six months, and the relationship variables still explained most of the variance in it. And these were women who had already left, which is a selected group: the study cannot tell you anything about what happens to attachment in people still in the relationship.
But the direction is the point. The one longitudinal observation in this literature shows the thing weakening with time and distance, which is unremarkable as a fact about attachment and remarkable only because of how it is usually described. Nobody has published a study showing that a trauma bond is more durable than attachment to an ordinary ex-partner, because nobody has run the comparison.
If what you are actually asking is how long it takes to stop missing someone, the honest answer is that it is a worse-studied question than it should be, and that what is known about recovery timelines, the difference between missing a person and missing the uncertainty, and why neither is a verdict on you is in what the evidence says about having moved on. More in this vein under Mindset.
The seven stages have no citation, and the real seven-stage model is about the abuser
The most specific page in these search results is the one promising seven stages: love bombing, trust and dependence, criticism and devaluation, manipulation and gaslighting, resignation and giving up, loss of self, and emotional addiction to the cycle.
It carries five references: a 2008 book on coping with trauma, Carnes's trade book, Dutton and Painter 1993, a 2021 book chapter on domestic violence in the Caribbean, and a 2020 paper on the therapeutic alliance across treatment. Not one of them is a study of stages. The staged sequence is presented as a framework the reader can locate themselves in, and no source in the piece supports the sequence, the number, or the order.
That is checkable against the literature rather than taken on faith. On 29 October 2026, a PubMed search for trauma bonding, trauma bond or traumatic bonding in a title or abstract, combined with any word beginning stage, returned zero records. There is no staged model of trauma bonding in the indexed research, with seven steps or any other number.
The twist is that a seven-stage model does exist in this area, and it is about the other person. Lesiak and Gelsthorpe cite Sinnamon's Seven-Stage Model of Adult Grooming — victim selection, gaining trust, meeting needs, priming, instigating abuse, exerting control — and use it to describe what the perpetrators in their interviews did. We have not read that chapter and report it as they cite it. But note the difference, because it is the whole argument of this page: the seven stages in the search results are stages a victim passes through; the seven-stage model in the literature is a sequence of things an abuser does.
Invented sequences with a fixed number are a signature of this whole genre rather than an accident of one topic. The same audit applied to a dating rule with digits in it is in where the 3-6-9 rule came from, and the answer there was also nowhere.
What the research literature on this subject actually contains
It is possible to measure the size of a field rather than assert it, and the result reframes the subject in a way no page on the first results page does.
Searching PubMed on 29 October 2026, in titles and abstracts:
- trauma bonding: 19 records.
- trauma bond: 4.
- traumatic bonding: 7.
- The three together, deduplicated: 27 records, the earliest from 1992.
- Stockholm syndrome: 36.
- love bombing, anywhere in the record rather than just the abstract: 2.
- narcissistic abuse: 1.
Now the comparison that matters:
- gaslighting: 153.
- coercive control: 455.
So the explanatory term that organises the entire first page of results has twenty-seven papers behind it, while the term for the behaviour being described has four hundred and fifty-five, and almost never appears in the headline.
Two honest qualifications. PubMed is a biomedical index and under-covers criminology, social work and family-violence journals, where a good deal of this work lives; the true counts across all databases are higher. And a count is not an argument — a small literature can be correct and a large one wrong.
What a count does establish is proportion. When a phrase has two indexed records and is presented to you as the opening stage of a documented process, you are being handed a vocabulary rather than a finding. The same arithmetic, run on the fight-flight-freeze-fawn diagram that circulates alongside this one, produced the same answer, and that audit is in what has actually been measured about being too accommodating.
Nearly all of the evidence is about trafficking, not about a partner
Read the twenty-seven papers by subject and the field stops looking like the search results entirely.
The modern core of it is a body of work on sex trafficking. Kaitlin Casassa, Logan Knight and Cecilia Mengo's scoping review searched ten databases and found fifteen articles meeting criteria. Casassa and colleagues then interviewed ten service providers — licensed social workers and counsellors — about how they define and observe trauma bonds in clients, and separately interviewed nineteen female survivors who were participants in or graduates of a human trafficking specialty court docket, twice: once about experiencing trauma bonding, once about healing from it.
Those studies are careful and they are small, qualitative and specific. The features the scoping review identified were an imbalance of power favouring the trafficker, the trafficker's deliberate use of positive and negative interactions, the victim's gratitude for the positive ones and self-blame for the negative, and the internalisation of the perpetrator's view. The service providers' phrase for it, which became the title of their paper, was that he loves her hard and then abuses her hard.
Here is why this matters for anyone arriving from a search box. The literature with the most empirical content on trauma bonding is about commercial sexual exploitation by a trafficker. It is being used to explain ordinary abusive relationships, long-distance situationships, difficult families, and in some posts simply an ex who was inconsistent. Findings from nineteen trafficking survivors in one court docket do not automatically transfer to someone who cannot stop texting a person who was unkind to them, and nobody has shown that they do.
This is the same transfer problem that runs through most of the genre: a result obtained in one population gets repeated until the population drops off. The version of it that this site has written about at most length is in what the phone-behaviour research really measured.
There is a measure for this. In fact there are four
The line you will read everywhere is that trauma bonding is not a diagnosis, is not in the DSM, and has no measure. The first two are true. The third is false, and the gap is checkable.
