Facial Width-to-Height Ratio Does Not Predict Self-Reported Behavioral Tendencies

Michał Kosiński

Introduction

For over a decade, evolutionary psychology has explored the hypothesis that certain physical traits serve as "honest signals" of an individual's personality and behavioral tendencies. A prominent focus of this research has been the facial width-to-height ratio (fWHR)—calculated as the distance between the cheekbones divided by the distance between the upper lip and mid-brow. Previous high-profile studies, often involving small sample sizes ($N < 50$), suggested that men with higher fWHR (wider faces) are more aggressive, prone to cheating, and likely to exhibit racial prejudice. This paper presents a massive-scale "reality check" on these claims, analyzing over 137,000 participants to determine if fWHR truly predicts personality, intelligence, or behavioral tendencies.

Methods

The research employed a two-stage approach to transition from manual validation to automated large-scale analysis.

Study 1: Validation

To ensure the reliability of automated measurement tools, two research assistants manually measured fWHR for 1,692 participants. These manual measurements were compared against those generated by Face++, a computer vision API. A high correlation ($r = .86$) between the manual and automated methods validated the use of AI for the subsequent large-scale dataset.

Study 2: Large-Scale Analysis

The validated automated measurement was applied to a massive sample of Facebook users from the myPersonality.org dataset. This repository includes psychometric scores and profile data donated by users for research purposes. The final sample for Study 2 consisted of 170,241 images from 137,163 participants.

Facial landmarking and head pose variations used for automated fWHR measurement.

Psychometric Mapping and Controls

Participants' facial data were linked to 55 well-established psychometric scales, including the "Big Five" personality traits (Openness, Conscientiousness, Extraversion, Agreeableness, Neuroticism), the Barratt Impulsiveness Scale, and Raven’s Progressive Matrices for intelligence. The analysis controlled for age and gender, factors known to correlate with both facial structure and personality.

Key Insights

The study's findings significantly challenge the existing fWHR literature:

  • Negligible Effect Sizes: While many correlations were statistically significant due to the enormous sample size, the actual effect sizes were negligible. For instance, fWHR explained only 0.5% of the variance in men’s personality traits.
  • Contradicting Gender Narratives: Contrary to the theory that fWHR is a male-specific marker of aggression, the study found that the (already weak) correlations were actually slightly stronger in women.
  • Directional Reversals: Surprisingly, where correlations did exist, they often contradicted the fWHR hypothesis. Broader-faced individuals reported being more agreeable and prosocial, and showed less interest in "Violent-Occult" topics—the opposite of predicted aggressive tendencies.
  • The "File-Drawer" Effect: The study highlights a systemic issue in previous research: of 10 landmark fWHR studies, half had samples smaller than 25. This suggests that earlier findings may have been artifacts of low-powered experiments and publication bias.

Results

The Five-Factor Model domains explained a tiny fraction of the variance in fWHR ($R^2 = .010$ for females; $R^2 = .005$ for males). No significant correlations were found between fWHR and intelligence, self-monitoring, or a sense of fairness. Even after adjusting for the limited reliability of self-reports (correction for attenuation), the strongest correlation—women’s conscientiousness—remained a meager $.101$.

Conclusion and Limitations

While this study relies on self-reported behavioral tendencies rather than direct lab observations, the sheer scale of the data suggests that fWHR is not a meaningful predictor of internal traits. Future research might investigate whether fWHR influences social perception (how others judge a person) rather than reflecting actual character. However, the current evidence indicates that judging character by facial width is scientifically unsupported.

Summary updated

Abstract

A growing number of studies have linked facial width-to-height ratio (fWHR) with various antisocial or violent behavioral tendencies. However, those studies have predominantly been laboratory based and low powered. This work reexamined the links between fWHR and behavioral tendencies in a large sample of 137,163 participants. Behavioral tendencies were measured using 55 well-established psychometric scales, including self-report scales measuring intelligence, domains and facets of the five-factor model of personality, impulsiveness, sense of fairness, sensational interests, self-monitoring, impression management, and satisfaction with life. The findings revealed that fWHR is not substantially linked with any of these self-reported measures of behavioral tendencies, calling into question whether the links between fWHR and behavior generalize beyond the small samples and specific experimental settings that have been used in past fWHR research.
BibTeX
@misc{kosinski2017facial,
  author = {Michał Kosiński},
  title = {{Facial Width-to-Height Ratio Does Not Predict Self-Reported Behavioral Tendencies}},
  year = {2017},
  eprint = {2410.09979},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2410.09979},
}
Figure