Psychology has assumed for decades that we bond with people like ourselves. A study of 1,484 people just found that personality similarity has almost no effect on how good a friendship feels

The phrase “birds of a feather flock together” has been part of the way people explain friendship for so long that it has moved from observation to assumption. Of course we gravitate toward people like us. Of course shared personality means shared comfort. Of course the closer your friend’s disposition is to your own, the better the friendship will feel. This assumption underpins everything from dating app algorithms to workplace team-building advice to the social psychology literature going back decades.

A new study has tested it at a level of precision that most previous research could not achieve and found that the assumption is mostly wrong. Friends are similar, but barely. They perceive each other as far more similar than they actually are. And the degree to which two people match on personality has essentially no relationship with how satisfied either of them is with the friendship. What determines friendship satisfaction turns out to be something simpler and considerably less flattering to our ideas about why we choose the people we keep close.

How the study was designed

Friendship research has a persistent methodological problem. Most studies examine pairs of friends in isolation, which misses the networked reality of how social bonds actually exist. Your friendship with one person does not occur independently of your friendship with others. You are embedded in a group, and the way you perceive each friend is shaped by the entire social context you share.

The Michigan State University team addressed this by recruiting 371 groups of exactly four friends, for a total of 1,484 participants. Each group joined a verified video call to confirm their genuine social connection. Every person then rated their own personality and the personality of all three other group members using the standard Big Five framework, which measures openness to experience, conscientiousness, extraversion, agreeableness, and neuroticism.

This round-robin design, in which every person evaluates every other person in the group, produces a dataset that most friendship studies never attempt. It allows researchers to separate two things that previous studies frequently confused: actual similarity, meaning how close two people’s self-reported personality scores genuinely are, and perceived similarity, meaning how similar a person believes their friend to be relative to themselves. The distinction matters enormously, because these two things turn out to behave very differently.

Friends are similar, but only barely

The first finding sounds like a confirmation of the folk theory. Friends are indeed similar in personality across four of the five Big Five traits. People with similar levels of openness, agreeableness, conscientiousness, and neuroticism do tend to end up as friends.

But the effect sizes are so small that they require closer examination. The correlations ranged from approximately .05 to .10, which in statistical terms is a real but extremely weak association. Translating that into practical terms: knowing how open, agreeable, conscientious, or emotionally stable one person in a friendship is gives you only slightly better than random odds of predicting where their friend falls on the same trait. The similarity is detectable but it is not strong, and it is certainly not the primary force shaping who ends up friends with whom.

Extraversion is the single exception: friends showed no statistically significant actual similarity on this trait at all. A person who is highly extraverted is no more likely to have highly extraverted friends than introverted ones. Whether this reflects the complementary nature of social roles, the contexts in which friendships form, or something else entirely, the data does not say. But the absence of similarity on extraversion, the trait most people would intuitively assume matters for social compatibility, is striking.

“This suggests that people likely do not have friends who are necessarily similarly extraverted or introverted,” said lead author Hyewon Yang.

We think our friends are far more like us than they are

The second finding introduces a more uncomfortable implication. When the researchers compared actual similarity to perceived similarity, the gap was large and consistent across all five traits. People systematically believe their friends are more similar to them than those friends actually rate themselves to be. The friendships that exist in people’s minds are considerably more matched than the friendships that exist in reality.

This phenomenon has a name in social psychology: assumed similarity, or social projection. Humans have a tendency to use themselves as a reference point when forming impressions of others, and this tendency is especially strong for people we are close to. The more comfortable and familiar a relationship feels, the more we fill in gaps in our knowledge of the other person with information drawn from ourselves.

The implication is that part of what makes a friendship feel like a connection between kindred spirits may be a cognitive construction rather than an accurate perception. You believe your close friend sees the world the way you do not because you have thoroughly verified that they do, but partly because closeness itself creates the impression of similarity even when that similarity is limited.

Similarity does not predict how good the friendship feels

This is the finding that most directly challenges the dominant assumption. Neither actual similarity nor perceived similarity predicted friendship satisfaction. Being matched on personality, in reality or in perception, did not make friendships better. Two people could be nearly identical across all five personality traits, or completely different, and their satisfaction with the relationship was equally unpredicted by that matching.

“Neither actual nor perceived personality similarity between friends predicted how satisfied they were with their friendships,” Yang said.

If matching on personality does not drive friendship quality, what does?

What actually determines friendship satisfaction

The answer the data points toward is less romantic than similarity but more actionable. Friendship satisfaction was driven primarily by the absolute level of positive traits in the friendship, not by how closely those traits were shared.

People who scored high on agreeableness and conscientiousness reported higher overall friendship satisfaction. More agreeableness means more kindness, empathy, and cooperative warmth in daily interactions. More conscientiousness means more reliability, follow-through, and consideration. These traits make someone easier to be around not because they mirror your own personality but because they produce behaviors that make a relationship feel safe and dependable regardless of who is in it.

The same logic applied to perceptions of friends. Participants who viewed their friends as highly agreeable and conscientious reported significantly more satisfaction with those friendships. Critically, what drove that satisfaction was not perceiving the friend as similar to oneself. It was perceiving the friend as simply good, in the sense of warm, kind, and reliable.

The role of neuroticism adds a complementary dimension. Participants who reported higher neuroticism, meaning a stronger tendency toward anxiety, emotional instability, and negative affect, were less satisfied with their friendships overall. Emotionally volatile people find social relationships harder to maintain and more difficult to feel settled within. And having friends who score high on neuroticism was similarly associated with lower friendship satisfaction, likely because frequent emotional turbulence strains even willing relationships over time.

Why this matters and what it does not mean

The study has real limitations. The sample was young, averaging 19 years old, with a large majority of women, and friendships averaged only three and a half years in length. How personality similarity affects longer, deeper friendships developed in adulthood and midlife remains an open question. The cross-sectional design captures a snapshot rather than tracking how friendships evolve. And the study examined only the Big Five traits; shared values, political beliefs, religious commitments, and life goals might show different patterns than personality.

But within those boundaries, the data makes a fairly clear argument. The folk theory of friendship, that we bond with people like ourselves and feel best about relationships with close personality matches, is not well supported when tested at this level of precision. Friends are somewhat similar, but the similarity is modest and does not predict satisfaction. What predicts satisfaction is whether the people you are closest to are genuinely good to be around: agreeable, dependable, emotionally stable.

The most comforting implication of the finding may be about what friendship actually requires. It does not require finding your psychological mirror. It requires finding people who are reliably kind and who show up consistently. Those qualities are not about similarity. They are about character.

Source

Hyewon Yang, Atea Nelson, Lisa Stuckman, Grace Yancho, Lindsay S. Ackerman, M. Brent Donnellan, William J. Chopik, Richard E. Lucas. “Friends’ Personality Similarity and Its Association with Friendship Well-Being.” Social Psychological and Personality Science, 2026.
DOI: 10.1177/19485506261450240