Dee Graham, Edna Rawlings and seven colleagues published a scale in 1995 — a 49-item instrument for identifying Stockholm syndrome reactions, a term they explicitly treated as interchangeable with traumatic bonding, in young women in heterosexual dating relationships. Factor analysis produced three factors: Core Stockholm Syndrome, Psychological Damage, and Love-Dependence. The scale and its factors had excellent internal consistency and good test-retest reliability, correlated negatively with a social-desirability measure, and correlated positively with the Impact of Event Scale, a borderline personality measure, the Passionate Love Scale, and the verbal-aggression and violence subscales of the Conflict Tactics Scales. It has been in the PsycTESTS database for thirty years. A Spanish translation was validated in Mexican women in 2019.
Joan Reid published the Trauma Bonding Scale for Adults in 2023, developed and evaluated in the context of sex trafficking, deposited in PsycTESTS the same year, validated in 619 young adults in the United States, and then separately in 538 young adults in Kenya.
And Reid's own review names two more instruments that contain a trauma-bonding component without being built for it: a subscale of the Post-Traumatic Stress Index, and an interpersonal-disturbance subscale of the well-established Trauma Symptom Inventory-2.
We went through the ten highest-ranking pages for this search on 29 October 2026 and counted. Carnes is named on five of them, Dutton on four, the scoping review on three. Not one names the Trauma Bonding Scale for Adults, Graham's scale, or either of their authors. The single acknowledgement that any measure exists is one clause in the Wikipedia entry, referring to “the psychometric scale for Stockholm syndrome” without giving its name, its author, its year or its length. A reader is told there is no way to measure the thing, by pages that have not looked.
Our own view of what a score is worth is on the page of our attachment style test, which says in its own copy what a result like that cannot tell you.
And the person who built the newest one says none of them is finished
Having found the measures, the honest thing is to say what their own author thinks of them, and she is blunter than this site would have dared to be.
In a 2024 chapter titled, plainly, Measures of Trauma Bonding, Reid writes that despite empirical evidence of trauma bonding among survivors of various forms of interpersonal violence, no measure of trauma bonding has been fully developed, evaluated, and standardised. That is the author of the newest instrument, in a textbook chapter about instruments, saying that the field does not yet have a finished one.
So the position is more interesting than either side of the usual argument. It is not true that nothing has been measured — four instruments exist, two of them with published psychometrics on more than a thousand people across two countries. It is also not true that the construct is settled enough to put a number on somebody. Both of those sentences are correct at once, and the pages in these search results carry neither.
What that combination implies for a reader is specific. A questionnaire that tells you whether you are in a trauma bond is not reporting a validated threshold, because no validated threshold has been published. The scales that exist were built to study groups — to ask whether a sample of trafficking survivors differs from another sample, whether a construct correlates with post-traumatic symptoms — not to classify an individual. The distance between those two uses is the distance between research and a verdict.
This is the same problem, in a different costume, as the one behind the questionnaire that tells people they are people-pleasers: its first proper psychometric validation appeared in 2025, and the version circulating online descends from a paperback with cut-off scores that were never calibrated. That audit is in what the evidence says about being too accommodating, and the pattern should be familiar by the third time it happens. More of it under Mindset.
The one study of who develops it, and what it controlled for
There is exactly one study in this literature that asks the question most readers are actually asking — why some people in abusive relationships form this attachment and others do not — and it deserves to be known.
Emma Shaughnessy, Raluca Simons, Jeffrey Simons and Hannah Freeman tested a path model in a high-risk sample of 354 participants who were currently in abusive relationships. They looked at childhood maltreatment and attachment insecurity as risk factors for traumatic bonding, and at whether traumatic bonding was associated with post-traumatic stress symptoms.
The results, as published. Childhood maltreatment and attachment insecurity both significantly predicted traumatic bonding over and above the effects of age, gender, and romantic love. Attachment insecurity moderated the first relationship: at higher levels of insecurity, the link between childhood maltreatment and traumatic bonding was stronger than at average or lower levels. And traumatic bonding was positively associated with post-traumatic stress symptoms. The authors describe it as the first study to examine a complex model of risk factors for traumatic bonding.
The control for romantic love is the part worth pausing on, because it is the methodological answer to the most obvious objection. If trauma bonding were just love in a bad relationship, statistically controlling for love would make the construct disappear. It did not. That is the strongest available evidence that the term is pointing at something, and it is a far better argument for the construct than anything on the first page of search results, none of which cites it.
The limits are the ordinary ones and the authors do not hide them: a cross-sectional path model cannot establish direction, self-report throughout, and a high-risk convenience sample. We could not read beyond the published abstract — the paper is paywalled, and the figures above are what its abstract reports. Why childhood maltreatment and attachment insecurity keep appearing together, and how little the second one is destiny, is in the longitudinal evidence on attachment styles changing.
The finding that has repeated across two countries
One result in this field has now been obtained twice, in two very different places, which in a literature of twenty-seven papers is close to a luxury.
Tiffany Chenneville, Joan Reid and colleagues validated the Trauma Bonding Scale for Adults in 619 young adults aged 18 to 29 in the United States, then in 538 young adults aged 18 to 29 in Kenya, using the same anonymous online survey each time: demographics, the scale, and a twelve-item open-source measure of post-traumatic stress disorder. In 2025 they published the comparison.
Post-traumatic stress correlated with and predicted trauma bonding in both samples — despite the two samples differing from each other in their levels of both post-traumatic stress symptoms and trauma bonding symptoms. The authors' framing is forensic: valid assessment is needed across settings in order to identify and address the needs of victims from different cultures.
Two things are worth drawing out, and the second cuts against the first.
