How to recognise your attachment style: questions that tell you more than an online test

“I did three tests and every one of them gave me a different result.”

She arrived with the screenshots on her phone. One said anxious, another said secure with avoidant leanings, the third said “fearful-avoidant” and added a percentage to two decimal places. Then I asked her something none of them had asked: when she last asked someone for help, and what happened when she did. She talked for five minutes. By the time she stopped, we both knew far more than any printout did.

This piece is about how you actually go about recognising your attachment pattern — and why questions about real events tell you more than statements rated from 1 to 7. I have written about how patterns repeat from one relationship to the next in the piece on attachment patterns; here I am after something more down to earth: how you work out where you stand at all.

Why an online test almost always tells you too little

I have nothing against questionnaires; I use their logic myself. The trouble is what happens once they move online: a measuring instrument yielding two continuous scores turns into a personality-assigning machine that returns one word and a confidence it has not earned.

The first problem is that most online versions are not really a measure at all — they are a copy of a copy. Established scales such as the ECR were developed on large samples and examined with item response theory. Pick six statements out of them, translate by feel, add a colourful result, and nothing of the psychometrics survives except the appearance.

The second problem runs deeper and applies to good questionnaires too: they measure what you think about yourself and are willing to admit. The very people who push closeness away often rate themselves as remarkably independent and entirely content — because that is genuinely what they feel. They are not lying. Their strategy consists precisely in not registering distress as distress.

The third problem is the most practical. A test asks what you are like in general, but attachment is not a trait you carry around like eye colour. It switches on in a particular relationship, at a particular moment, when you are tired, frightened or alone. “Do I like relying on others” and “what did you do on Tuesday at eleven at night” are very different questions.

The four types are really two scales

If one thing here is worth keeping, it is this. Modern research on adult attachment measures not four boxes but two continuous dimensions: how much anxiety you carry in relationships (fear of being left, need for reassurance) and how much avoidance (discomfort with closeness, need for self-sufficiency). The four familiar types are simply names for the corners of that space.

The distinction is not hair-splitting. When researchers used taxometric methods to test whether people sort into classes or spread continuously, they found a continuous distribution. There is no group of “the anxious” and a group of “the secure” with a clean line between them, only a smeared-out crowd mostly somewhere in between.

So the most useful question to put to yourself is not “which type am I” but “how much anxiety and how much avoidance do I have in this relationship, in this stretch of my life”. Those are two numbers, not one label — and both of them can be moved.

The questions I ask instead of a test

I do not hand out questionnaires in my room. I ask about events. An event question is harder to slip past, because it has details: the hour, the place, what was said, what you did with the phone in your hand.

Some you can put to yourself: when did you last ask for help you did not strictly need but did want? What happens in your chest when your partner says “we need to talk”? Who in your life would come at two in the morning — and have you ever called them? When things go badly, does your body notice first, or do your people?

These questions are not a test and they carry no score. They are an invitation to look at a recording instead of a summary. Summaries are polite. Recordings are precise.

The first twenty minutes after an argument

If I were allowed one window in which an attachment pattern shows itself most cleanly, I would choose this one. Not the argument — what comes immediately after it.

In those twenty minutes one person cannot bear to leave it unresolved. They text, they call, they follow their partner down the hallway, they repeat the same sentence in different words. Another puts their shoes on and goes out. Opens a laptop. Vacuums. Says “nothing is wrong” and means it, because inside it genuinely does feel calm.

Both are doing the same thing — bringing their arousal down — in opposite directions. One lowers it with closeness, the other with distance. That is why this moment tells you more than twenty statements on a scale: it shows where your system turns under pressure, not where you would like it to turn when calm.

Who you called when it was bad

That question has a strange power. People answer quickly and in general terms at first: “I have a few people”, “my sister”, “friends”. Then I ask about a specific event — the last bad news, the last fright, the last genuinely hard evening — and the answer often changes.

Sometimes it turns out they called nobody — not because there is nobody, but because the thought never crossed their mind. That matters. It means the system that ought to look for a person in distress switched that option off somewhere along the way, most likely because it once proved unreliable or expensive.

Other times they called four people in an hour and after each call nothing was better. That is information too: comfort is not getting in — reception is on, but the channel is shut.

