"Beauty" premium for social scientists but "unattractiveness" premium for natural scientists in the public speaking market

The study investigates whether, and how, facial attractiveness influences scientists’ success outside academia, particularly in the public speaking market where fees proxy audience willingness to pay. Motivated by growing expectations for scientists to engage the public and evidence that nonexperts may have difficulty evaluating scientific merit directly, the authors ask whether facial appearance—known to shape judgments via a beauty halo effect—affects external prominence and market value. They consider potential stereotypes distinguishing natural versus social scientists and hypothesize that attractiveness may differentially relate to external influence and speaking fees across fields. The purpose is to clarify whether facial attractiveness correlates with academic performance, external influence (media/public measures), and speaking fees, and whether these relationships vary between natural and social sciences, thereby informing understanding of how appearance shapes science communication and societal recognition.

Prior work documents a beauty premium and halo effects in social perception and labor markets, with attractive individuals often evaluated more positively and earning higher incomes. In academia, attractiveness relates to higher teaching evaluations and career outcomes, though experimental evidence suggests attractiveness boosts confidence rather than productivity. Public interest in attractive scientists may be higher, yet perceived research quality may be judged lower for more attractive scientists. Stereotypes about scientists—especially in STEM—portray them as socially awkward and physically unattractive, potentially shaping public expectations and assessments. Media representations often reinforce narrow images, which could differentially influence natural versus social scientists. The literature also debates the universality and measurement of attractiveness (averageness/symmetry vs. culturally specific canons) and the validity of citation metrics as performance indicators. This background motivates testing whether attractiveness correlates with academic output, external attention, and market value in public speaking, and whether field stereotypes moderate these effects.

Data: 734 public speakers (collected Dec 2013–Jan 2014) from eight North America–based speaker agencies. Composition: 217 full-time academics, 151 part-time academics, 366 nonacademics. Minimum speaking fee defined as the lower end of listed fee ranges; for speakers listed on multiple sites, the highest lower-end value was used. Facial attractiveness: Frontal portraits were found for 726 speakers via public sources. Anaface.com computed beauty scores (1–10) based on facial geometry (e.g., symmetry and ratios). For each image, three research assistants independently placed 17 facial landmarks (June 5–20) to generate scores; the average score was used. Prior work shows weak but significant correlation of Anaface with human ratings; the authors additionally cross-validated with geometry-based measures in the supplement. Academic performance (academics only): Publication and citation metrics from Google Scholar and Scopus: total publications, total citations, citations per paper, and h-index. External influence: Google search results (non-.edu domains) for the speaker’s full name (automated on April 14, 2014 via Google API; manual exclusion to avoid spurious matches; counts deemed invalid if ≥5 pages mismatched, N=154). Additional measures: number of TED talks (pre-2013), number of books (US Library of Congress), New York Times Best Seller counts and weeks, and major nonfiction book awards. Biographical controls: Field of expertise categorized via automated keyword searches into natural sciences (N=187), social sciences (business-related; N=358), and social sciences (others; N=189). Other covariates: gender; ethnicity (African, Asian; Caucasian reference); professional age (years since highest degree) and its square; QS field-specific university rank of highest degree (2004–2018; non-listed assigned rank 500); Nobel laureate; US-based; academic engagement (full-time vs part-time). Statistical analysis: OLS regressions with heteroscedasticity-robust standard errors. Dependent variables included academic performance metrics, external influence metrics, and log minimum speaking fee. Independent variable of interest: facial attractiveness; models include interactions between attractiveness and field (natural science reference; interactions for social science (business) and social science (others)) to assess between-field differences. Skewed variables (publications, citations, Google webpage counts, fees) were log-transformed as appropriate. Significance reported at four levels (0.1, 0.05, 0.01, 0.001). For multiple testing on external influence outcomes, Benjamini–Hochberg FDR=0.2 critical values were reported alongside p-values.