The replication is real, and it is the kind this field badly needs. A construct that only behaves consistently in one country is usually a construct about that country.
But look at who was measured. These are general young-adult samples recruited through an online survey panel, not trafficking survivors, not people in violent relationships, not a clinical population. The psychometric work establishing that a trauma-bonding instrument behaves sensibly was done largely on people who were not selected for having experienced any of this. That is standard practice in scale validation and it is not a criticism of the work. It does mean the sentence the scale is validated carries less than a reader would assume, and the authors of the measurement chapter say as much themselves.
And a correlation between a bonding score and a post-traumatic stress score cannot tell you which produced which, or whether a third thing produced both.
Stockholm syndrome was never a diagnosis either
Nearly every page on trauma bonding reaches for Stockholm syndrome as its explanatory parent, usually in a sentence beginning you may have heard of. The parent is in worse shape than the child.
In 2008, M. Namnyak, N. Tufton, R. Szekely, M. Toal, S. Worboys and Elizabeth Sampson published a systematic review in Acta Psychiatrica Scandinavica with a title that does not hedge: 'Stockholm syndrome': psychiatric diagnosis or urban myth? They searched PubMed, EMBASE, PsycINFO and CINAHL, and compared the features of cases widely reported in English-language media to see whether a recognisable syndrome emerged.
What they found: twelve papers met the inclusion criteria. The existing literature consists mostly of case reports. There is ambiguity in how the term is used. No validated diagnostic criteria have been described. Among the five media cases they examined, four common features appeared. Their conclusion was that there is little published academic research on the subject, that similarities between well-publicised cases may reflect reporting and publication bias, and that the diagnosis is not described in any international classification system.
That is the state of the term in 2008, and it has not been superseded. One proposal since then has been to replace Stockholm syndrome with a different concept altogether — that attempt, and the problem with the theory it leans on, is set out in the audit of the fight-flight-freeze-fawn model rather than repeated here.
So the chain the search results offer you runs: a staged model with no study behind it, resting on a construct with four unfinished measures, explained by reference to a syndrome that a systematic review found to be mostly case reports with no validated criteria. Each link is presented to the reader as the solid part that justifies the next one.
And the number underneath all of it is the subject of the next section, because somebody did count.
Somebody did count, and the figure is 92%
The one hard number in this whole area comes from an unlikely place: the FBI's own in-house magazine, in July 1999. We downloaded the issue and read the article.
G. Dwayne Fuselier, writing in the FBI Law Enforcement Bulletin, reports that according to the FBI's Hostage/Barricade System — a national database then containing data from over 1,200 reported federal, state and local hostage and barricade incidents — 92 per cent of victims reportedly showed no aspect of the Stockholm syndrome. When victims who showed only negative feelings toward law enforcement were counted as well, usually from frustration at how slowly negotiations were going, the figure rises to 95 per cent.
His own summary of the state of the idea is sharper than anything a cautious website would write: over the preceding twenty-five years, he says, the Stockholm syndrome has been “overemphasized, overanalyzed, overpsychologized, and overpublicized,” and the occasions where it actually occurs remain exceptions to the rule. Most hostages, he writes, do not identify or sympathise with the hostage taker and do not see the police as adversaries.
He also names why the misconception persists, and the list is uncomfortably applicable to the modern version. People find it provocative: in the original 1973 bank robbery a hostage was alleged to have had a sexual relationship with one of the captors. Training reviews naturally focus on extraordinary incidents. Exceptions are more interesting, prompt more discussion, and therefore seem more prevalent.
So the figure that should accompany every use of this analogy is that in the only large dataset anybody has published, the reaction appeared in fewer than one case in ten. It has not been updated since the late 1990s, which is itself worth stating: there is no modern prevalence estimate, for the hostage version or for the relationship version.
This is the second time on this page that a number has gone the opposite way from the story built on top of it, and it will not be the last.
The precondition that gets inverted when the model is imported
The same FBI article contains a passage that, as far as we can tell, nobody writing about trauma bonding has quoted, and it undoes the analogy at its foundation.
Setting out what has to be true for the syndrome to have the potential to develop at all, Fuselier lists three conditions established by FBI researchers after interviewing numerous flight attendants who had been through hijackings. First, a significant length of time must pass. Second, the hostages and hostage takers must remain in contact rather than the hostages being hooded or isolated. And third — the hostage takers must treat the hostages kindly, or at least not physically abuse or verbally threaten them.
He goes further in the conclusion. If the victim receives or witnesses physical or psychological abuse, the syndrome is extremely unlikely to occur. Even if some aspect of it has developed, it can and usually will cease if the captors verbally or physically abuse any of the hostages.
Read that beside the popular claim. Trauma bonding is explained to readers as Stockholm syndrome arriving in a relationship, and as the mechanism by which abuse itself manufactures attachment. The source model says the opposite about its own subject matter: abuse makes the reaction extremely unlikely, and tends to end it where it has started.
Be precise about what this does and does not establish, because overstating it would be the same error in the other direction. It does not show that people do not become attached to partners who hurt them — they plainly do, and the rest of this page is about the evidence for it. What it shows is that the hostage literature cannot be the source of that explanation. The analogy is doing rhetorical work it was never entitled to do, and it is doing it in the one direction its own authors ruled out.
Which leaves the attachment real and the explanation vacant. The field's recent attempts to fill that vacancy are in the sections below, and they point at the other person rather than at the reader.
Where the folk explanations came from, and what they share
Before the current vocabulary there were three others, and Lesiak and Gelsthorpe's account of them is the most useful paragraph written on this subject in years.
