Do Opposites Attract? What the Research Actually Shows

"Opposites attract" is one of the most durable ideas in dating, partly because everyone knows a couple who seems to prove it: the quiet engineer and the loud founder, or the planner married to the improviser. The research says those couples are real but unusual. They also tend to be less opposite than they look, because the differences you notice at a dinner party are rarely the ones that show up in the data.

Do Opposites Attract? The Largest Answer So Far

The most comprehensive test is a 2023 paper in Nature Human Behaviour by Tanya Horwitz, Jared Balbona, Katie Paulich and Matthew Keller. The team did two things.

First, they ran meta-analyses on 22 traits, pooling 480 partner correlations from 199 peer-reviewed studies of married, engaged, cohabiting and co-parenting couples. Second, they calculated correlations for 133 traits directly from up to 79,074 male–female couples in the UK Biobank.

The UK Biobank results are the cleanest answer to the "opposites" question. Of the 133 traits, 118 showed a partner correlation that was statistically significant after correcting for multiple comparisons. Of those 118, 115 were positive and 3 were negative. The mean partner correlation across traits was .19.

That fits what earlier research had found. Summarizing the prior literature, the authors write that partner correlations "are overwhelmingly in the positive direction, with only a handful of examples of significant negative partner correlations." Opposites attracting is the exception, not the rule.

Which Traits Do Couples Share Most?

Not all similarity is equal. The meta-analysis found a clear ranking. Here are the mean correlations between partners for selected traits (a correlation of 0 means no more alike than strangers; 1 would mean identical):

In the UK Biobank data, the same pattern held. Political and religious attitudes, education and some substance-use traits sat at the top. Birth year was the highest correlation of all (.87), which simply reflects that people partner with people their own age. Height came in at .27 and body mass index at .25.

The shape of that table is the most useful takeaway. Couples are strongly matched on values and life trajectory, moderately matched on physical traits, and only weakly matched on personality.

Where Opposites Actually Do Attract

The three negative correlations deserve a look, because they're the only real evidence for the folk wisdom. The largest was chronotype. In UK couples, whether someone was a morning or an evening person had a partner correlation of −.18. Morning people were somewhat more likely to be paired with evening people than chance would predict.

That's a weak effect, and the study doesn't explain it. It could be attraction to a complementary rhythm. It could also be something after the fact, like one partner shifting schedules around the other, or the couples who stayed together long enough to be in a biobank of 40-to-69-year-olds being the ones whose schedules fit. The paper measures resemblance; it doesn't measure why.

Personality is the other place where "opposites" gets its reputation. With extraversion at .08, an introvert is barely more likely than chance to end up with another introvert. That isn't evidence that opposites attract. It's evidence that on extraversion, it mostly doesn't matter either way. That's why the loud-and-quiet couple is common enough to become a cliché.

What the Data Can't Tell You

The Horwitz paper is strong, but it has limits the authors spell out:

• Resemblance isn't the same as preference. Couples can end up similar because they chose each other for it, because they met inside the same schools, jobs and neighborhoods, or because they grew more alike over time. The authors call that last one convergence, and it isn't a type of mate choice at all. Education is a clear case: people often meet within their profession, so similar schooling can come with the territory.
• The UK Biobank is not a dating pool. Participants were 40 to 69 when recruited, were healthier and less economically deprived than the general population, and were mostly of European ancestry. The couples were also inferred by the researchers from household data, not self-reported.
• Results vary a lot by sample. The meta-analysis found high heterogeneity for most traits, even objective ones like height. Partner similarity seems to differ genuinely across places, times and cultures, so these numbers are averages, not laws.
• Only male–female couples were analyzed. The authors excluded same-sex samples because the patterns may differ and there's less data.

Feeling Similar vs. Being Similar

A second body of research complicates the tidy "birds of a feather" story in a way that matters for early dating.

A 2008 meta-analysis by R. Matthew Montoya, Robert Horton and Jeffrey Kirchner pooled 460 effect sizes from 313 laboratory and field studies of similarity and attraction. Both actual similarity (r = .47) and perceived similarity (r = .39) predicted attraction overall. But the pattern changed as people got to know each other. Actual similarity mattered most in studies where people never interacted or interacted briefly, and its effect in existing relationships was not significant. Perceived similarity, meaning how alike you think you are, predicted attraction at every stage, including in existing relationships.

A speed-dating study by Natasha Tidwell, Paul Eastwick and Eli Finkel, published in Personal Relationships in 2013, tested this in a live romantic setting. Participants had four-minute dates with about 12 people each. Actual similarity, calculated from questionnaires they filled out beforehand, did not significantly predict romantic liking. Perceived similarity did. (The same research group's work on what predicts attraction in speed dating found a related gap between what people say they want and whom they actually like.)

Put the two literatures together and you get a coherent picture:

1. Long-term couples are genuinely similar on values, education and background.
2. In the moment of attraction, people respond to a sense of similarity, the feeling of "we get each other," more than to measured overlap.
3. That feeling can be wrong in both directions. You can feel instantly aligned with someone whose politics you'd find hard to live with. You can also feel lukewarm about someone whose values match yours closely.

What This Means for Choosing a Partner

The research doesn't say you should only date your clone. Personality similarity barely registers in lasting couples, which is good news for anyone who's worried their partner is too quiet, too social or too different in temperament. What the data does suggest is that the differences that tend to go the distance are differences of style, not of values.

That's also the argument for screening on values earlier than most people do. Dating apps surface photos, heights and jobs. Those are mostly the moderate-correlation physical and status traits, and they say little about the values that couples in the data share most. Curated matching tries to put that ordering right. At Tenr, matches are chosen by people rather than an algorithm optimizing for swipes, and the first meeting is a 10-minute video date. That's enough time to check whether the sense of "we're alike" survives a real conversation, without spending an evening on it.

Sources:

• Horwitz, T.B., Balbona, J.V., Paulich, K.N. & Keller, M.C. (2023). Evidence of correlations between human partners based on systematic reviews and meta-analyses of 22 traits and UK Biobank analysis of 133 traits. Nature Human Behaviour, 7, 1568–1583.
• Montoya, R.M., Horton, R.S. & Kirchner, J. (2008). Is actual similarity necessary for attraction? A meta-analysis of actual and perceived similarity. Journal of Social and Personal Relationships, 25(6), 889–922.
• Tidwell, N.D., Eastwick, P.W. & Finkel, E.J. (2013). Perceived, not actual, similarity predicts initial attraction in a live romantic context: Evidence from the speed-dating paradigm. Personal Relationships, 20(2), 199–215.