Compatibility talk usually assumes a matching problem. Whether the folk theory is “opposites attract” or its more evidence-backed rival, “birds of a feather” – what researchers call assortative mating – the underlying picture is the same: there is some alignment between two people’s traits, and relationship quality depends on getting the alignment right. Dating profiles, matchmaking algorithms, and most advice about “finding your person” all inherit this assumption.
A 2026 study in the Journal of Research in Personality by Marta Kowal at the University of Wrocław suggests the picture is more interesting than either folk theory. Kowal analyzed survey data from 41,606 partnered adults across 74 countries – a deliberate corrective to a literature built mostly on Western, educated samples. Participants rated themselves and their partners on nine traits, including kindness, physical attractiveness, health, religiousness, resources, education, and political orientation, and completed a short form of Sternberg’s Triangular Love Scale along with a relationship satisfaction measure. The analysis compared what happens when partners are rated as evenly matched on a trait against what happens when one is rated higher than the other.
For the traits that predicted relationship quality most strongly, matching turned out not to be the point. Kindness was the single strongest predictor – self- and partner-ratings together accounted for roughly 21% of the variance in satisfaction, which is unusually large for a single trait measured with a single item – and the happiest participants were not those who saw themselves and their partner as equally kind. They were those who believed their partner was kinder than they were. The same asymmetry held for physical attractiveness. Researchers call this partner idealization, and it sits close to the older literature on positive illusions: the well-documented tendency of satisfied partners to view each other more generously than the evidence strictly warrants. Alongside the asymmetry there was a plainer additive effect – quality was highest when both partners were rated high – so this is not a case for pairing with someone out of your league, but for a surplus of perceived good qualities, tilted in your partner’s favor.
Political orientation behaved differently. There, similarity mattered in the symmetric way the folk theory predicts: the further apart the ratings, the lower the relationship quality, regardless of which partner leaned which way. And culture moderated the whole pattern. In wealthier, more individualistic countries with high relational mobility, kindness and attractiveness dominated; in less modernized ones, matching on social class and education carried more weight. “Opposites attract” found little support anywhere, but the reason depends on which trait, and which country, you ask about.
The caveats are familiar but worth stating plainly. Only one member of each couple was surveyed, so the study captures perceived traits and perceived similarity, not actual ones – there is no way to know whether the idealizing partners are describing kind partners or describing partners kindly. It is cross-sectional, and the causal arrow may well run backwards: Kowal is candid that people in love rate their partners more favorably, which would inflate the idealization effect. The trait measures are single items. And an effect this dependent on perception is best read as a fact about how satisfied people see their partners, not yet a recipe for becoming satisfied.
Still, the reframe survives the caveats. The compatibility question most of us were taught to ask – are we alike enough? – appears to be the right question only for values, politics chief among them. For the traits that matter most, the better question is closer to the one I wrote about in the context of the emergent love model: not whether you found a matching person, but what the two of you perceive and practice toward each other. Kindness is unusual among the nine traits in being both a perception and a behavior – it can be noticed more, and it can be done more, and the additive finding suggests both directions help.
When you describe your partner to a friend, do they come out sounding a little better than you? And if not, is that an assessment – or a habit?