The explanations offered over the past half-century for why a woman stays with or returns to a man who hurts her were, in order: that she enjoys it — masochism; that she is too helpless to leave — learned helplessness; and that she enables his behaviour for her own gratification — codependency.
Now their provenance, which the authors lay out without ceremony. Masochism comes from Freud. Learned helplessness comes from experiments on electrocuted dogs. Codependency comes from stories told in Alcoholics Anonymous. None of the three was built on research into the experiences of the women it was describing.
And they share a structure, which is the authors' actual point. Every one of them locates the cause inside the victim and attaches a defect to her personality — masochistic, helpless, codependent. None of them examines the role of the other person in creating or maintaining the attachment. Their analogy for how strange this is: if road accidents were investigated this way, drivers would rarely be looked at, on the assumption that pedestrians are inherently problematic.
Now apply that test to the seven stages. Love bombing, trust and dependence, criticism, gaslighting, resignation, loss of self, emotional addiction. It is a sequence of things that happen to and inside the reader, with an abuser present as weather. The fourth folk explanation has the same shape as the first three, and it is the one currently being handed out at the top of the search results.
The general version of this — being given a vocabulary in which the problem is your nervous system rather than someone else's conduct, and how long that framing keeps people in place — is argued at more length in the piece on love and anxious attachment.
The 2026 study that turns the construct around
The most recent study in this literature was published in September 2026 and sets out to repair precisely that. It is open access, and we read it in full rather than from the abstract.
Mags Lesiak and Loraine Gelsthorpe, at the Institute of Criminology in Cambridge, interviewed 18 women who had experienced repeat domestic abuse and sustained attachment to their perpetrators. The recruitment criterion is what makes the study unusual. All participants were financially independent, often lived separately from their partners, faced no physical captivity and no threats on leaving, and were screened out if they reported any other reason for staying — financial, cultural, social, or related to mental health or addiction. Thirty-four women came forward; eighteen met those conditions and were interviewed for two to three hours each.
In other words, the sample was built to isolate the attachment from every circumstance that normally explains staying. The compulsion was still there. Three of the women had moved to other cities to keep themselves away; one described the separation as worse than the abuse.
The analysis names the mechanism attachment weaponisation, and identifies a perpetrator profile: volatility, manipulativeness, performative vulnerability. Three themes carried it — the two-faced soulmate, shared childhood trauma, and a pull the women themselves described in the language of addiction and craving. The shared-trauma theme is the sharpest. Every participant reported adverse childhood experiences without being asked about them directly, and every one said her partner had them too; the paper then documents that disclosure being used against them — the partner recentring his own suffering, or repeating hers in front of friends, which one woman read afterwards as a way of separating her from them.
The authors' conclusion is a single sentence: trauma bonding is not a symptom of weakness, it is a strategy of control. Not an attachment that happens to form, but one that somebody manufactures.
What that study can and cannot carry
Having given a paper that much space, the limits belong in the same type size, and most of them are visible in its own methods section.
Eighteen women, all in the United Kingdom, all self-selected. They were recruited through social media advertising and through one domestic violence charity in Nottingham, and the advertisement specified the inclusion criteria — so participants knew in advance what sort of experience was being sought. That is a known route to confirmation.
No standardised instruments were used. Domestic violence and adverse childhood experiences were both assessed from the interview narratives rather than from questionnaires, which the authors state explicitly. The strength of that choice is depth; the cost is that nothing here can be compared with anybody else's numbers.
One researcher conducted every interview and performed the analysis. Thematic saturation is reported at twelve transcripts of the eighteen.
And the screening criterion that makes the study valuable also bounds it. By design this sample excludes everyone whose situation involves financial dependence, immigration status, children, threats, or physical constraint — which is to say most people in abusive relationships. The finding is that captivity is not necessary for the attachment to form. It is not a finding about how common any of this is, and a qualitative study cannot produce one.
So what survives is a shift in where the explanation points rather than a measurement. That is still worth a great deal in a field whose last four explanations all pointed at the person being hurt — and it is the kind of claim that can be checked by somebody else, which the staged model cannot.
The habit of reading a qualitative paper for what it establishes rather than for its most quotable line is the same one this site applies to the research on supporting an avoidant partner, where the quotable version of the advice turned out to point at the worst measured outcome.
Trauma-coerced attachment, and why a renaming is not pedantry
There is a parallel attempt to fix the same problem by changing the name, and it has a practical object that is easy to miss.
Kendra Doychak and Chitra Raghavan proposed trauma-coerced attachment in the Journal of Human Trafficking in 2018, in a paper whose title quotes a participant: No voice or vote. The aim, as later authors describe it, was to offer a thorough operational definition, to examine how coercion specifically contributes to the attachment, and to move the explanation away from victim-blaming framings. In 2023 the same authors went further, arguing in the European Journal of Trauma and Dissociation that the phenomenon maps onto an existing DSM-5 category — identity disturbance due to prolonged and intense coercive persuasion.
The word doing the work is coerced. It makes the other person's conduct part of the name of the thing, which trauma bonding does not: a bond is something two people have, and the passive construction leaves the question of who built it unasked.
And the stated motivation is not conceptual tidiness. Doychak and Raghavan's argument for aligning the phenomenon with DSM criteria is to open legal avenues for victims who face criminalisation — people prosecuted for acts committed under a trafficker's control, for whom a recognised diagnostic category is the difference between a defence and a sentence. That is what a name is for in this corner of the field.
