Case Study 3.1: The Speed Dating Studies — What We Say We Want vs. What We Actually Choose
Background
In 2005, psychologists Eli Finkel and Paul Eastwick began a research program that would quietly upend one of the most confident claims in the mate preference literature: that people's stated partner preferences are meaningful predictors of whom they actually find attractive.
Prior to their work, much of what we thought we knew about partner preferences came from survey-based methods: ask people what characteristics they want in a romantic partner, compile the results, and treat those results as evidence about human mating psychology. These surveys had generated broadly consistent findings — people said they wanted partners who were physically attractive, kind, and (with gender-moderated variation) financially successful. Evolutionary psychologists cited these results as evidence of universal mate preferences shaped by natural selection.
Finkel and Eastwick took a different approach. Rather than asking people what they want, they created conditions under which people actually chose — and then compared stated preferences to real-world choices.
The Design
Their core paradigm was speed dating: brief structured interactions, typically three to five minutes, between rotating pairs of singles. Participants completed pre-event surveys describing their ideal partner preferences — how important was physical attractiveness? earning potential? humor? emotional stability? After each interaction, participants rated the person they'd just met and indicated whether they wanted to see them again. Follow-up assessments captured longer-term romantic outcomes.
The crucial methodological innovation was this: by comparing pre-event stated preferences to post-event attraction ratings, they could test whether people who said they valued, for instance, physical attractiveness actually showed stronger attraction to physically attractive partners than people who rated attractiveness as less important. This is a direct behavioral test of whether preferences predict choices.
The Key Findings
The results were striking and, to many researchers, surprising. Stated partner preferences showed essentially no correlation with actual romantic interest in specific individuals encountered in the speed-dating context. People who said physical attractiveness was paramount showed no stronger tendency to prefer attractive partners than people who said it mattered less. People who said they wanted an ambitious partner did not show stronger interest in partners who were (objectively rated as) more ambitious.
What did predict romantic interest? Individual difference variables that people rarely listed as explicit preferences: how physically attractive the person was judged to be by a group of raters (though this predicted ratings for everyone, not specifically for people who said they valued attractiveness), and — interestingly — how enthusiastic and energetic the person seemed during the brief interaction.
Why the Design Works
The speed dating paradigm has several features that give its findings higher ecological validity than typical laboratory tasks. Participants are genuine singles who believe they may be meeting real potential partners; the stakes feel real. The interaction, while brief, is dynamic and bilateral — unlike rating a photograph. The outcome measure (desire to see the person again) is a genuine behavioral choice rather than a Likert-scale rating. And the context, while somewhat artificial, bears more resemblance to real social encounters than most laboratory attraction paradigms.
Finkel and Eastwick also used this platform to test the "matching hypothesis" — the idea that people tend to end up with partners of roughly equivalent social desirability. In the speed-dating context, the matching effect largely disappeared: people with high and low desirability showed similar preference patterns in who they wanted to see again; it was only in who wanted to see them that high-desirability individuals differed.
Methodological Limits That Remain
Speed dating is not the real world. The interactions are brief and constrained by format. The settings tend to attract participants who are already relationship-motivated and willing to engage with a somewhat formal structure. Self-selection means speed daters may not represent singles broadly. And three-to-five minute interactions don't capture the processes by which attraction typically develops in ongoing social relationships — through repeated exposure, shared experience, and the gradual discovery of compatibility.
The research also concentrated heavily on heterosexual participants; same-sex speed-dating designs have produced broadly similar findings, but the dataset is much smaller.
Discussion Questions
- How does the Finkel-Eastwick finding about stated preferences versus actual choices relate to the methodological distinction between self-report and behavioral measures discussed in Chapter 3? What does it suggest about the limits of preference surveys?
- Speed dating is an improvement on many laboratory paradigms but still not the real world. What methodological design could you imagine that would have even higher ecological validity? What would you sacrifice?
- The matching hypothesis was tested and found to be weaker in speed-dating contexts than in real relationship formation. What does this suggest about the importance of research context for attraction findings?