Attachment styles: what the research does and doesn't show

What did Mary Ainsworth find?

Mary Ainsworth, a colleague of Bowlby, is best known for the Strange Situation (as a 1997 interview with her puts it), a way of watching how infants behave with a caregiver. In the procedure, a child is observed in an unfamiliar room, with two brief separations from the caregiver and a stranger present. What the researchers look at most closely is what the child does when the caregiver comes back.

• Secure: the child gets comfort when distressed, then goes back to play.
• Avoidant: the child avoids the caregiver on reunion and focuses on exploring. These children can seem unaffected by the separation, yet, according to the 47-author review, behavioural and physiological measures indicate distress.
• Resistant (sometimes called ambivalent): the child's attention stays on the caregiver, even when exploring might be expected.
• Disorganized: the child may show mostly the same behaviours as the others, but with moments that don't fit any of the three, such as seeming fearful or freezing.

These are patterns of a child with one caregiver, observed in a lab. The review authors stress that the classifications are useful for studying group trends, and are 'not (currently) validated or intended to be used clinically at the individual level'. A 2023 meta-analysis of 97 samples of preschool and early-childhood children (8,186 children) found a spread across the four categories, and noted that 89% of the samples were from North America or Europe. So even the infant and child research describes a particular slice of the world.

What did Hazan and Shaver add in 1987?

In 1987, Cindy Hazan and Phillip Shaver published a paper asking whether romantic love might be an attachment process, a bond between adult lovers like the bond between infants and parents. They took the three infant patterns, secure, avoidant and anxious/ambivalent, and wrote them as descriptions of how an adult might experience love. Their abstract reports two questionnaire studies in which the three kinds of adult described their romantic love in different ways, and in which the styles were related to people's mental models of themselves and others and to the way they described their relationships with their parents.

Two things about that paper are worth keeping in mind. First, it's based on questionnaires: people described themselves, and no one watched them the way the Strange Situation watches infants. Second, the abstract itself ends by pointing to 'measurement problems' for future tests. Later research looked at how best to measure attachment, as the next sections show.

Are attachment styles fixed types, or a scale?

Hazan and Shaver's approach treated styles as categories. Later work found something different. In an overview of adult attachment research, the researcher Chris Fraley describes how Brennan, Clark and Shaver (1998) found two underlying dimensions: attachment-related anxiety and attachment-related avoidance. A secure adult, on that view, scores low on both. Fraley writes that most researchers now measure attachment dimensionally (as of 2018), and that attachment styles 'are things that vary in degree rather than kind'. That overview is a researcher's own summary, not a peer-reviewed paper.

A 2015 study by Fraley and colleagues tested this directly with a statistical method for telling categories from dimensions. In two samples, they found that individual differences looked more like a dimensional model than a categorical one, for general attachment and for attachment in specific relationships such as with parents and peers.

In plain terms, 'anxious' or 'avoidant' describes where someone sits on a scale. Researchers also study attachment in specific relationships, such as with parents or peers, as well as in general. It isn't a box you belong to. That's one reason the quiz format is a poor fit for what the research shows.

Can your attachment style change?

Research suggests attachment is not fixed, with caveats. A 2013 meta-analysis of 127 papers (21,072 attachments) looked at attachment to mothers and fathers from infancy to early adulthood, not at adult romantic styles. It found a moderate link (r = .39, where 0 is no link and 1 a perfect one) between attachment security at one time and another, no significant stability across intervals longer than 15 years, and higher stability over short intervals.

The 47-author review, which is about children, adds that childhood attachments are 'by no means fixed', and that classifications can change through new experiences, relationships or therapeutic interventions.

So the research doesn't back the idea that your early years have sealed your relationships, and it doesn't back the idea that change is quick or easy. It also doesn't tell us whether reading a guide or filling in a worksheet changes anything, and we know of no study that does.

Where do adult attachment styles come from?

A 2019 review by Chris Fraley and Glenn Roisman, of longitudinal research (studies that follow the same people over time) on where adult attachment styles come from, adds three points. Adult attachment appears to have origins in early caregiving, but those links are 'weak and inconsistent across measurement domains'. In plain words, the connection between early care and adult styles looked small, and it didn't show up the same way on every kind of measure. Styles appear more malleable in childhood and adolescence than in adulthood. And early experiences do not determine adult outcomes.

