Attractiveness studies: what famous papers found
Attractiveness studies, correctly cited: what landmark papers found, what they measured, and the numbers later writers got wrong.
You are six blog posts deep, and the same statistic repeats until the trail ends at a blog post with no journal, abstract, or source.
The landmark papers matter; transmission is the problem. Here are correct citations, findings, and unsupported claims.
This is a citation audit, not a claim that research is weak. Much is replicated work. Four correctly cited findings beat forty embellished ones.
The short version
-
Fast judgement: Willis and Todorov (2006) found that trait judgements after a 100-millisecond face exposure correlated highly with judgements without a time constraint; extra time mainly increased confidence.
-
Agreement and treatment: Langlois et al. (2000) described eleven meta-analyses showing agreement within and across cultures, plus more positive judgement and treatment of attractive children and adults.
-
Brief behaviour: Ambady and Rosenthal (1992) found that expressive behaviour observed, often under 30 seconds, predicted interpersonal outcomes with meaningful accuracy.
-
Stated preferences: Buss (1989) found consistent cross-cultural patterns in stated partner preferences across 37 cultures, n ≈ 10,047; preferences were traits weighted by context.
-
Shape, halo and approach: Singh (1993), Thorndike (1920), Dion, Berscheid and Walster (1972), and Moore (1985) addressed waist-to-hip ratio, observer bias, assumptions about attractiveness, and nonverbal approach signals.
How fast a first impression forms — Willis & Todorov (2006)
Correct citation: Willis, J., & Todorov, A. (2006). First impressions: Making up your mind after a 100-ms exposure to a face. Psychological Science, 17(7), 592–598.
The paper found that trustworthiness, competence and attractiveness judgements after a 100-millisecond exposure correlated highly with judgements without a time constraint. Extra time increased confidence. The point is speed, not a countdown.
Commonly misquoted as: “You have 100 milliseconds to make a first impression” or “7 seconds.” It measured judgement from a static photo, not a deadline or seven-second window.
A still face is not a conversation, moving body, or interaction. A fast read is not final.
Caveat: this is evidence about rapid judgements from faces, not an irreversible impression. Our test reads perceived first-impression appeal, not your whole character.
Whether people agree on who is attractive — Langlois et al. (2000)
Correct citation: Langlois, J. H., Kalakanis, L., Rubenstein, A. J., Larson, A., Hallam, M., & Smoot, M. (2000). Maxims or myths of beauty? A meta-analytic and theoretical review. Psychological Bulletin, 126(3), 390–423.
The review described eleven meta-analyses. It found agreement about who is attractive within and across cultures, plus more positive judgement and treatment of attractive children and adults, who also exhibited more positive behaviours and traits.
Commonly misquoted as: “919 studies,” “919 effect sizes,” “over 1,800 articles,” “12,261 raters,” or “12,000 raters.” All of those are fabrications. The correct description is eleven meta-analyses.
The finding supports shared perception patterns, not universal attraction or moral truth. See the halo-effect explanation.
Caveat: agreement in ratings does not erase individual taste or context, or control every later interaction.
Whether brief observations predict anything — Ambady & Rosenthal (1992)
Correct citation: Ambady, N., & Rosenthal, R. (1992). Thin slices of expressive behavior as predictors of interpersonal consequences: A meta-analysis. Psychological Bulletin, 111(2), 256–274.
The paper found that very brief observations of expressive behaviour — “thin slices”, often under 30 seconds — predicted interpersonal outcomes with meaningful accuracy. Posture, expression and manner carry information early.
Commonly misquoted as: proof that “first impressions are always right.” A meta-analysis finding better-than-chance prediction is not a finding of accuracy in any individual case. A useful signal can still be misread.
Appearance advice that ignores behaviour is incomplete: expressive behaviour changes the initial read. Our first-impression statistics guide puts that boundary plainly.
Caveat: a thin slice is not a personality diagnosis. Treat it as provisional and leave room for contradiction.
