What Speed-Dating Research Reveals About Who You'll Actually Like
Psychologists have run hundreds of controlled speed-dating events to study attraction. The core finding: people can't predict who they'll want. Here's the data.
Quick Answer
Controlled speed-dating research — where psychologists run real events with 4-minute dates and track who actually wants whom — consistently finds that people's stated preferences don't predict their live attraction. In Eastwick and Finkel's 2008 Northwestern study, participants' pre-event ratings of how much they valued looks or earning prospects failed to predict which real partners they desired, and the well-documented sex differences in stated preferences (men say looks, women say earnings) disappeared entirely in vivo. One earlier dataset found stated preferences correlated with actual yes/no decisions at just .00 to .17. The takeaway for anyone filtering matches by checklist: the checklist is measuring something, but it isn't who you'll want.
Why Speed Dating Became a Research Tool
For most of the 20th century, attraction research ran on questionnaires: rate how important looks, income, kindness, and ambition are in an ideal partner. That method produced one of psychology's most replicated findings — men say physical attractiveness matters more, women say earning prospects matter more, across dozens of cultures. What it couldn't do was check whether those stated preferences had anything to do with who people actually wanted when a real person sat down across from them.
Speed dating fixed that. Starting in the mid-2000s, researchers — most prominently Paul Eastwick and Eli Finkel at Northwestern — began running full speed-dating events under controlled conditions: real participants, real 4-minute conversations, real yes/no decisions afterward, and follow-up tracking of what happened next. The method captures the exact moment stated preferences meet a live human, which is the moment every dating-app filter is implicitly betting on.
The Core Finding: Stated Preferences Don't Predict Live Desire
The study's headline result is stark. Participants who said looks mattered most to them were not measurably more driven by looks in their actual attraction to real partners. Participants who said earning prospects mattered were not measurably more driven by earning prospects. As the authors put it, ideal preferences assessed before the event "failed to predict what inspired their actual desire at the event." People appear to lack introspective access to what actually moves them — a finding the paper explicitly connects to Nisbett and Wilson's classic work on how poorly people report the true causes of their own judgments.
This wasn't a one-off. The paper cites an earlier speed-dating dataset (Iyengar, Simonson, Fisman, and Mogilner, 2005) in which correlations between daters' stated preferences and their actual yes/no decisions ranged from .00 to .17 — effectively noise. And a 2005 study of professional speed-dating events with older participants (mean age 32) found that income did not make either sex significantly more desirable to partners, despite "good earning prospects" ranking high on every stated-preference survey ever run.
The practical implication is uncomfortable for anyone who has set age, height, income, or education filters on an app: the research says that what you tell a form you want is a weak guide to who you'll actually feel something for. It's the empirical version of the case made in what having a "type" costs you.
The Sex Difference That Vanished
The second finding matters because it dismantles a piece of conventional wisdom. Eastwick and Finkel replicated the classic result perfectly on paper: their participants stated the traditional sex-differentiated preferences, men weighting attractiveness more, women weighting earning prospects more. Then, in the live data, the difference disappeared. Physical attractiveness predicted romantic interest about equally for men and women. Earning prospects predicted it about equally — and weakly — for both.
That reframes a lot of dating discourse. Debates about what "men want" or "women want" are largely debates about what men and women say, which the live data suggests is a different and less informative question than what actually draws them to a specific person in a room.
What Does Predict Attraction in the Room
If checklists don't, what does? Later work from the same research program, including Eastwick, Hunt, and Finkel's studies on how consensus about desirability breaks down as people get to know each other, points toward something less filterable: attraction is substantially relational — a product of the specific interaction between two particular people rather than a sum of either person's traits. Two daters with identical checklists will be drawn to different people, and the same person will be highly desirable to one partner and not another, in ways that no profile can forecast.
This is the strongest research case for meeting people quickly rather than screening them extensively first. The 10-minute video date format exists because a short live conversation captures the relational signal that filters structurally cannot — the same logic that makes "no spark" a noisy first read but a live meeting an informative one. Speed-dating research, in effect, has been validating the format for two decades.
What This Means for How You Date
Three practical conclusions follow directly from the data:
- Loosen the filters. Hard cutoffs on the traits that show up in stated-preference surveys — height, income, education — are screening on variables the research finds weakly predictive of live attraction. They're removing people you might have wanted.
- Trust the room over the profile. A 4-minute conversation produced better information about mutual interest than any pre-event questionnaire. That ratio holds for apps: weeks of profile evaluation and messaging are a worse predictor than a short real conversation.
- Be skeptical of your own theory of what you want. The people in these studies were sincere. They were also wrong about themselves. If your last three relationships didn't match your stated type, the research suggests the type is the thing to revise.
Sources:
- Eastwick, P. W., & Finkel, E. J. (2008). Sex Differences in Mate Preferences Revisited: Do People Know What They Initially Desire in a Romantic Partner? Journal of Personality and Social Psychology, 94(2), 245–264.
- Finkel, E. J., & Eastwick, P. W. (2008). Speed-Dating. Current Directions in Psychological Science, 17(3).
Frequently Asked Questions
Do people know what they want in a partner?
Not reliably, per controlled speed-dating research. In Eastwick and Finkel's 2008 study, participants' stated ideal preferences — how much they said they cared about looks or earning prospects — failed to predict which real people they were actually attracted to after a 4-minute date. An earlier dataset found correlations between stated preferences and actual yes/no decisions ranging from .00 to .17.
Do men and women want different things in a partner?
On paper, yes: across decades of surveys men rate physical attractiveness as more important and women rate earning prospects as more important. In live speed-dating, the difference vanishes — the 2008 Northwestern study found no sex differences in how much attractiveness or earning prospects actually predicted romantic interest in real partners.
How long is a speed date in research studies?
Typically about 4 minutes. In Eastwick and Finkel's study, each participant had 4-minute dates with 9 to 13 opposite-sex partners at a 2-hour event, then rated every partner and reported yes/no interest afterward, with a 30-day follow-up on any matches.
Why do speed-dating studies matter for regular dating?
They're one of the few methods that captures attraction to real people in real time rather than reactions to hypothetical profiles. The consistent finding — that filter-style preferences don't predict live attraction — is a direct critique of screening partners by checklist, which is how most dating-app filters work.
Does income make someone more attractive in speed dating?
Not measurably. A 2005 study of professional speed-dating events (participants averaging 32 years old) found that income did not make either men or women significantly more desirable to partners, despite earning prospects ranking high in stated preferences.
Related reading
Most Single Americans Aren't Looking to Date. Here's the Data
Pew found 57% of single adults aren't looking for a relationship or dates — and interest has been falling, especially among men. What that means for the people who are.
How Many Adults Are Actually Single? The Census Numbers
The 'everyone is single now' narrative peaked in 2019. Pew's analysis of Census data shows the unpartnered share has ticked down — and where singles actually concentrate.
Living Together Before Marriage: What the Research Actually Found
Does cohabiting before marriage raise divorce risk? A Stanford analysis of 216,000 couple-years says the answer depends on when you look — and on who you lived with before.
Ready to find your person?
Tenr is the invite-only 10-minute date app for high achievers.