Can Attachment Styles Change? What Trait Stability Says About It

Attachment styles change. They change slowly, in a predictable direction, and at roughly the pace a personality trait moves — which is to say years, not weeks, and by degree rather than by conversion.

That answer is less satisfying than either version you usually hear. It is not "your style is set by age two," and it is not "do the work and become secure." Both of those are stories about categories. The measurements underneath are not categorical at all, and once you see what is actually being measured, the rate of change stops being mysterious.

The two numbers that answer the question

Rank order holds, loosely. Chris Fraley pooled every longitudinal study he could find that measured attachment security in infancy and again later in life, some as far out as early adulthood. The stability correlation came out at about .39 ¹. That is a tilt, not a track. The anxious one-year-old is more likely than average to be the anxious twenty-year-old, and plenty of people move a long way.

Mean levels drift with age. In a sample of over 86,000 adults aged 18 to 70, attachment anxiety was highest among the youngest respondents and declined steadily with age ². Not in a step at any particular life event — gradually and continuously across five decades.

Put those together and you get the shape of the change. Most people become somewhat more secure as they age, while mostly keeping their position relative to everyone else. That is exactly the pattern personality research finds for traits: Neuroticism falls and Agreeableness rises across adulthood in nearly every sample ever collected, without shuffling who is more anxious than whom ³.

Why the label moves more than you do

There is a second reason people report that their style changed, and it has nothing to do with change.

The four styles are not four kinds of person. They are four quadrants drawn on two continuous dials. The first is attachment anxiety: how much you worry about being abandoned, how badly a partner's distance lands. The second is attachment avoidance: how uncomfortable you are with closeness and dependence. Everyone has a score on both, and the styles are just the corners you get by cutting each dial in half — secure is low-low, anxious is high-anxiety, dismissive-avoidant is high-avoidance, fearful-avoidant is high on both.

By the late 1990s the field had pooled the many self-report scales in use and found they all reduced to those two dimensions. The categories survive because they are memorable, not because they are what anyone measures.

So consider someone who scores just above the midpoint on anxiety. They get the anxious label. Someone just below gets secure. The two are nearly identical, and a few points of ordinary movement flips one into the other. If you have taken two attachment quizzes and received two different styles, this is almost certainly why: you live near a cut line, and the quizzes drew it in slightly different places.

This is also the test of whether a quiz is worth taking. The research instruments report two scores and show you how far from the midpoint each sits. A quiz that hands you a single label and no numbers measured two dimensions and then threw the measurement away to give you a type. The label is more shareable. The numbers are the only part that was ever true of you. The Big Five ran into this problem thirty years earlier and settled it the same way, which is why a percentile is the honest way to report a trait and a type is not.

The trait underneath the anxiety dial

If attachment changes at trait speed, the obvious question is how much of it is a trait. Erik Noftle and Phillip Shaver put this directly to the test, giving several hundred adults both a Big Five inventory and the standard attachment measure ⁴.

Attachment anxiety correlated with Neuroticism at around .4. That is substantial. Neuroticism is a general readiness to experience threat, and a partner going quiet for six hours is a threat, so the trait shows up here the same way it shows up in exam stress or health worry. Knowing someone's Neuroticism percentile tells you a fair amount about how they behave when a relationship feels at risk.

But .4 is not 1.0. Squared, it says the two measures share about 16 percent of their variance. The other 84 percent is the part that is not general threat-sensitivity but something specific to closeness — a history, a set of expectations about what people do when you need them. Someone can be low in Neuroticism and still carry a learned alarm about abandonment. That person exists, and a Big Five profile alone will not find them.

The avoidance dial has a looser fit. Higher avoidance goes with lower Agreeableness and lower Extraversion, but the correlations are modest, generally .2 to .3 ⁴. That makes sense: avoidance is not a mood, it is a strategy — a preference for self-reliance over dependence — and a strategy can sit on top of several dispositions at once.

The finding that most often gets dropped from summaries is the one that matters most here. After controlling for the Big Five, the attachment dimensions still predicted relationship quality. The overlap is real and so is the residue. Neither map replaces the other.

What this means for changing it

The trait overlap explains the rate. Roughly a sixth of the anxiety dial is general threat-sensitivity, and that part moves the way Neuroticism moves — slowly, over years, with age doing much of the work. The remaining part is relationship-specific and learned, which is the part that responds to what actually happens in your relationships.

That is why the realistic target is a few points on a dial rather than a new label, and why the change tends to show up as "this bothers me less than it used to" rather than as a conversion experience. It is also why a relationship in which reassurance is cheap and reliable does more than any amount of reading about attachment: the specific expectations are the movable part.

The place this gets genuinely useful is where the two maps disagree. Low Neuroticism with high attachment anxiety points at a history rather than a temperament, and histories respond to different work than temperaments do. High Neuroticism with low attachment anxiety means you have built something in your relationships that your general disposition did not hand you, which is worth knowing so you can protect it.

Worth saying plainly, because the vocabulary invites it: all of this describes normal-range variation, meaning where you sit relative to other people on two ordinary dials. If closeness is a source of real distress rather than a pattern you have noticed and want to understand, that is a different conversation, and a questionnaire is the wrong instrument for it.

If you have the attachment side and want the wider map: screen one trait free (3 min) → or take the full Big Five test and see all thirty facets →. It will not tell you your attachment style. It will tell you how much of that style was already visible in your temperament, which is the question a quiz cannot answer and the one that decides how much of it is likely to move.

References