The Strange Situation at 50: Are Attachment Styles Scientifically Real?

Most people meet attachment theory backwards. First comes the quiz result, the TikTok explainer, or the friend who announces that your partner is "so avoidant" — and only later, if ever, the question of where any of this came from and whether it deserves the confidence people bring to it.

It came from a twenty-minute laboratory procedure involving a toddler, two chairs, some toys, and a stranger. Mary Ainsworth's Strange Situation is now more than half a century old, and it has been tested about as hard as any method in developmental psychology. This article covers what the procedure actually was, what has held up after fifty years of scrutiny, what has quietly not held up — and what all of that means if you have ever been handed a label like "anxious" or "avoidant" and wondered how seriously to take it.

What Was the Strange Situation, Exactly?

Mary Ainsworth was a developmental psychologist — Ohio-born, trained in Toronto — who had spent years observing mothers and infants in their homes, first in Uganda and then in Baltimore. The Strange Situation, which she formalized with her colleagues in the 1978 book Patterns of Attachment (Ainsworth, Blehar, Waters & Wall), compressed what she had seen in living rooms into a standardized lab procedure.

It runs about twenty minutes, in eight short episodes. An infant of roughly twelve to eighteen months plays in an unfamiliar room with their caregiver. A stranger enters. The caregiver leaves briefly, returns, leaves again, returns again. The point was never how much the baby cries — some secure babies cry a lot, some very little. The classification hinges on the reunions: what the infant does with the caregiver when they come back.

• →Secure: seeks the caregiver, settles, and returns to playing. The relationship works as a base.
• →Avoidant: turns away, keeps to the toys, and treats the return as a non-event — a strategy of not showing the need, rather than an absence of it.
• →Resistant (ambivalent): clings and protests at the same time, and cannot quite be comforted.
• →Disorganized: added later by Mary Main and Judith Solomon (1990) — contradictory, frozen, or disoriented behaviour, with no coherent strategy at all.

One detail matters enormously and is almost always lost in translation: the Strange Situation classifies a relationship, not a child. The same infant can be classified secure with one parent and insecure with the other. It was never designed to stamp a personality type onto a baby — let alone onto the adult that baby becomes.

Are Attachment Styles Scientifically Real?

For the infant procedure itself, the answer is a fairly confident yes. The classifications are not one lab's artifact: trained coders reach strong agreement, and the patterns replicate far beyond the Baltimore families Ainsworth studied. The landmark check was a cross-cultural meta-analysis by Marinus van IJzendoorn and Pieter Kroonenberg (1988, Child Development), pooling nearly two thousand Strange Situation classifications across eight countries. Secure attachment was the largest group everywhere — roughly two in three infants — and, strikingly, the variation within countries was greater than the variation between them. Whatever the procedure measures, it is not an American peculiarity.

The honest caveat: the procedure does not carry identical meaning everywhere. In Japanese samples studied by Kazuo Miyake and colleagues in the 1980s, infants who had rarely been apart from their mothers found the separations far more distressing, inflating resistant classifications — a reminder that the "strangeness" of the Strange Situation depends on what a baby's ordinary life looks like. Cross-cultural researchers treat the categories as real but read them in context.

Is the Attachment Quiz You Took Online the Same Thing?

No — and this is the single most common confusion in how attachment theory gets used today.

Adult attachment is measured in two separate research traditions that grew up largely independently. The first is the Adult Attachment Interview, developed by Mary Main and colleagues: a roughly hour-long structured interview about childhood, coded not on what happened to you but on how coherently you can tell the story. The second is the self-report tradition from social psychology — questionnaires such as the Experiences in Close Relationships scale (Brennan, Clark & Shaver, 1998), where adults rate statements about closeness and worry in romantic relationships. When Glenn Roisman and colleagues (2007, Journal of Personality and Social Psychology) directly tested the overlap between the two, the associations were trivially small. Two instruments, both called "attachment," measuring largely different things.

The free quizzes circulating online are informal cousins of the second tradition — a handful of self-report items, no validation, instant label. They can be a useful conversation starter about how you experience closeness. What they are not is a diagnosis, a childhood X-ray, or anything resembling the Strange Situation. And the most common use of all — reading a listicle and classifying your partner ("is she avoidant?" is one of the most repeated questions in online attachment communities) — is the least reliable use of the framework that exists. A pattern in you is hard enough to assess from the inside; a pattern in someone else, assessed through the fog of your own hurt feelings, is not an assessment at all.