• Facial attractiveness and academic achievement (academics): No statistically significant association between attractiveness and publication/citation metrics after controls. Examples:
  • ln(Publications, Google Scholar): b=0.102, p=0.096; ln(Publications, Scopus): b=−0.032, p=0.589.
  • ln(Citations, Google Scholar): b=0.084, p=0.361; ln(Citations, Scopus): b=0.045, p=0.675.
  • Citations per paper (Google Scholar): b=−2.163, p=0.362; (Scopus): b=0.709, p=0.723.
  • h-index (Google Scholar): b=0.286, p=0.813.
  • No significant between-field differences in these relationships (interaction terms non-significant across outcomes).
• Facial attractiveness and external influence (academics): Positive associations where appearance is salient to the public:
  • TED invitations: b≈0.056–0.057, p≈0.059–0.072.
  • NYT Best Seller weeks: b≈0.996–1.026, p≈0.081–0.079.
  • ln(Google webpages, non-.edu): b=0.207, p=0.046; controlling for productivity b=0.156, p=0.055.
  • No significant effects for number of books or nonfiction book awards.
  • No statistically significant moderation by discipline for these external influence measures.
• Facial attractiveness and speaking fees: Strong field-dependent effects.
  • Academics:
    • Natural sciences: unattractiveness premium; one-unit decrease in attractiveness increases minimum speaking fee by about 19% (implied by b≈−0.189; p≈0.011).
    • Social sciences (business): beauty premium; interaction b=0.204, p=0.034.
    • Social sciences (others): beauty premium; interaction b=0.364, p<0.001.
  • Nonacademics: similar divergence by field background
    • Natural sciences: b=−0.147, p=0.059.
    • Social sciences (business): b=0.189, p=0.053.
    • Social sciences (others): b=0.262, p=0.056.
  • Combined sample (academics + nonacademics):
    • Natural sciences: b=−0.176, p=0.001.
    • Social sciences (business): b=0.201, p=0.004.
    • Social sciences (others): b=0.327, p<0.001.
• Interpretation: Attractiveness does not track academic quality (publications, citations), but correlates with visibility metrics where appearance is salient. In the speaking market, social scientists benefit from beauty, whereas natural scientists benefit from being less attractive (consistent with stereotypes and the "ugly Einstein" effect).

The findings indicate that facial appearance plays a meaningful role in how the public values scientists in the speaking market. Despite no link between attractiveness and scholarly productivity, attractiveness is positively associated with external prominence where faces are visible (TED talks, web mentions), suggesting that visual exposure amplifies attractiveness effects on attention. Crucially, the relationship between attractiveness and speaking fees diverges by field: social scientists receive a beauty premium, while natural scientists receive an unattractiveness premium, aligning with stereotypes that associate natural science expertise with less attractive appearances and, perhaps, perceived higher research quality in less attractive individuals. These patterns persist among nonacademic speakers with similar field backgrounds, suggesting a broader market mechanism rather than purely academic factors. The null link between attractiveness and citations/publications argues against productivity-based explanations for fee differences; instead, taste-based discrimination, stereotypes, or differences in perceived communication style may underlie the fee patterns. Overall, facial appearance appears to shape public valuation of scientific expertise differently across disciplines, affecting knowledge dissemination and science communication outcomes.

This study contributes evidence that facial attractiveness is unrelated to core academic performance but is modestly associated with public-facing visibility where appearance matters and, importantly, is differentially priced in the public speaking market across disciplines. Social scientists benefit from a beauty premium, while natural scientists experience an unattractiveness premium, consistent with prevailing stereotypes about scientific personas. These results highlight that facial appearance can meaningfully influence how scientists are perceived and compensated outside academia. Future research should: (1) validate algorithmic attractiveness measures against human ratings across relevant audiences; (2) examine cross-cultural variability in attractiveness standards; (3) investigate psychological and communication-related channels (e.g., sociability, social intelligence, oral communication skills) linking attractiveness to market outcomes; (4) explore disciplinary epistemic or dialectic differences that may shape audience expectations; and (5) test generalizability with representative samples of scientists across career stages and fields, using robust performance metrics beyond simple citation counts.

• Attractiveness measurement: Used Anaface (geometry-based) with prior evidence of only weak positive correlation with human ratings; the study lacked an independent in-sample validation due to budget constraints. Perceived attractiveness depends on evaluator samples; cross-cultural validity of neoclassical canons and golden ratio is problematic and Eurocentric.
• Cultural context: Sample relies heavily on Western scholars and North American speaker agencies, limiting cross-cultural generalizability.
• External influence measure validity: Google webpage counts required exclusions for name ambiguity; residual noise may remain.
• Performance metrics: Citation-based indicators (including h-index) have known limitations and potential for manipulation, though robustness checks with variants were conducted.
• Observational design: Correlational analyses cannot establish causality; unobserved traits (e.g., sociability, communication skills) may mediate relationships.
• Sample selection: Focus on speakers already active in the public speaking market may bias attractiveness distributions and limit inference to the broader population of scientists.