What the data actually shows
Attachment theory began with John Bowlby's mid-20th-century work and Mary Ainsworth's "Strange Situation" studies in the 1970s, which classified infants into secure and insecure patterns based on how they responded to separation from and reunion with a caregiver. The consistent finding was that sensitive, responsive caregiving was associated with more secure attachment, and this is one of the more replicated observations in developmental psychology.
But when researchers follow the same people from infancy into adulthood, the correlation between early attachment and adult attachment is generally modest. Longitudinal studies find real continuity, but plenty of discontinuity too — many people classified as insecure infants are securely attached adults, and vice versa. Researchers also distinguish childhood attachment to a specific caregiver from the broader "attachment style" measured in adult romantic relationships; these are related but not the same thing.
The literature also documents change in both directions. "Earned security" describes adults who report difficult early caregiving but show secure attachment now, often linked to a later supportive relationship or reflective work on their past. Reviews of attachment-informed therapy and of relationship effects suggest patterns can shift, though the effects are gradual rather than dramatic and the research is still developing.
Why this feels different from how it actually is
Attachment language has spread far beyond the research, and the popular version is more deterministic than the science. Online, "attachment style" is often presented as a fixed personality type that explains your whole relational life and traces neatly back to your parents. That framing feels satisfying because it offers a single, tidy origin story — but the data describe tendencies that interact with current circumstances, not a permanent label.
It also feels fixed because attachment patterns are partly self-reinforcing. If you expect closeness to be unreliable, you may behave in ways that make it so, which seems to confirm the original belief. That loop can make a learned pattern feel like an unchangeable trait, even when the underlying tendency is more malleable than it appears in the moment.
And memory plays a role. We tend to reach for early-childhood explanations for adult difficulties partly because the story is compelling and partly because it locates the cause somewhere we cannot be blamed for it. That is understandable, but it can overstate how much of the present is dictated by the distant past.
Childhood is one input among several. It tilts the odds rather than writing the script.
What the research says to do about it
The most encouraging consistent finding is that attachment patterns can change, and current relationships are one of the main routes. A stable, responsive partner or close friend is associated with movement toward security over time. So is the slower work of noticing your own patterns rather than treating them as facts about the world.
For people who want structured help, attachment-informed and emotionally focused therapies have supportive evidence for improving relationship functioning, and the broader therapy literature shows that relational patterns are workable in adulthood. The honest framing is that change tends to be gradual and partial rather than a clean reset, but it is well documented.
Reflecting coherently on your own history — being able to tell a clear, balanced story about it, including the hard parts — is itself associated with security in the research on "earned security." It is less about having had a good childhood and more about having made sense of the one you had.
What the research says does not help
Treating your attachment style as a fixed diagnosis does not match the evidence and can become self-fulfilling. The data describe a tendency that responds to current relationships and effort, so labelling yourself permanently "anxious" or "avoidant" tends to entrench the pattern rather than explain it.
Using a childhood origin story to predict the failure of present relationships is not supported. Discontinuity is common in the longitudinal data, so the link from a difficult early start to a difficult adult relationship is real but far from guaranteed.
The quick online quizzes that sort you into a permanent type oversell their precision. They can be a useful prompt for reflection, but the underlying research measures tendencies that shift, not a stable personality category — and treating the result as fate misreads what the science actually claims.
'Earned security' describes adults who had a difficult early start but show secure attachment now — childhood matters, but it is not destiny.
What this looks like in real life
A difficult start, a secure now
Someone describes a genuinely hard early childhood but reports stable, trusting relationships as an adult — often after a supportive long-term partner or friendship, or after doing the reflective work of making coherent sense of their past. Researchers call this 'earned security,' and it is well documented. It is less about having had a good childhood than about having made sense of the one you had.
The label that becomes self-fulfilling
A quiz sorts someone into 'anxious,' and they start treating it as a fixed diagnosis that explains their whole relational life. But the research measures a tendency that responds to current relationships and effort. Expecting closeness to be unreliable can nudge you to behave in ways that make it so — which seems to confirm the belief and can entrench a pattern that is more malleable than it feels.
Real numbers in context
Attachment research deals in patterns and probabilities, not clean percentages, and the field is careful about this. The headline finding — that sensitive caregiving relates to secure attachment — is well replicated, but the strength of the link from infant attachment to adult attachment is generally described as modest, with substantial change over time documented in long-running studies.
Population estimates of adult attachment categories vary by sample and measure; many studies place a majority of adults in the secure range, with the rest distributed across insecure patterns, but these figures are sensitive to how attachment is measured. The most defensible takeaway is qualitative: childhood matters, it is not destiny, and "earned security" is a real and documented path.