What people say they want in a partner — Buss (1989)
Correct citation: Buss, D. M. (1989). Sex differences in human mate preferences: Evolutionary hypotheses tested in 37 cultures. Behavioral and Brain Sciences, 12(1), 1–49.
The study covered 37 cultures, n ≈ 10,047, and found consistent cross-cultural patterns in stated mate preferences: traits weighted by context.
Commonly misquoted as: evidence that a single trait — height, income or jawline — is decisive. The study measured what people say they prefer across many traits. It did not show that one trait determines any real outcome.
A preference list is not a vending-machine formula for desire, and stated preference is not behaviour. What women actually find attractive keeps the context idea intact.
Caveat: cross-cultural patterns can inform without predicting an individual woman’s choice. Do not use them to argue with someone about her preferences.
Body-shape ratios — Singh (1993)
Correct citation: Singh, D. (1993). Adaptive significance of female physical attractiveness: Role of waist-to-hip ratio. Journal of Personality and Social Psychology, 65(2), 293–307.
The paper found waist-to-hip ratio was a replicable predictor of attractiveness ratings. It concerns female figures, not permission to rank women or a complete theory of attraction.
Commonly misquoted as: a universal “ideal ratio” that applies to men. The paper is about ratings of female figures; the male shoulder-to-waist literature is separate work.
This narrow finding gets stretched online. A body-shape measure linked to ratings is not universal for every body, culture, setting or person. Is the golden ratio of the face real? applies the same caution.
Caveat: describing this study does not endorse ranking women. We are auditing transmission, not turning it into a target for anyone’s body.
Why one good trait colours everything — Thorndike (1920) and Dion, Berscheid & Walster (1972)
Correct citations: Thorndike, E. L. (1920). A constant error in psychological ratings. Journal of Applied Psychology, 4(1), 25–29. Dion, K., Berscheid, E., & Walster, E. (1972). What is beautiful is good. Journal of Personality and Social Psychology, 24(3), 285–290.
Thorndike found that officers rating soldiers gave correlated ratings across unrelated qualities: the original halo effect. Dion, Berscheid and Walster found attractive people were assumed to have desirable traits and life outcomes.
Commonly misquoted as: “Attractive people are more successful, kinder or smarter.” Both papers are about the judgements observers make, not about the people being judged. The bias is in the inference.
Attractiveness can influence treatment without proving attached traits. First impressions deserve attention, not worship.
Caveat: halo effects are real social forces, not destiny. Warmth, competence and conduct can update an early impression — and our product cannot measure all of them from photos.
Signals that start an interaction — Moore (1985)
Correct citation: Moore, M. M. (1985). Nonverbal courtship patterns in women: Context and consequences. Ethology and Sociobiology, 6(4), 237–247.
Moore’s catalogue described nonverbal signals preceding approaches being made and accepted in naturalistic settings. It records patterns, not a decoding system.
Commonly misquoted as: a checklist of “signs she likes you” that can be read with certainty, or used as a script. It was observational work. Any reading of it must stay consent-forward and must not become a technique for pressing an approach.
A signal is not consent. Approach respectfully, accept a clear or uncertain no, and do not touch or kiss anyone without explicit, ongoing consent. Body language of attraction is worth reading only with that limit in view.
Caveat: context changes meaning. No catalogue removes the need to communicate plainly and respect the other person’s choice.
How to check a statistic yourself
Start with the claimed author and year, not the repeated sentence. Search the author, year and paper title, then check the journal record and abstract. Ask whether the number appears and describes the sample, method, or result.
Be suspicious of any figure with no journal. A statistic first traceable to a blog post is unsourced until someone produces the original paper. Compare the claim with the subject: stated preferences cannot prove what determines behaviour, and a face-exposure study cannot prove a real-life conversation deadline.
Keep the original wording beside your summary. Citation errors begin when a narrow sentence becomes a universal one.
Caveat: abstracts are efficient checks, not substitutes for a full paper when the claim matters. These summaries are deliberately narrower than the internet versions.