It has not been settled. Three terms are now in circulation for overlapping phenomena — traumatic bonding, trauma bonding, trauma-coerced attachment — alongside the older Stockholm syndrome, and the scoping review found ambiguity in how all of them are used. A reader is entitled to notice that a field still arguing about what to call something is not a field in a position to tell anybody which of seven stages they are in.
Meanwhile the construct that has been measured and legislated for most, and which almost never appears in these search results, is the subject of the next section.
The construct that does have a finished instrument
If the practical question is what to call this in a way that opens doors — with a service, a lawyer, a court — the answer is not trauma bonding, and it has existed for years.
Coercive control has 455 records in PubMed on the title-and-abstract search run for this article, against twenty-seven for the whole trauma-bonding family. It has a 2017 review in Aggression and Violent Behavior devoted specifically to its conceptualisation and measurement. And it has an instrument that was finished.
Peter Lehmann, Catherine Simmons and Vijayan Pillai validated the Checklist of Controlling Behaviors in 2012: an 84-item assessment covering ten subscales — physical abuse, sexual abuse, male privilege, isolation, minimising and denying, blaming, intimidation, threats, emotional abuse, and economic abuse. It was administered to 2,135 women taking refuge in a domestic violence shelter, and principal component analysis of the subscales produced ten factors with goodness-of-fit values above the normative level of .90. It is in PsycTESTS. It is designed for exactly the population that searches for trauma bonding.
We are not telling anyone to reach for a checklist instead of a person. The point is narrower and it is about vocabulary. Trauma bond describes the feeling and, as the previous sections show, descends from a staged model nobody validated and a hostage analogy that excludes abuse. Coercive control describes conduct, it is a recognised pattern of abuse with a measured structure, and in a number of jurisdictions it is a criminal offence under that name. One of those two terms has a definition that somebody else is obliged to act on.
The behaviours themselves — monitoring, vetting who you see, an account demanded of ordinary hours — and where the line falls between a feeling and a pattern, are worked through in the psychology of jealousy and again, for the version that runs through a phone, in what the research on back burners found.
Nobody has studied how to break one
Every page on this subject has a section on breaking the bond. Here is the sentence that section is missing, and it comes from the only systematic review of the literature.
Casassa, Knight and Mengo searched ten databases, included fifteen articles, and wrote in their published abstract: no article indicated how trauma bonds could be severed and replaced with healthy attachments. They describe the gap as representing potential targets for urgently needed interventions.
That is the state of the evidence on the most actionable-looking content in these search results. The seven steps to break a trauma bond, the thirty-day no-contact protocol, the stages of withdrawal — none of it has a study behind it. It is not that the studies are weak; it is that the review that went looking for them found none.
One piece of work has asked the question since. The same group interviewed 19 female sex trafficking survivors about how they had experienced healing from trauma bonding, and three themes emerged: building trusting and honest relationships, their relationship with themselves, and education. Read it for what it is — nineteen women, one court docket, qualitative, no comparison and no measurement of outcome — and it is still the closest thing to an answer that exists, and notably it contains no steps, no sequence and no timeline.
What can be said with more confidence is the unglamorous thing from much earlier on this page: in the one study that measured attachment over time in women who had left, it fell by about a quarter in six months, with nothing being administered. Distance and time did something. That is thin, and it is honest, and it is more than the protocols have.
For anyone whose situation is a difficult relationship rather than an abusive one — and those are different, and the difference is not a matter of degree — our free conflict style test takes about two minutes and states plainly what it cannot tell you. For the question of why leaving is hard even where nobody is being controlled, the measured answer is in why people stay in unhappy relationships.
What survives, and what to do with it
Put the audit in one place, because it is unusual for a subject to come apart this cleanly.
The seven stages are not research. Zero indexed papers pair this construct with a staged model. The page that ranks for them cites five works, none of which is about stages. The only seven-stage model in the area describes what a perpetrator does.
The attribution is commonly wrong. The theory is Dutton and Painter's, in an unindexed journal whose date is given variously as 1981 and 1983; Carnes's 1997 book is a trade book and coined a different phrase.
The theory's third condition is missing from every popular account, and it is the one that made the hostage analogy legitimate.
The hostage analogy excludes abuse. The FBI's own account says the reaction appeared in under 10% of more than 1,200 incidents, requires captors who do not abuse or threaten, and tends to stop when they do.
Something real is nonetheless being measured. Four instruments exist; one study of 354 people in abusive relationships found childhood maltreatment and attachment insecurity predicting it above and beyond romantic love; post-traumatic stress tracks it in two countries. The author of the newest scale says none of them is finished.
And the direction of explanation has started to turn. The newest work argues the attachment is manufactured rather than developed, by someone, on purpose.
What to do with that is short. If you recognise the conduct in the second section of this page, the useful next step is a helpline and not a framework — and the term that opens doors is coercive control, not trauma bond. If what you have is an attachment you cannot explain to yourself, the research says it does not make you masochistic, helpless, codependent or defective, because every one of those explanations was invented without asking anyone who had lived it. And if you want a sense of your own patterns rather than a verdict, our attachment style test is free, scores in your browser and sends nothing anywhere.
The claims this site has had to withdraw are listed on the corrections page. More of this under Mindset and Dating.
