Run Game The game

What people look for in a partner: what research finds

6 min read

Ask people what they want in a partner and the answers are fairly consistent. Watch who they actually get interested in and the picture gets messier. Both findings are real, and the gap between them is one of the more useful things in the research.

Everything below describes groups and averages. A study can say that one group, on average, rates a trait higher than another. It can't tell you what the person in front of you wants.

What people say they want

Two early studies by David Buss and Michael Barnes (1986) asked US adults to rate or rank traits in a mate. In the first, 184 people (92 married couples, aged 18 to 40) rated 76 characteristics. The ten rated highest across the whole sample were good companion, considerate, honest, affectionate, dependable, intelligent, kind, understanding, interesting to talk to and loyal. Things like being tall or wealthy sat near the bottom of that list.

In the second study, 100 unmarried undergraduates ranked 13 traits. "Kind and understanding" came first. "Exciting personality" came second.

So when the question is broad, character traits dominate. That's a pair of small, US-based samples from the 1980s, but the pattern of kindness near the top shows up often enough that it's a reasonable starting point.

Where stated preferences differ by sex

The same studies, and many since, found averages that differ. In the 1986 ranking study, men gave physical attractiveness more weight than women did, and women gave good earning capacity more weight than men did.

Buss's 1989 study widened this to 37 samples from 33 countries (10,047 people in total). Women valued cues to resources more highly than men did, and men valued youth and physical attractiveness more highly than women did.

A 2020 replication by Kathryn Walter and colleagues, including Buss, used a new sample: 14,399 people in 45 countries. The abstract reports that men, more than women, preferred attractive, young partners, and women, more than men, preferred older partners with financial prospects. It also found that couples' ages sat closer together in countries with more gender equality. The abstract doesn't give effect sizes, so how large these averages are, compared with the variation among individuals, isn't something it settles.

Where stated preferences stop predicting

Here's the complication. In a 2008 study by Paul Eastwick and Eli Finkel, 163 Northwestern undergraduates filled in a questionnaire about their ideal partner, then went speed dating, with follow-up afterward. The usual sex differences appeared in what people said they wanted. But once they met real people, the link between a partner's attractiveness or earning prospects and how interested someone was showed no sex difference. And what people said they wanted beforehand didn't predict who they got excited about on the night.

A 2014 meta-analysis by Eastwick and colleagues, as summarized by UT Austin News, pooled 95 studies from 15 countries and found the same thing: attractiveness predicted positive evaluations of real partners about equally for men and women, and earning prospects had a smaller effect that was also similar across sexes. Sex differences did show up for online dating profiles, where men weighted attractiveness more. A 2015 replication of the speed-dating study (307 participants) reached a similar result. The authors of the meta-analysis said their results don't undermine evolutionary approaches. They said sex differences look different when people judge abstract traits than when they evaluate someone they've met.

There's a counterpoint. Daniel Conroy-Beam and Buss (2016) studied three samples of people in relationships (214, 259 and 294 people; the first was newlywed couples) and found that partners tended to sit close to each person's stated ideals across several traits. The authors list limits: preferences and partner traits were measured at the same moment, so it can't show whether the ideals guided the choice or people adjusted ideals to fit who they were with. Two of the three samples relied on self-report.

A large 2024 registered report from the Psychological Science Accelerator (10,358 people in 43 countries, abstract via OpenAlex) landed in between. People did evaluate partners somewhat more positively when they matched their stated ideals, but the effect for specific traits was very small (an average standardized effect of .04).

Read together, the studies don't say preferences are meaningless. They say stated preferences describe one thing, and live attraction in a room describes another.

Chemistry is hard to predict

In 2017, Samantha Joel, Eastwick and Finkel ran two speed-dating studies where people completed more than 100 measures of traits and preferences first. Machine-learning models could predict some of how much a person tended to like others, and how much they were liked. But they couldn't predict who liked whom in particular, beyond those general tendencies. The authors suggest that this kind of compatibility is difficult to predict before two people meet. The usual caveats apply: it's four-minute dates, and a better method might do better.

For long-term relationships, a 2020 PNAS paper by Joel and co-authors from 29 labs analyzed 43 datasets (11,196 couples). People's own judgments about the relationship, such as how committed their partner seemed and how appreciative they felt, were the strongest predictors of how satisfied they were. Personality traits added little, and none of the measures predicted whose satisfaction would rise or fall later. That's about satisfaction in existing couples, not about who people are first drawn to, so it answers a different question.

What this means in practice

What to do next

  1. Write down your own list of five traits you'd want in a partner, then circle the ones you'd notice in a first conversation (humor, curiosity, warmth) versus those that only show over weeks (reliability, honesty).
  2. After your next three conversations, note in one line what actually held your attention. Compare it to your list.
  3. Ask one open question about values or habits on a first meeting ("What's a weekend you'd call perfect?") instead of running through a checklist.
  4. Don't screen out someone over a single stated trait after ten minutes. Stated preferences are weak predictors, for you as much as for anyone.
  5. Read how to be more attractive for the parts you can change, and choosing a compatible partner once you're deciding who fits you. Status and value in dating covers a related set of terms.

If you log approaches in Run Game, the notes field is a good place for item 2: a line on what made a conversation good tells you more over a month than the success rate does. For more on one researcher's view of attraction, see Helen Fisher, and for the broader picture of what's known, how attraction works.

Sources