Cultural evolution of human beauty standards

TL;DR

This study tackles how human beauty standards evolve in mass media over 25 years by assembling a large, longitudinal dataset of 793,199 model records across four media types. It employs a comprehensive statistical framework (including Mann–Kendall tests, Sen's slope, bootstrap CIs, PCA, and network analysis) to quantify changes in anthropometrics, appearance traits, and demographics, benchmarked against NHANES data and using Relative Fat Mass as a key body-fat metric. The findings reveal a paradox: mean model physiques remain largely stable while diversity increases in size, origin, and appearance, driven mainly by tail expansion via selective plus-size inclusion; however, this diversity is uneven and intersectionally concentrated among non-White identities. Policy experiments show numeric thresholds (e.g., Milan’s BMI floor) can reduce underweight appearances, whereas non-threshold approaches (like Paris’s medical certificate) have little detectable effect, highlighting the importance of explicit criteria in regulating representation. Overall, the study documents meaningful but uneven progress toward inclusivity, with persistent alignment of the industry’s thin ideal at the top and pronounced asymmetries across race and body-size categories that shape the trajectory of cultural beauty norms and policy implications.

Abstract

Beauty standards shape self-perception and health through social comparison and objectification, while exposure to idealized imagery exacerbates body-image concerns. Media and fashion are central arbiters of these ideals, yet long-term, quantitative, intersectional studies on how representation has changed remain scarce. We assembled a dataset of 793199 records spanning 25 years of advertising, magazine covers, runway shows, and editorials to quantify changes in anthropometric and demographic representation. We find a paradox in the evolution of beauty ideals: while representational diversity has increased, the median model physique remains stable. This is driven by selective plus-size inclusion at the upper tail, while the typical physique continues to diverge from the US population. Intersectionally, non-white models are 4.5 times more likely to be plus-size, indicating that progress in size inclusivity falls disproportionately on multiple underrepresented identities. Stratifying the industry via a data-driven prestige hierarchy, we find that thinness is overrepresented at the top tier. Finally, comparing two regulatory interventions we observe that numeric thresholds are more effective at reducing underweight appearances. Our results quantify the cultural evolution in media and fashion, revealing that inclusion has increased; however, gains are uneven and intersectionally concentrated on size and ethnicity, whereas the prevailing thin ideal remains largely unchanged.

Figures (32)

• Figure 1: Evolution of representational diversity for female models (2000--2024).(a) Mean height, bust, waist, hips, and RFM (see Methods: RFM) by work type (fashion shows, advertisements, magazine covers, and editorials). Anthropometric means remain stable across the period; only bust shows a modest decline. Shaded areas indicate 95% confidence intervals. (b) Standard deviations of the same measures increase across all work types, indicating growing variability in represented body sizes despite stable central tendencies. (c) Hair color distribution shifts away from lighter phenotypes: blonde declines while darker shades gain share. (d) Eye color distribution shows similar diversification, with blue eyes declining and brown increasing. (e) National origins by world region: Eastern European representation peaks in the early 2000s then declines, while contributions from Sub-Saharan Africa, East Asia, and South Asia increase. See Fig. \ref{['fig:entropy']} for entropy trends and Supplementary Sec. \ref{['si:worldregions']} for region definitions.
• Figure 2: Distribution shape and population benchmarking.(a) Interquartile ranges (IQR) for bust, waist, hips, and RFM remain stable over time, indicating that the central 50% of the distribution is unchanged. (b) Skewness shifts from near-zero to positive, indicating growing right-tail asymmetry (more large-size observations). (c) Excess kurtosis increases over time, reflecting heavier tails; estimates are noisy due to sensitivity to outliers. (d) RFM distributions for US-based female models and US women aged 16--29 (NHANES NHANES2021_2023) show minimal overlap; dashed lines indicate group means. (e) Annual gap $\Delta\mathrm{RFM} = \mathrm{RFM}_{\text{NHANES}} - \mathrm{RFM}_{\text{Models}}$ with fitted trend (${\sim}0.01$ percentage points per year), showing no convergence over two decades. (f) PCA projection separates models and population along RFM isolines, indicating selection primarily on body composition. (g) Component loadings: bust, waist and hips drive PC1; height dominates PC2. See Fig. \ref{['fig:pca']} and Supplementary Multidimensional Analysis.
• Figure 3: Prestige-tier dependence and policy impacts on model Relative Fat Mass (RFM).(a--b) Industry stratification by brand prestige. Brands are assigned to four tiers (Elite, High, Mid, Low) on the model--brand bipartite network (see Methods: Network Hierarchy). (a) The share of low-RFM models (bottom decile of the industry-wide distribution) declines monotonically from Elite to Low tiers, indicating that thinness selection intensifies at the top of the hierarchy. (b) The share of high-RFM models (top decile) follows a U-shaped pattern: over-represented at both Elite and Low tiers, with the smallest shares in the middle tiers---suggesting that mid-tier brands are less willing to cast plus-size models than either elite houses or lower-prestige outlets. (c--f) Policy impact analysis using difference-in-differences. (c) Mean RFM at Milan Fashion Week (green) versus control cities (grey), 2002--2010. Following the 2006 BMI floor ($\geq$18.5 kg/m$^2$), Milan exhibits a discrete upward shift of approximately 0.5 RFM units relative to the stable control trend. (d) Mean RFM at Paris Fashion Week (blue) versus control cities, 2013--2022. The 2017 French medical-certificate law produces no visible discontinuity; both Paris and control cities experience parallel increases coinciding with rising plus-size representation industry-wide (cf. Fig. \ref{['fig:intersectional']}). (e) The 25th percentile of RFM at Milan jumps by more than one unit after 2006, confirming that the policy effect is concentrated in the lower tail of the distribution. (f) The 25th percentile of RFM at Paris shows no differential change relative to control cities following the 2017 law. Shaded bands indicate 95% confidence intervals. Vertical dashed lines mark policy implementation.
• Figure 4: Intersectional minorities carry visible diversity(a) Evolution of the proportion of non-White models (Methods: Ethnicity Attribution). (b) Evolution of the proportion of plus-size model by ethnicity (White and non-White) (Methods: Plus-size Attribution). (c) Odds ratio of being a plus-size model for the two populations, It represent how more likely are non-White models to be plus-size compare to White models. (d) Difference between the proportion of plus-size model per ethnicity. Non-White models are almost always over-represented among the plus sized models (Methods Odds ratio and intersectionality analysis)
• Figure 5: Temporal evolution of diversity across body measurements and demographic characteristics Entropy measures quantify diversity for nine variables across four career event types: fashion shows, advertisements, magazine covers, and editorials (2000-2024). (a-c) Body size measurements (height, bust, waist) show stable differential entropy with modest temporal variation. (d-e) Hip and RFM entropy trends reveal consistent diversity patterns across all fashion contexts. (f) Hair color Shannon entropy demonstrates stable categorical diversity over time. (g) Eye color diversity shows slight increasing trends in recent years for some event types. (h) Geographic diversity (world region) exhibits distinct patterns between event types, with fashion shows maintaining higher entropy than other contexts. (i) Ethnicity diversity shows temporal increases across all event types.