The same person, three relationships

This is the most common surprise. Asked the same question three times instead of once — about a partner, a mother, a closest friend — the same person gives three different profiles. Someone excessively guarded with a partner is entirely open with a friend. Someone calm in their couple turns into a fourteen-year-old the moment their mother rings.

This is well documented: questionnaires built to measure attachment relationship by relationship show that much of the variation belongs to the particular relationship rather than to the person as a whole. There is a general leaning you carry, and there is what happens between you and this one human being.

Which is why “I am the anxious type” is usually too imprecise to be useful. Better: “in this relationship, when he does not touch me for three days, my alarm goes off”. That sentence can be addressed. A label cannot.

What the way you tell it reveals

There is a second route to measurement: a structured interview about childhood, rated by trained coders. Its peculiarity is that it counts not what happened to you but how you speak about it — whether the account holds together, whether you give examples when you make a claim, whether your sentences fall apart halfway through.

Interestingly, that interview and the questionnaires overlap far less than you would expect. A questionnaire measures how you experience your close relationships now. The interview measures how much access you have to your own history without being swept away by it.

You cannot do this properly at home, but you can notice what happens to your speech when you talk about your childhood. Is the account suspiciously smooth and free of detail? Or does it carry you off so you lose the thread? Either is a sign that something has not been put away.

Where the scales break down

Questionnaires measure worst of all for people with serious early trauma. Their answers often swing between extremes within a single week, because the system genuinely does work in contradiction — calling out and fleeing at once. A test smooths that into an average describing nobody.

It is similarly complicated in acute distress. Anyone mid-separation, mid-burnout or mid-insomnia will score high on anxiety, and that says nothing about their enduring pattern. A measurement taken in a storm describes the storm.

And some items measure culture, not attachment. “I am not comfortable having to rely on anyone” means one thing where self-sufficiency is a virtue and you grew up believing you do not burden others, and something else where that is not so.

Where my own questions get it wrong

Now against myself. Everything above has limits, and honesty requires that I list them.

First: event questions depend on memory, and memory is reconstruction. Someone who feels abandoned today will more easily call up scenes of abandonment. That is not a lie but recall bias, and no question of mine removes it.

Second, and more importantly: the four types are not boxes, and I am not convinced they are even the most useful language. Most people are a mixture. The pattern shifts between relationships and over time — long-term studies show adult attachment moves, driven mainly by life events and long relationships rather than insight alone. Someone high on anxiety at twenty is not necessarily there at thirty-five.

Third: the label can become an excuse. “I am just avoidant” sounds like insight but works as a defence — it explains the behaviour and thereby removes it from the conversation. Much as with diagnoses, which I wrote about in the piece on what a diagnosis does and does not tell you: a label is a map, not a justification and not an identity.

And fourth: for some people attachment is not the relevant subject. Someone who has not slept for three months, someone burnt out, someone with a genuinely unreliable partner does not need an analysis of their pattern but sleep, rest or a decision. Attachment is a useful explanation, not an explanation for everything.

What to do with the answer

Say you have worked something out about yourself. What then?

The least useful move is to present it to your partner as a fact they must accommodate. The most useful is to turn it into one concrete request. Not “I am anxious, so I need more attention”, but “when you go out in the evening, text me when you arrive — then I stop churning”. Not “I need space”, but “I need twenty minutes, then I come back and we carry on; I am not leaving because I am gone”.

If that is where it sticks — if you know what is happening and still do the same thing — that is a good moment for therapy. Not because something is badly wrong, but because attachment patterns change most reliably inside a relationship rather than in thinking about one. If you are considering it, I have written about what to look for in the piece on choosing a therapist.

And if what troubles you most in all of this is the night-time churning over what somebody meant by that message, it is probably more useful to start there — with rumination and with setting boundaries, rather than with typology.

What to remember

An online test gives you a word. It is not wrong, but too coarse to do anything with.

Ask two questions instead of one: how much anxiety and how much retreat you carry, in this relationship, in this stretch of your life. Look for answers in events, not self-descriptions: the first twenty minutes after an argument, and who you called when it was bad — or whether it even occurred to you.

And take seriously that you are not the same person in every relationship. That is not inconsistency — it is the most hopeful thing here. If the pattern changes depending on who you are with, it is not your fate but something made between two people.