Pulling it together
The honest summary of this subject is that both confident versions of it are wrong. It is not true that trauma bonding is a documented seven-stage process with established steps for escape: no study pairs the construct with a staged model, the systematic review found no article describing how such a bond could be severed, and the hostage literature the whole analogy rests on specifically excludes abuse as a route to it. It is also not true that the term is empty. Four instruments exist, one of them validated in 619 Americans and 538 Kenyans; a study of 354 people in abusive relationships found childhood maltreatment and attachment insecurity predicting it even with romantic love controlled; and post-traumatic stress tracks it across both samples.
What has changed most recently is where the explanation points. For fifty years the available answers — masochism, learned helplessness, codependency — all located a defect inside the person being hurt, and none was built by asking her anything. The 2026 interview study built its sample specifically to exclude every external reason for staying, found the compulsion anyway, and concluded that the attachment is manufactured rather than developed.
If you take one practical thing from this page, take the vocabulary. Trauma bond names a feeling. Coercive control names conduct, has a finished 84-item instrument behind it, and in several jurisdictions is a criminal offence. One of those words obliges somebody else to act.
Frequently asked questions
What are the seven stages of trauma bonding?
They are not a research finding. On 29 October 2026, a PubMed search combining trauma bonding, trauma bond or traumatic bonding in a title or abstract with any word beginning “stage” returned zero records: there is no staged model of this construct in the indexed literature, with seven steps or any other number. The best-known version — love bombing, trust and dependence, criticism and devaluation, manipulation and gaslighting, resignation, loss of self, emotional addiction — comes from a single website, which lists five references, none of which is a study of stages. There is a seven-stage model in the adjacent literature, but it describes adult grooming, meaning a sequence of things a perpetrator does, not phases a victim passes through. The distinction matters: one framing asks what was done to you, the other asks what stage you are in.
Is trauma bonding a real diagnosis?
No, and neither is Stockholm syndrome. Namnyak and colleagues' systematic review in Acta Psychiatrica Scandinavica searched four databases, found twelve papers mostly consisting of case reports, noted ambiguity in how the term is used, and concluded that no validated diagnostic criteria have been described and that the diagnosis does not appear in any international classification system. Trauma bonding is in the same position. What does exist is measurement: a 49-item scale from 1995, a Trauma Bonding Scale for Adults published in 2023 and validated in the United States and Kenya, and trauma-bonding components inside two broader trauma instruments. The author of the newest of those writes in a 2024 review chapter that no measure of trauma bonding has been fully developed, evaluated and standardised. Both halves of that are true at once.
Is trauma bonding the same as Stockholm syndrome?
The two terms have been used interchangeably since at least 1995, when Graham and colleagues titled their scale a measure of Stockholm syndrome reactions and noted in the abstract that the phenomenon is also referred to as traumatic bonding. But the borrowing creates a specific problem. In the FBI's account of the hostage version, the reaction requires captors who treat hostages kindly or at least do not abuse or threaten them, and if a victim receives or witnesses physical or psychological abuse the syndrome is described as extremely unlikely to occur — and as usually ceasing if it had started. So the model is being used to explain how abuse creates attachment, in a literature whose own authors say abuse prevents the thing being described. The attachment people report is real. The hostage analogy is not where its explanation can come from.
How long does it take to break a trauma bond?
Nobody has published a figure, and the only systematic review of the literature states that no article indicated how trauma bonds could be severed and replaced with healthy attachments. Any specific number of days or weeks you read is invented. The one relevant measurement comes from Dutton and Painter's 1993 study: among 75 women assessed just after leaving an abusive relationship and again six months later, felt attachment to the former partner had decreased by about 27%, with no intervention being tested. That is a decline rather than a disappearance, in a group who had already left, and it is the entire evidence base on the question. The closest thing to an answer about what helps comes from interviews with 19 trafficking survivors, who described trusting relationships, their relationship with themselves, and education — notably with no steps, sequence or timeline attached.
Does intermittent reinforcement explain why I cannot leave?
It is the most-cited mechanism and the evidence for it in human relationships is thinner than its confidence suggests. The schedule effect itself is solid and old: behaviour reinforced unpredictably is harder to extinguish than behaviour reinforced consistently, which comes from Ferster and Skinner's work on operant conditioning in animals. The bridge to relationships is Dutton and Painter's finding that the extremity of intermittent maltreatment and the power differential predicted how strongly 75 women remained attached to a former partner, with those relationship variables accounting for 55% of the variance at six months. That is real, and it is a study with no comparison group of non-abusive relationships, which means it cannot establish that unpredictability produces more attachment than consistency does. The same mechanism, as recommended deliberately in dating advice, is taken apart in our piece on the 72-hour rule.
Is there a test for trauma bonding?
Four instruments exist, and none of them is the online quiz you are likely to find. Graham and colleagues' 1995 scale has 49 items and three factors, was built on college women in heterosexual dating relationships, and has a validated Spanish translation. Reid's Trauma Bonding Scale for Adults was developed in the context of sex trafficking, deposited in PsycTESTS in 2023, and validated in 619 young adults in the United States and 538 in Kenya. Two further measures contain trauma-bonding components without being designed for it. The limitation that matters for an individual reader is that these were built to compare groups in research, not to classify a person, and no validated cut-off score has been published for any of them. A quiz that tells you whether you are in a trauma bond is not reporting a threshold anyone established.
Can you have a trauma bond with a parent, a friend or a boss?
