Modern WisdomHuge New Study Reveals What People Really Want In A Partner - Dr Paul Eastwick
CHAPTERS
- 0:00 – 1:45
How well people know what they want (and where insight fails)
Chris asks whether people truly know what they want in a romantic partner. Paul explains that people are good at identifying universally desirable traits, but struggle to predict what they uniquely prefer relative to others.
- •Strong agreement on broadly desirable traits (attractive, intelligent, considerate, honest)
- •Strong agreement on broadly undesirable traits (disorganized, careless, anxious)
- •The real challenge is knowing one’s idiosyncratic (unique) preferences
- •Why studying individual differences in preference accuracy matters
- 1:45 – 3:39
Why “unique preferences” matter: gender differences as a test case
Paul describes how his interest grew out of classic research on gender differences in mate preferences. He frames gender as an individual-difference variable that should predict downstream attraction if stated differences are meaningful.
- •Long history of surveying what men vs. women say they want
- •Typical stated differences: men rate attractiveness higher; women rate earning potential higher
- •Gender differences should show predictive power if they reflect real preferences
- •Motivation: test whether stated differences appear in revealed attraction patterns
- 3:39 – 6:47
Stated vs. revealed preferences: definitions and how revealed preferences work
They clarify the core distinction: stated preferences come from what people say they want, while revealed preferences come from what traits actually predict attraction/liking across targets. Paul emphasizes revealed preference is about predictive association, not merely who someone ‘chooses.’
- •Stated preferences: rating traits in an ‘ideal partner’ (or context-specific ideal)
- •Revealed preferences: does a trait predict liking across a range of people?
- •Speed dating is an intuitive model for revealed preference measurement
- •‘Liking’ can be measured as ratings, choices, dates—any evaluative outcome
- 6:47 – 8:46
Inside the 10,000-person, 43-country study design—and the standout result
Paul outlines the dataset: ~10,000 participants across 43 countries rating either a current partner or someone they’re interested in. He explains how the team ranked traits by how strongly they predicted positive feelings, revealing surprising top predictors.
- •Participants rated a real target on 35 traits plus their ideals for the same traits
- •Dependent variable: positivity/desire toward the target
- •Revealed preference operationalized as trait → positive feelings association
- •‘Good lover’ emerged as the strongest revealed predictor despite being ~12th stated
- 8:46 – 10:35
“Ideal partner preference matching”: do we like partners who match our ideals?
They introduce the matching question: whether people feel more desire when a partner matches their personal ideal-trait profile. Paul notes the statistical complexity and explains the paper’s approach of testing multiple matching methods.
- •Matching asks: do partners who fit your ideal profile feel more desirable?
- •Many competing methods (profile correlations, scoring approaches) complicate inference
- •The study compares multiple approaches rather than betting on one
- •Goal: isolate the ‘unique’ individual-differences component from normative desirability
- 10:35 – 14:05
What matching shows (small but real) and why single-trait predictions collapse
Paul reports that overall matching effects exist but are modest (a few percent of variance). However, when focusing on any single trait, interaction effects are typically tiny—sometimes even absent for traits people assume would matter most.
- •Across 35 traits, unique matching effects are detectable but small
- •Single-trait matching relies on interaction tests that tend to be very small
- •Large samples are needed to detect most single-trait matching effects
- •Religiosity shows a comparatively stronger matching component than most traits
- 14:05 – 16:15
Biggest stated vs. revealed discrepancies: sex, scent, and ‘warm’ traits
They dig into where people most ‘miss the mark’ between what they say and what predicts attraction. Traits like ‘good lover’ and ‘smells good’ are underweighted in stated ideals but strongly predict positive feelings; some warm traits are overstated relative to their predictive power.
- •Underestimated in stated ideals: good lover; smells good (very high revealed rank)
- •Other underestimates: sexy/nice body and related attractiveness markers
- •Overestimates: some warm/prosocial traits (e.g., patient; emotionally stable)
- •Ranking comparisons highlight systematic self-misperception patterns
- 16:15 – 22:01
Sex differences: why stated gender gaps persist while revealed gaps shrink
Paul explains a central finding: men and women show similar revealed preferences for attractiveness and earning-related traits, despite persistent stated differences. The apparent gap often reflects different degrees/directions of misestimation rather than different underlying attraction drivers.
- •Revealed preferences for attractiveness-related traits are similar for men and women
- •Stated preferences still show classic gender differences
- •Women underestimate how much attractiveness matters more than men do
- •For earning traits: women slightly overestimate; men slightly underestimate—netting a stated gap
- 22:01 – 31:32
Interpreting the ‘purple pill’: first impressions, social reality, and what changes with familiarity
Chris frames the results as validating both ‘red pill’ and ‘blue pill’ narratives in different ways. Paul argues attraction is strongly influenced by first impressions (consensus traits like attractiveness), but as people interact over time, consensus fades and idiosyncratic attraction can emerge.
