Cultural evolution of beauty standards

Data and code to reproduce the figures of

Louis Boucherie, Sagar Kumar, Katharina Ledebur, August Lohse and Karolina Śliwa. Cultural evolution of beauty standards. Proceedings of the National Academy of Sciences 123 (21), e2602380123 (2026). doi:10.1073/pnas.2602380123

Citation

@article{boucherie2026cultural,
  title   = {Cultural evolution of beauty standards},
  author  = {Boucherie, Louis and Kumar, Sagar and Ledebur, Katharina and Lohse, August and {\'S}liwa, Karolina},
  journal = {Proceedings of the National Academy of Sciences},
  volume  = {123},
  number  = {21},
  pages   = {e2602380123},
  year    = {2026},
  doi     = {10.1073/pnas.2602380123}
}

Repository

evolution-beautystd/
├── data/             plot-ready CSV / JSON inputs
├── scripts/          one Python script per figure
├── figures/          figure outputs (PDF; the scripts also write PNG)
└── requirements.txt  Python dependencies

The repository is self-contained: each script in scripts/ reads from data/ and writes to figures/.

pip install -r requirements.txt
cd scripts
python <script_name>.py

utils.py holds the shared plot style and is not run on its own.

Runway-show network

data/network_shows_edges.csv is the network of runway shows: which models walked for which brands, and in which year (model_id, brand, year). Models are anonymous IDs, not linked to the model IDs in the other files.

Brands ranked by eigenvector centrality in the model–brand network. Degree centrality is the share of all models who walked for the brand.

Rank	Brand	Models	Degree centrality	Eigenvector centrality
1	Dolce & Gabbana	2,477	0.185	0.119
2	Valentino	1,808	0.135	0.114
3	Fendi	1,403	0.105	0.102
4	Dior	1,436	0.107	0.100
5	Prada	1,558	0.116	0.098
6	Chanel	1,123	0.084	0.096
7	Louis Vuitton	1,602	0.120	0.091
8	Hermès	1,245	0.093	0.090
9	Kenzo	1,267	0.095	0.087
10	Givenchy	1,242	0.093	0.087

All brands are in figures/network_shows_brand_centrality.csv; scripts/network_shows.py rebuilds the figure and the table.

Figures

Script	Shows
age_distribution.py	Age distribution of records, overall and per event type
age_evolution.py	Mean age of records per event type, 2000–2024
body_tails.py	Tail quantiles (P10–P99) of female body measurements, 2000–2024
career_evolution.py	Records per year, by category and by gender
choropleth_models.py	Models and records by country of origin
entropy_evolution.py	Entropy of categorical and numerical attributes, female models
event_study_panel.py	Event study around the 2006 Milan charter and the 2017 Paris regulation
forest_plot_outliers_trends.py	Trends in IQR, skewness and kurtosis of body measurements
forest_plot_std_trends.py	Trends in the standard deviation of body measurements
heterogeneous_policy.py	Policy effects across hierarchy tiers and cities
hierarchy_table.py	Top-30 brands by network prestige (table)
intersectionality_panel.py	Race × plus-size analysis, 2010–2024
measurement_evolution_plots.py	Mean and standard deviation of body measurements per event type (female and male)
network_analysis_panel.py	Brand–model network: embeddings, tiers and prestige
network_shows.py	Runway-show network and brand centralities
outliers_comparison.py	Shape of measurement distributions, models vs NHANES, with PCA (female and male)
pca_temporal_plot.py	Yearly trajectory of female body composition in PCA space
quantile_policy_analysis.py	RFM quantiles, Milan and Paris vs control
rfm_by_age_group.py	RFM of models vs NHANES women, by age group
rfm_tiers_sensitivity.py	Sensitivity of top/bottom-10 % RFM rates to the number of tiers

External data

• NHANES (US CDC), public-use data: data/nhanes_female_17_40.csv, data/nhanes_rfm_{female,male}_filtered.csv, data/pca_nhanes_projection_{female,male}.csv. https://www.cdc.gov/nchs/nhanes/index.htm
• Natural Earth, public domain: the country shapefile in data/naturalearth_110m_countries/. https://www.naturalearthdata.com/