Does an 'insecure' attachment style mean something is wrong?

For children, not necessarily. The 47-author review, which is about young children, says insecure classifications can represent adaptive strategies to cope with a specific context, and do not necessarily predict negative outcomes. It says research supports only small-to-moderate associations between a child's attachment classification and later developmental outcomes. It also notes that insecure or disorganized attachment is not a formal diagnosis. That review does not tell us how this applies to adults.

The link to how people feel is not simple either. A 2021 paper pulling together four large meta-analyses calls the evidence on insecure attachment and internalising symptoms (such as low mood and anxiety) 'divergent'. In children, elevated symptoms were found with deactivating (avoidant) patterns but not hyperactivating (resistant) ones. In adolescents and adults it was the other way round: elevated symptoms with hyperactivating (preoccupied) classifications but not deactivating (dismissing) ones. The authors suggest the age group studied explains at least part of the difference, and put forward a theory they say still needs testing. These findings are about research groups, not about you, and they don't show that an attachment style causes anything.

Why is there no attachment style quiz on this page?

You may have come here for a quiz. Here is why one wouldn't tell you much.

• A quiz would sort you into a category, but researchers' own tests suggest attachment is better described as a scale.
• A quiz measures what you say about yourself, on one day. Researchers measure attachment in general and in specific relationships, and the research above suggests it can change.
• The 47-author review warns that overemphasising classifications 'leads to trait and typological thinking' and can lead to 'stigmatizing labels, leading to self-fulfilling prophecies'. The authors are writing about children's classifications in clinical practice. Applying the caution to a web quiz is our judgement, but we think it travels.
• We can't check how any particular online quiz was built.

If you've taken a quiz and the result rings true, that's fine. Treat it as a prompt to notice something, not a diagnosis or a fixed fact about who you are.

How to use the idea without a label

You can use the idea of attachment as a way of noticing, with no score at all. These are questions to think about for a few minutes, for one relationship at a time. They aren't a test, and there are no right answers. If any answer makes you feel afraid of this person, stop. That is not about your attachment style: see 'When to get more support' below.

1. 1When I feel close to this person, what do I want from them? What do I do?
2. 2When I feel distant from them, what do I do? What do I hope they will do?
3. 3What helps me feel safe with them? What makes me feel unsafe?
4. 4Is it different with other people in my life? What does that tell me?
5. 5What is one thing I'd like to be able to say to them, and what would help me say it?

If you'd like to put something into words to a partner or friend, the I-statements worksheet can help. The guide to setting boundaries and the assertive communication guide cover how to ask for what you need. If putting others first has become a habit, the people-pleasing worksheet is there. And the guide to practising self-compassion is for the harsh voice that can follow a rocky patch in a relationship. These are not for use with someone who frightens or controls you.

When to get more support

This guide is an explainer, not therapy, and it doesn't diagnose anything or treat anxiety, depression or any other condition. It's worth talking to your GP or another qualified professional if a relationship is affecting your sleep, work or health, if feeling low or anxious is getting in the way of your life, or you're worried about how long it has lasted. In England you can refer yourself to NHS Talking Therapies for anxiety and depression support without seeing a GP first (you need to be registered with a GP and aged 18 or over, or 16 or over in some areas). Relate lists relationship counselling and individual counselling on its website. Our helplines and therapist directory lists routes in other countries too.

If a partner or family member controls what you do, who you see or your money, or you feel afraid of them or find yourself 'walking on eggshells', that is not about your attachment style, and it is not your fault. In the UK, the National Domestic Abuse Helpline is free and confidential on 0808 2000 247, 24 hours a day (it supports women), and the Men's Advice Line is on 0808 801 0327, Monday to Friday, 10am to 5pm. In the US, call the National Domestic Violence Hotline on 1-800-799-7233, or text START to 88788. If someone might be monitoring your devices, use a safe device. In an emergency, call 999 (UK) or 911 (US). The helplines page lists more.

If you ever have thoughts of harming yourself, please reach out straight away: in the UK and Ireland, call Samaritans on 116 123; in the US, call or text 988; elsewhere, findahelpline.com lists free, confidential lines. You deserve real support, not just a page to read.