Do Attachment Styles From Childhood Determine My Relationships?

Here is where fifty years of follow-up research pushes back hardest on the pop-psychology version. The infant classifications are real. Their grip on your adult life is much looser than the internet suggests.

On stability, the landmark cohorts disagree with each other. In one twenty-year follow-up, Everett Waters and colleagues (2000, Child Development) found that just under three-quarters of participants kept the same secure-versus-insecure status from infancy to early adulthood — but in the same journal issue, Michael Lewis and colleagues reported no significant continuity at all in their cohort, with life events like parental divorce doing much of the reshuffling. Attachment patterns can persist; they are also demonstrably open to revision by what life does next.

On prediction, the strongest evidence is meta-analytic. Ashley Groh and colleagues (2017, Child Development Perspectives) pooled the longitudinal studies: early security predicted later peer competence at r = .19, and fewer externalizing (r = −.15) and internalizing (r = −.07) problems. Those are real signals — and they are small. More recently, registered analyses of two landmark longitudinal cohorts published in 2023 found partial correlations of about .10 to .15 once background factors were controlled. In plain terms: knowing how an infant behaved in a twenty-minute procedure tells you something, on average, across hundreds of children — and very little, with confidence, about any one adult's marriage.

This is the part worth sitting with, because the accurate science and the useful message turn out to be the same sentence: attachment patterns are real, meaningful, and modest — not destiny. Anyone telling you that your childhood classification sealed your relational fate is claiming precision the data has never supported. Anyone telling you attachment is meaningless is ignoring one of the most replicated frameworks in developmental psychology.

Can Therapy Change Your Attachment Style?

The research above cuts both ways, and the second direction is the hopeful one. If attachment patterns were fixed at eighteen months, the Lewis cohort could not have scattered the way it did. Patterns learned in relationships appear open to revision in relationships. The Adult Attachment Interview literature even has a name for people who describe genuinely hard childhoods with the coherence and balance of security: earned secure. They exist in every sample.

That is the premise attachment-based therapy works from — not erasing history, but examining the strategies you built inside it and practising different ones inside a steadier relationship. What that looks like differs by pattern: the work for anxious attachment tends to centre on the alarm that fires around distance and reassurance; avoidant attachment work tends to centre on what closeness costs and why self-sufficiency became the safer bet; and disorganized attachment work moves more slowly, because the pattern usually formed where comfort and threat came from the same person. No responsible therapist promises an outcome or a timeline — but the direction of the evidence is that these patterns are practised, not carved.

The Verdict at Fifty

• →Held up: the infant classifications themselves, their replication across eight countries with a secure majority everywhere, and a genuine — if small — predictive signal for later development.
• →Revised: lifelong stability is modest and contested between landmark cohorts; prediction of adult outcomes shrinks to partial correlations around .10–.15 under controls; and the procedure classifies relationships, not people.
• →Never was the science: the online quiz as diagnosis, the four styles as personality types, and the classification of partners from listicles.

The Strange Situation earned its fiftieth birthday. What it never earned is the fatalism sold in its name. If you want the story of where the whole framework began — with infant monkeys, a wire mother, and a cloth one — that history, ethical cost included, is covered in Harlow's Monkeys: Why Comfort Beat Food.

When to Reach Out

A quiz label is not a reason to seek therapy. Noticing the same painful pattern across relationships — the alarm at distance, the wall at closeness, the push-pull you can describe but not stop — is. You do not need to arrive with the right vocabulary, or any vocabulary.

If you are in crisis right now, call or text 9-8-8, the 9-8-8 Suicide Crisis Helpline, available across Canada at no cost, any time. In British Columbia you can also call 1-800-SUICIDE (1-800-784-2433).

A Low-Pressure First Step

TEO Counselling offers virtual sessions across British Columbia, Ontario, Alberta, Saskatchewan, Manitoba, and Newfoundland and Labrador, and a free 15 minute consultation to ask about approach, training, and fit — a conversation, not a commitment. Book a free consultation through the Jane booking platform, or call (519) 760-5211.

You don't have to carry it alone.