The term is used that way constantly online and the research base does not extend there. Counting the indexed literature, the modern empirical work is overwhelmingly about sex trafficking: a scoping review across ten databases found fifteen articles, and the primary studies interview ten service providers or nineteen survivors from one specialty court docket. The older core is Dutton and Painter's work on women leaving violent intimate relationships. There is a 1992 review on traumatic bonding in incest and a 2019 concept analysis in the commercial sexual exploitation of children, so the family and child literature is not empty — but nothing in it supports applying the construct to a demanding boss or an inconsistent friend. The honest position is that outside intimate partner violence and trafficking, this is a metaphor being used as an explanation.
Does having a trauma bond mean I am codependent?
No, and the history of that word is the reason to be careful with it. Lesiak and Gelsthorpe set out the three explanations offered over the past fifty years for why someone stays with a partner who hurts them — that she enjoys it (masochism), that she is too helpless to leave (learned helplessness), and that she enables it for her own gratification (codependency) — and trace their provenance: Freud, experiments on electrocuted dogs, and stories told in Alcoholics Anonymous. None was built on research into the experiences of the people it described, and all three locate the defect inside the victim while leaving the other person's conduct unexamined. The staged model now circulating has the same shape. What the data support is narrower: childhood maltreatment and attachment insecurity are risk factors, measured once, in 354 people, cross-sectionally.
Why do I miss someone who treated me badly?
Because attachment does not require good treatment, and nothing in the research suggests that missing someone is evidence about your character. Participants in the 2026 Cambridge interview study — all financially independent, most living separately, screened to exclude every external reason for staying — described the pull in the language of addiction and craving, and three had moved to other cities to keep themselves away. One described the separation as worse than the abuse. That is a qualitative account of eighteen self-selected women rather than a prevalence figure, and it is reported here for the shape rather than the numbers. The measured version is Dutton and Painter's: attachment persisted after leaving, was predicted by the intermittency and extremity of the maltreatment, and declined by roughly a quarter over six months. Missing someone is not a verdict on you, and it is not a reason.
What should I call this if I need help from a service or a court?
Coercive control, which is the term with the finished instrument and, in a number of jurisdictions, a legal definition. On the title-and-abstract count run for this article, coercive control returned 455 PubMed records against 27 for the whole trauma-bonding family. The Checklist of Controlling Behaviors, validated by Lehmann, Simmons and Pillai in 2012, has 84 items across ten subscales — physical abuse, sexual abuse, male privilege, isolation, minimising and denying, blaming, intimidation, threats, emotional abuse, economic abuse — and was administered to 2,135 women taking refuge in a domestic violence shelter. The practical difference is that trauma bond names how something feels, while coercive control names what someone did. The second is the one another person is obliged to act on. Find A Helpline lists free, confidential services by country.
Sources
What this article draws on. For the broader bibliography see our sources page.
- Dutton, D. G. & Painter, S. (1981) — Traumatic Bonding: The development of emotional attachments in battered women and other relationships of intermittent abuse. Victimology: An International Journal, 6(1-4), 139-155 — the founding paper. It has no DOI, is indexed neither in Crossref nor in PubMed, and sources disagree on whether to date it 1981 or 1983; we were not able to read it and report it only as the theory its 1993 test describes
- Dutton, D. G. & Painter, S. (1993) — Emotional attachments in abusive relationships: a test of traumatic bonding theory. Violence and Victims, 8(2), 105-120 — 75 women assessed just after leaving and again six months later; 55% of the variance in attachment at six months; attachment down about 27%. There is no comparison group of non-abusive relationships
- Graham, D. L. R., Rawlings, E. I., Ihms, K., Latimer, D., Foliano, J., Thompson, A., Suttman, K., Farrington, M. & Hacker, R. (1995) — A scale for identifying “Stockholm syndrome” reactions in young dating women: factor structure, reliability, and validity. Violence and Victims, 10(1), 3-22 — a 49-item scale with three factors; also deposited in PsycTESTS
- Reid, J. A. (2023) — Development and Evaluation of the Trauma Bonding Scale for Adults© in the Context of Sex Trafficking. American Journal of Criminal Justice, 48(4), 945-966 — paywalled; the scale itself is in PsycTESTS at doi 10.1037/t90296-000
- Chenneville, T., Reid, J. A., Wasilewski, S. & Adeli, S. (2024) — Validation of the Trauma Bonding Scale for Adults© in Kenya. Journal of Interpersonal Violence, 39(9-10), 2148-2164 — 538 young adults aged 18-29; paywalled, and what is reported here comes from the published abstract
- Chenneville, T., Bilali, K., Wasilewski, S. & Reid, J. (2025) — Similarities and Differences in Trauma Bonding Among Young Adults in the U.S. and Kenya: Implications for Forensic Assessment. Behavioral Sciences & the Law, 43(4), 374-384 — US N = 619 and Kenya N = 538; paywalled, reported from the published abstract
- Reid, J. A. (2024) — Measures of Trauma Bonding. In Trauma Bonding and Interpersonal Crimes, ch. 3, pp. 23-36. Wiley — the chapter stating that no measure of trauma bonding has been fully developed, evaluated and standardised, and naming the Stockholm Syndrome Scale, a Post-Traumatic Stress Index subscale and a Trauma Symptom Inventory-2 subscale. Behind an anti-bot wall; quoted from the publisher's abstract