Our own correction
This site carried an inflated Langlois figure. We removed it because a page auditing others’ citations while hiding its corrections is worthless.
The wrong numbers were “919 studies,” “919 effect sizes,” “12,261 raters,” and “over 1,800 articles.” All were false. The correct description is eleven meta-analyses. Research is not weak; transmission was broken. Four correctly cited findings serve readers better than forty embellished ones.
Caveat: correcting one page does not make every internet claim reliable. Check the original source whenever a number matters.
The bottom line
The famous attractiveness papers do not say a face is destiny, first impressions are always right, or one trait decides who gets chosen. They say: judgements form quickly, observers often agree, brief behaviour carries information, preferences come in bundles, and bias colours inference.
That is enough to take first impressions seriously without feeding looks anxiety. Appearance matters at the start, but it is not your entire social value and no photo-based tool can read your whole life. If you want a practical read, the free Real World Appeal test has no paywall on your score and shows the result before you decide whether to continue. Its limit is simple: it estimates perceived first-impression appeal from your images, not your worth or future.
Studies referenced
- Willis, J., & Todorov, A. (2006). First impressions: Making up your mind after a 100-ms exposure to a face. Psychological Science, 17(7), 592–598.
- Langlois, J. H., Kalakanis, L., Rubenstein, A. J., Larson, A., Hallam, M., & Smoot, M. (2000). Maxims or myths of beauty? A meta-analytic and theoretical review. Psychological Bulletin, 126(3), 390–423.
- Ambady, N., & Rosenthal, R. (1992). Thin slices of expressive behavior as predictors of interpersonal consequences: A meta-analysis. Psychological Bulletin, 111(2), 256–274.
- Buss, D. M. (1989). Sex differences in human mate preferences: Evolutionary hypotheses tested in 37 cultures. Behavioral and Brain Sciences, 12(1), 1–49.
- Singh, D. (1993). Adaptive significance of female physical attractiveness: Role of waist-to-hip ratio. Journal of Personality and Social Psychology, 65(2), 293–307.
- Thorndike, E. L. (1920). A constant error in psychological ratings. Journal of Applied Psychology, 4(1), 25–29.
- Dion, K., Berscheid, E., & Walster, E. (1972). What is beautiful is good. Journal of Personality and Social Psychology, 24(3), 285–290.
- Moore, M. M. (1985). Nonverbal courtship patterns in women: Context and consequences. Ethology and Sociobiology, 6(4), 237–247.
Frequently asked questions
What are the most important studies on attractiveness?
The key papers cover fast first impressions, agreement about attractiveness, brief behavioural observations, stated mate preferences, body-shape ratios, the halo effect, and nonverbal courtship signals. Read each citation carefully because the findings are narrower than the internet versions. The first-impression research guide is a useful next step.
Did research prove that first impressions take 100 milliseconds?
No. Willis and Todorov found that judgements from a 100-millisecond face exposure correlated highly with judgements made without a time limit. That is evidence that a judgement can form quickly, not a deadline saying you must make an impression within 100 milliseconds. See the first-impression statistics guide.
What did Langlois et al. actually find?
Langlois et al. reviewed eleven meta-analyses and reported agreement about who is attractive within and across cultures, along with more positive treatment of attractive children and adults. It did not establish the inflated study, rater, or article counts repeated online. The halo-effect explanation covers the judgement mechanism beside this finding.
Does the halo effect mean attractive people are better people?
No. Thorndike and Dion, Berscheid and Walster studied observers’ judgements. Their work shows that one favourable impression can colour assumptions about unrelated traits and life outcomes; it does not prove that attractive people are kinder, smarter, or more successful. Read the halo effect on attractiveness for the distinction.
How can I check an attractiveness statistic?
Search the author and year first, then inspect the original abstract and journal record. Ask what was measured, who was studied, and whether the claimed number appears in the paper. A statistic whose earliest traceable appearance is a blog post is unsourced. If you want a practical comparison, start with the free Real World Appeal test, which shows a result before you decide whether to continue.