- •Top revealed traits blend intimacy/physicality with loyalty/honesty/supportiveness
- •First-impression contexts amplify consensus traits (attractiveness, confidence)
- •Over time, people disagree more about who is attractive (idiosyncrasy increases)
- •Practical implication: repeated-interaction environments create more pathways to attraction
- 31:32 – 38:58
Why ‘good lover’ ranks #1: abstract ideals vs visceral partner experience
They unpack why ‘good lover’ is so predictive despite being rated lower in ideals. Paul suggests the abstract concept feels limited in day-to-day impact, but in a real relationship it bundles many valued qualities (care, sensitivity, generosity) and visceral connection.
- •Abstractly, ‘good lover’ seems like a small slice of life; concretely it’s central
- •It signals broader relational competencies (giving, caring, attuned)
- •Other ‘shallow’ traits can proxy for upstream qualities (discipline, hygiene, consideration)
- •People may answer ‘ideal partner’ thinking about life logistics vs romantic/visceral reality
- 38:58 – 42:51
Unanswered questions: can matchmaking use these findings, and what is “compatibility”?
Paul identifies key limitations for real-world matchmaking: the study relies on a participant’s subjective ratings of a target, which platforms don’t have pre-meeting. They discuss what data might help predict who likes whom versus merely predicting who is generally popular.
- •Matchmaking needs traits from both sides (self-reports or independent ratings) pre-meeting
- •Predictive accuracy typically drops when moving from partner-judgments to self-judgments
- •Easy problem: predicting popularity; hard problem: predicting compatibility (fit)
- •Skepticism toward ‘secret sauce’ matchmaking algorithms without strong evidence
- 42:51 – 47:16
Bias and motivated reasoning: how we ‘rewrite’ our partner’s traits
Chris challenges whether ‘revealed’ preferences are still contaminated by perception biases. Paul agrees, explaining motivated reasoning: people interpret a partner’s traits in the most flattering way when invested, and even reframe conflict or sensitivity to protect relationship narratives.
- •Partner ratings are filtered through desire, loyalty, and self-justification motives
- •Experiments show people adopt positive frames when told a trait predicts good relationships
- •Many traits have ‘good’ and ‘bad’ versions; satisfaction biases which version is seen
- •This limits the use of subjective partner ratings for predicting attraction before people meet
- 47:16 – 54:45
Dating’s missing ingredient: networks, staged attraction, and the chaos of falling in love
They argue modern dating over-focuses on optimizing early-stage selection while neglecting social-network compounding and the time needed for attraction to develop. Chris emphasizes the unmeasurable, phenomenological ‘falling in love’ factor; Paul describes it as chaotic and hard to forecast from checklists.
- •Relationship formation is staged: you must clear early gates to reach deeper compatibility
- •Social networks create compounding opportunities that apps and ‘Petri dish’ dating miss
- •Falling in love can hinge on idiosyncratic, unpredictable interaction details
- •Researchers are moving toward more creative models beyond one-size-fits-all trait scales
- 54:45 – 1:00:49
Can preferences change over time—and can we deliberately change them?
They explore whether mate preferences shift historically and culturally (earning potential, status, gender roles), with examples from film and literature. Paul notes experimental attempts to change stated preferences show it’s difficult to create enduring changes, though expectations and narratives still matter for gender relations.
- •Uncertainty about historical continuity vs cultural shifts in what people value
- •Earning/status dynamics may be influenced by women’s workforce participation and inequality
- •Deliberately changing stated preferences is hard; lab effects don’t persist strongly
- •Even if revealed preferences converge, mismatched expectations can drive conflict
- 1:00:49 – 1:03:05
Where positive feelings about partners come from: global judgments and turning points
Paul explains that relationship positivity isn’t just a stable stored attitude; it can be shaped by memorable ‘turning point’ moments. Small events can disproportionately shift how people feel, in ways that are partly personal and unpredictable.
- •Partner positivity predicts major outcomes (breakup/divorce, health)
- •People may retrieve a general attitude, but also weight key episodic moments
- •Turning points can be negative (feeling uncared for) or positive (support in a bad day)
- •How people integrate global impressions with moments remains poorly understood
- 1:03:05 – 1:06:10
Wrap-up: improving self-knowledge, future research, and where to follow Paul
Chris asks whether better insight into one’s own revealed preferences would improve relationship outcomes, and Paul notes the idea is promising but under-tested. They close with where to find Paul’s work and mention his upcoming book.
- •Hypothesis: smaller stated–revealed gaps might relate to better relationship success
- •Polyamory/multi-partner data could help reveal within-person preference patterns
- •Many valuable follow-ups require different data than typical relationship studies collect
- •Follow Paul on X/Twitter @PaulEastwick; book planned in ~1.5 years
