Do Opposites Attract? A Study of 133 Traits Says Almost Never

“Opposites attract” may be the single most repeated claim in dating — repeated at weddings, in movies, by the friend explaining why the mismatch will work this time. It is also among the best-evidenced things we can call wrong. Not lightly wrong, not wrong-in-one-study wrong: in 2023, researchers assembled roughly a century's worth of couple research and one of the largest single datasets ever pointed at the question, and the answer came back the same way from both. Across trait after trait, people who end up together are alike far more often than they are opposite — and genuinely opposite pairings are rare enough to be the exceptions that get named individually.

What the Study Actually Did

The anchor is a 2023 paper by Tanya Horwitz, Jared Balbona, Katie Paulich, and Matthew Keller in Nature Human Behaviour: “Evidence of correlations between human partners based on systematic reviews and meta-analyses of 22 traits and UK Biobank analysis of 133 traits.” The title is the method. It ran two arms and let them check each other.

The first arm gathered what everyone else had already measured: a systematic review and meta-analysis of partner similarity across 22 traits, drawing on roughly 200 peer-reviewed studies of real couples — married, engaged, cohabiting, co-parenting — published across more than a century of research. The second arm went to the UK Biobank and measured 133 traits directly, in nearly 80,000 actual couples. One arm aggregates the field's history; the other tests it fresh at scale. If “opposites attract” described real pairings, somewhere in 133 traits it would have to show up.

What It Found

The rule, across both arms, is similarity. For the overwhelming majority of traits, partners resembled each other — the correlations ran positive, not negative. The resemblance was strongest exactly where you might guess once you hear it: political attitudes, religious attitudes, educational attainment, and some substance-use habits. The person you build a life with tends to vote like you, believe like you, have sat in classrooms about as long as you, and take roughly your position on drinking and smoking. Psychological traits — personality, mental health — and physical measurements showed weaker resemblance, but still resemblance: weaker similarity, not opposition.

And the opposites? Genuinely negative correlations — cases where having a trait makes your partner less likely to have it — were rare enough that the paper's abstract names its example in the singular: chronotype. Morning people pair with night owls slightly more often than chance would predict. That is roughly the size of the concession the evidence makes to one of our culture's favorite romantic claims: whether you are a morning person. Almost everywhere else the folk wisdom points, the data points the other way.

The Trap: Similarity Is a Pattern, Not a Mechanism

Here is where most write-ups of this research go wrong, and where we want to be careful. The finding is that partners are similar. It is not a finding about why — and “you are drawn to people like you” is only one of at least three explanations that would all produce the same numbers.

You might choose similar people — that is the attraction story. But similar people also meet: schools, workplaces, neighborhoods, and social circles are already sorted by education, belief, and habit, so the pool you draw from resembles you before any preference kicks in. And couples converge: two people who share meals, elections, and decades quietly grow more alike, which means some measured similarity is a consequence of the relationship rather than a cause. The paper documents the pattern; disentangling those three engines is genuinely hard, and anyone who tells you the data proves you are magnetically drawn to your mirror image has skipped that entire problem. What the evidence rules out is narrower and cleaner: whatever the engine, systematic opposite-seeking is not what's happening.

The Counterexample Everyone Raises

There is one pairing that seems to defy all of this, and if you have dated much you have probably lived it: the anxious–avoidant pull, where a person who craves closeness and a person who guards distance find each other with what feels like magnetic reliability. We wrote a full piece on why that pull feels like chemistry and works like a trap, and we will not relitigate it here — except to note that it is not actually a counterexample. Attachment styles are a dynamic — two anxieties interlocking — not a trait one partner has high and the other low, and the intensity of that loop is precisely why it gets mistaken for the “opposites” magic. The research above is about who durably pairs; the anxious–avoidant loop is famous for producing the other kind of relationship.

Who This Describes — and Who It Might Not

The honest caveat, as always, gets its own section. The largest arm of this research is the UK Biobank: adult couples in the United Kingdom, mostly of European ancestry, from specific birth cohorts, and — crucially — already partnered, often for a long time. That is a picture of who ends up and stays together in one society, not a log of who swipes on whom, and not a guarantee that the same pattern holds at the same strength across every culture, generation, or ancestry group. The meta-analysis arm reaches wider, but historical couple research has its own skews toward Western, married samples. Strong evidence about the pattern where it was measured; appropriate humility everywhere else.

What This Is Not

This is not advice to date your clone. Nothing in the research says difference is bad for a relationship — it says difference is not the organizing principle of who pairs with whom. Those are different claims, and collapsing them produces advice as bad as the myth it replaces. Plenty of strong couples differ loudly in temperament, hobbies, and taste; what the data suggests they are unlikely to differ about is the deeper substrate — beliefs, values, education, habits. Which is why, if you want the practical version of this article, it already exists: shared values matter more than shared interests. The research here is the evidence floor under that piece, not a replacement for it.

The Bottom Line

“Opposites attract” survives as a story because it is a better story — tension is narrative, and two people discovering they already agree is not a movie. But as a description of real pairings, it fails across roughly 200 studies, nearly 80,000 couples, and 133 measured traits, surviving mainly in whether you like mornings. The practical takeaways are quiet ones. Stop treating friction as a sign of destiny — that instinct is the myth talking, and often the nervous system talking, which we have also written about. Stop apologizing for wanting someone who shares your fundamentals; the evidence says that is what lasting pairs overwhelmingly look like. And when something like Intently's compatibility scoring weights alignment on values and life direction over novelty, this body of research is a large part of the reason why.