Representation data

Dataset from professional pages on https://models.com  and https://www.fashionmodeldirectory.com. After de-duplication at the name-agency-URL level, the corpus comprises 15 183 unique profiles. Available attributes include anthropometrics (height, bust, waist, hips, dress size, shoe size), identity fields (name, nationality; when available, country of origin, age, weight, gender, ethnicity), Instagram handle, and profile photographs. All anthropometrics were standardized to SI units. Using the same sources, we assembled a work-level dataset that links models to credited jobs across format categories: Advertisements (92 283 records), Magazine Covers (86 382), Fashion Runways (413 265) and Editorials (146 503). Each record includes the model identifier, job type, and at least one associated image. These images are used for downstream annotations and, when profile fields are missing, to extend gender and ethnicity labels via machine learning algorithms (see Supplementary Gender and Ethnicity Attribution). The Instagram follower count were downloaded from https://www.instagram.com from the model Instagram handle available on the professional pages website.