- Shaughnessy, E. V., Simons, R. M., Simons, J. S. & Freeman, H. (2023) — Risk factors for traumatic bonding and associations with PTSD symptoms: A moderated mediation. Child Abuse & Neglect, 144, 106390 — 354 participants in current abusive relationships; childhood maltreatment and attachment insecurity predicted traumatic bonding over and above age, gender and romantic love. Paywalled; reported from the published abstract
- Casassa, K., Knight, L. & Mengo, C. (2022) — Trauma Bonding Perspectives From Service Providers and Survivors of Sex Trafficking: A Scoping Review. Trauma, Violence, & Abuse, 23(3), 969-984 — ten databases, fifteen articles; the review that found no article indicating how trauma bonds could be severed
- Casassa, K., Ploss, A. & Karandikar, S. (2024) — “He Loves Me Hard and Then He Abuses Me Hard”: How Service Providers Define and Explain Trauma Bonds Among Sex Trafficking Survivors. Violence Against Women, 30(5), 1354-1377 — interviews with 10 licensed social workers or counsellors
- Casassa, K., England, G. & Karandikar, S. (2024) — “It's Not Love, Baby Doll. It's Science”: Experiences of Trauma Bonding Among Survivors of Sex Trafficking. Violence and Victims, 39(5), 598-617 — 19 survivors from a human trafficking specialty docket
- Casassa, K., England, G. & Karandikar, S. (2025) — “I Had to Allow Myself to Heal”: How Survivors of Sex Trafficking Have Experienced Healing From Trauma Bonding. Violence Against Women, 31(9), 2350-2372 — the same 19 survivors, asked about healing; three themes and no protocol
- Lesiak, M. & Gelsthorpe, L. (2026) — The Invisible Abuser: Attachment, Victimization, and Perpetrator Perception in Repeat Abuse. Violence Against Women, 32(11), 4195-4214 — open access; 18 UK women screened to exclude every external reason for staying. Read in full
- Namnyak, M., Tufton, N., Szekely, R., Toal, M., Worboys, S. & Sampson, E. L. (2008) — “Stockholm syndrome”: psychiatric diagnosis or urban myth? Acta Psychiatrica Scandinavica, 117(1), 4-11 — four databases, twelve papers, mostly case reports, no validated diagnostic criteria. Paywalled; reported from the published abstract
- Fuselier, G. D. (1999) — Placing the Stockholm Syndrome in Perspective. FBI Law Enforcement Bulletin, 68(7), 22-25 — the HOBAS figures (92% and 95% of over 1,200 incidents) and the three conditions, including that the captors must not abuse or threaten. The linked file is the whole July 1999 issue; the article begins on its page 22
- Doychak, K. & Raghavan, C. (2018) — “No voice or vote”: trauma-coerced attachment in victims of sex trafficking. Journal of Human Trafficking, 6(3), 339-357
- Doychak, K. & Raghavan, C. (2023) — Trauma-coerced attachment: Developing DSM-5's dissociative disorder “identity disturbance due to prolonged and intense coercive persuasion”. European Journal of Trauma & Dissociation, 7(2), 100323
- Lehmann, P., Simmons, C. A. & Pillai, V. K. (2012) — The Validation of the Checklist of Controlling Behaviors (CCB). Violence Against Women, 18(8), 913-933 — 84 items, ten subscales, 2,135 women in a domestic violence shelter
- Hamberger, L. K., Larsen, S. E. & Lehrner, A. (2017) — Coercive control in intimate partner violence. Aggression and Violent Behavior, 37, 1-11 — a review of the conceptualisation and measurement of coercive control; listed for the field rather than for a specific figure, as we could not obtain its text
- Ferster, C. B. & Skinner, B. F. (1957) — Schedules of Reinforcement — the original mapping of reinforcement schedules, and the source of the claim that intermittently reinforced behaviour resists extinction. The subjects were pigeons, and the outcome measured was persistence rather than attachment
- Carnes, P. J. (1997) — The Betrayal Bond: Breaking Free of Exploitive Relationships. Health Communications — a trade book, sixteen years after Dutton and Painter, and the origin of the phrase “betrayal bond” rather than of the construct
- PubMed (US National Library of Medicine) — Title/abstract counts run on 29 October 2026: trauma bonding 19, trauma bond 4, traumatic bonding 7, Stockholm syndrome 36, gaslighting 153, coercive control 455, narcissistic abuse 1; love bombing 2 anywhere in the record; zero records pairing the construct with any word beginning “stage”
- National Domestic Violence Hotline (United States) — Warning signs of abuse — 24/7 confidential support: 1-800-799-7233, or text START to 88788
- Find A Helpline — Free, confidential support lines by country, including domestic abuse services
Keep reading
- MindsetWhy Smart People Keep Dating the Wrong People: What Is Measured, and What Is Only RepeatedHaving a type is real and has been measured properly once. Almost everything usually said about why has not been measured at all.
- MindsetCan Attachment Styles Change? How Much Moves, How Slowly, and What Moves ItYes, and the real version is slower than the genre promises: rank-order stability over ten adult years sits near .50, and not all movement is progress.
- MindsetWhy the Honeymoon Phase Ends: What Is Measured and What Is FolkloreSatisfaction does decline, slowly for most couples. The neurochemical story everyone repeats rests on two brain scans of 34 people and one blood test.
- DatingModern Dating Statistics: The Numbers That Hold Up, and the Ones Nobody PublishesWhich dating numbers have a real survey under them, what each one measured, and where every one of them stops.
- DatingThe 3-6-9 Rule in Dating: Where It Came From and What the Research SaysThree checkpoints, three months apart, and not one study behind them. What has actually been measured about the first year says something else.
- Body LanguageEye Contact: What It Predicts, What It Creates, and How Long Is Too LongThe average comfortable look lasts 3.3 seconds. The famous staring study is small, and the dim room it was run in moved its numbers as much as the gaze did.