What the data actually shows
The most direct evidence comes from research on similarity and attraction. A large meta-analysis by Montoya, Horton and Kirchner pulled together decades of studies and found a consistent positive link between similarity and attraction — people reliably like others who share their attitudes and traits. The effect was strongest for perceived similarity (believing you are alike), but actual similarity also mattered.
Studies of real couples point the same way. A large-scale analysis by Horwitz and colleagues (2023), examining many traits across millions of couples, found that partners tend to be similar on a wide range of characteristics — from attitudes and education to some health and behavioural traits — and found little evidence that opposites systematically pair up. This pattern, where like associates with like, is what researchers call homophily, and it shows up in friendships and partnerships alike.
Where complementarity does appear, it tends to be narrow and specific rather than a general principle. Couples may divide some roles or preferences in ways that fit together, but on the broad dimensions that predict satisfaction and stability — shared values, similar outlooks — the evidence favours similarity, not difference. The research does not support the idea that being opposite is, on its own, what creates lasting attraction.
Why this feels different from how it actually is
The opposites-attract idea persists partly because differences are memorable. A couple who clash on temperament or habits makes a better story than a couple who quietly agree on most things, so the vivid exceptions stick in the mind while the unremarkable similar pairings fade into the background.
Early attraction can also be confusing. Novelty and difference can feel exciting at first — someone unlike you can seem intriguing precisely because they are unfamiliar. But that initial spark of novelty is not the same as the deeper similarity that research links to lasting relationships, and the two are easy to conflate in the early weeks.
There is also a kernel of truth that gets over-generalised. Most couples differ somewhere, and people notice and talk about those points of contrast. From there it is a short step to the tidy but inaccurate conclusion that the differences are what hold the relationship together, when the data suggests the shared ground is doing most of the work.
What the research says to do about it
If anything, the research is reassuring about being drawn to people who are like you. Shared values and outlook are among the more consistent correlates of relationship satisfaction, so feeling at ease with someone who sees the world similarly is not a lack of spark — it is closer to what the evidence associates with durability.
Where differences exist, the research on relationships points less to whether you are alike and more to how you handle the gaps. Work on conflict and communication — by researchers such as the Gottmans and others — suggests that managing disagreement well matters more for outcomes than the raw number of things you differ on. Differences are workable; how a couple navigates them is the more important variable.
It can also help to distinguish surface differences from core ones. Differing in hobbies, energy levels or small preferences is common and easily accommodated. Differences in fundamental values, goals or how you want to live are the ones worth taking seriously, since those are the dimensions where similarity tends to matter most.
What the research says does not help
Treating 'opposites attract' as relationship advice does not help, because it is not well supported. Seeking out a partner mainly because they are your opposite, or worrying that a similar partner means there is no chemistry, both rest on a premise the data does not back.
Equally, assuming that two similar people will automatically have an easy relationship is a mistake. Similarity is associated with attraction and stability on average, but it is not a guarantee — communication, life circumstances and how a couple handles conflict all matter, and no degree of likeness removes the ordinary work of being together.
Reading too much into early novelty also misleads. The excitement of dating someone very different can fade as unfamiliarity wears off, so mistaking that initial buzz for a sign of long-term fit can point people in the wrong direction. Early intensity is a poor predictor; shared values and how you treat each other are better ones.
Real numbers in context
The headline pattern is consistency, not a single dramatic statistic. The Montoya, Horton and Kirchner meta-analysis combined many studies and found a reliable positive relationship between similarity and attraction across them — the kind of broad, repeated result that gives a finding weight, even though any one effect size is moderate.
Large couple studies reinforce it. Horwitz and colleagues (2023) reported that across a long list of traits, partners resemble each other more often than chance would predict, while traits where opposites pair up are rare. The honest summary is that similarity is the rule and opposites attracting is, at best, the uncommon exception — closer to a memorable story than a measured pattern.