There's an enormous amount of research on romantic compatibility and personality. There's almost nothing on friendship.
This is partly a funding problem — relationship science research gets resources because it addresses pressing social questions (divorce, marriage quality, family stability). Friendship doesn't have the same institutional weight. But it also reflects a cultural assumption: that friendships are more spontaneous, less structured, and less in need of scientific understanding than romantic relationships.
That assumption is probably wrong. Longitudinal studies show that friendship quality is one of the strongest predictors of life satisfaction, health, and longevity — equal to or stronger than marriage quality in some measures. The research on what makes friendships work is thin, but what exists is worth understanding.
Similarity vs. complementarity: which predicts friendship?
The intuitive answer differs by question. For romantic partners, people often believe in complementarity — opposites attract. For friends, people often believe in similarity — you get along with people like you. Research supports the friendship intuition more than the romantic one.
Similarity in values and interests is one of the strongest predictors of friendship formation. Studies on college roommates, work colleagues, and neighborhood acquaintances consistently find that initial friendship formation is predicted by proximity and perceived similarity. We become friends with people who seem to share our worldview, interests, and orientation.
Big Five similarity has a more nuanced relationship with friendship quality. High Openness shows clear homophily — people high in openness tend to form friendships with others high in openness. This makes sense: the primary activity of openness-high friendships is intellectual exploration, and that requires a partner who's also interested in ranging across ideas.
Agreeableness shows a different pattern. High-agreeableness people are easier to befriend — they're warm, cooperative, and conflict-avoidant. But high-agreeableness friendships can become stagnant because conflict-avoidance prevents the kind of honest feedback that deepens relationships over time.
Conscientiousness similarity predicts friendship stability more than formation. Low-conscientiousness people often form friendships with high-conscientiousness people (the responsible friend who manages logistics and remembers birthdays) — but these friendships are more likely to fray over time as the asymmetry generates resentment.
Attachment style in friendship
Attachment theory was developed in the context of romantic relationships but applies to friendship in ways that are often underappreciated.
Secure attachment in friendship looks like: comfortable with closeness and distance, able to be present without anxiety when apart, able to address conflict directly without catastrophizing. Secure friends are the easiest to maintain relationships with over long periods and across geographic distance.
Anxious attachment in friendship produces characteristic patterns: high investment in the friendship, high sensitivity to signs of distance or withdrawal, difficulty with friends who are unreliable or intermittent. Anxiously attached people often have intense, close friendships that are also high-maintenance — they need more reassurance of the friendship's stability than securely attached people do.
Avoidant attachment in friendship produces a different pattern: comfortable with less frequent contact, uncomfortable with high emotional intensity or dependence, tends to keep friendships in a more bounded, activity-based mode. Avoidantly attached people can have many long-lasting friendships that stay relatively shallow.
The attachment compatibility question in friendship is less well-studied than in romantic relationships, but the structural dynamics are similar. Anxious-avoidant friendship pairings can be particularly difficult: the anxiously attached friend seeks more contact and reassurance, the avoidantly attached friend pulls back in response to pressure, which escalates the anxious friend's need for reassurance. This cycle is most common in romantic relationships but appears in close friendships too.
Anxious-anxious pairings can be intensely close and mutually validating, but also high-drama when both people need reassurance simultaneously.
Secure-secure pairings are the most stable across time and circumstances — lower maintenance, higher resilience to conflict and distance.
What the research shows about what kills friendships
More is known about friendship dissolution than friendship quality. The predictors of friendship loss are informative:
Geographic distance. The most well-documented predictor of friendship loss. Most friendships maintained primarily through proximity dissolve or significantly weaken after relocation. The friendships that survive distance share specific features: higher baseline closeness, more deliberate maintenance effort, and communication that isn't entirely contingent on shared context.
Life transitions. Marriage, parenthood, career changes, and health crises all restructure social networks. Friendships that were embedded in a prior life context (college, a job, a neighborhood) often don't survive the transition intact. The ones that do tend to have a shared identity that extends beyond the context.
Asymmetric investment. Friendships where one person consistently invests more time, emotional labor, and logistical effort than the other tend to erode — even when both people value the friendship. The asymmetry generates resentment even when it isn't named.
Unaddressed conflict. Conflict avoidance is a short-term friendship preservative and a long-term friendship killer. Friendships that survive decades of close contact have almost always navigated explicit disagreement, hurt feelings, and periodic rupture-and-repair cycles.
Why personality-based friendship compatibility matters now
Until recently, personality-based compatibility was only available as a quiz result — "here's your type, here are types that get along with yours." The insight was limited because the underlying data was self-report, and the output was generic.
The availability of behavioral personality data changes what's possible. If your personality profile is built from how you've actually operated across hundreds of real interactions — not how you've described yourself on a questionnaire — then compatibility predictions have a different foundation.
This is part of what Memrov Match will do when it launches: offer compatibility analysis built from finalized personality profiles derived from behavioral data. Not quiz-based type matching, but actual Big Five, HEXACO, and attachment style profiles compared across the dimensions research has shown to matter for connection quality.
The friendship application is one Memrov is thinking about deliberately. Romantic compatibility gets all the attention. But the research on what makes friendships work — and what makes them fail — is real, and the same profile that informs romantic compatibility is equally relevant to the friendships you want to invest in and maintain over decades.
Memrov builds your personality profile from your AI conversation history — the foundation for understanding how you connect, what you need from relationships, and who you're likely to work well with. Start with the free personality test →