Race- and gender-based under-representation of creative contributors: art, fashion, film, and music

The study examines how gender and race/ethnicity are represented among influential creative contributors in contemporary art, high fashion, Hollywood box office film, and popular music. Motivated by the societal benefits of art and the importance of diversity for access and audience representation, the authors quantify under-representation using socially inferred demographics. They compare contributors’ inferred identities to U.S. population benchmarks to highlight loci of exclusion and emphasize intersecting identities (gender by race/ethnicity). The work contributes by: comparing four domains side-by-side; jointly analyzing gender and race/ethnicity; benchmarking against U.S. demographics; and documenting a transparent, semi-automated approach suitable for repetition over time. The study is observational and uses a convenience sample to identify and quantify under-representation rather than to infer causal mechanisms.

Background across four domains shows persistent inequities. Contemporary art: Historical activism (e.g., Guerrilla Girls) and empirical studies reveal under-representation of women and artists of color in galleries, auctions, and museum collections and exhibitions. Fashion: Research documents harmful effects of narrow body ideals and persistent racial homogeneity among models; preliminary evidence suggests limited diversity among designers and decision-makers. Film: Studies (e.g., USC Annenberg, UCLA) show low representation of women and non-white characters and creators, with strong influence of behind-the-camera demographics on on-screen diversity; the industry has historically reinforced racial hierarchies. Music: Corporate-level inequalities in signing and pay disadvantage women; Annenberg reports show persistent gender gaps among artists, songwriters, producers, and award nominees, with varying representation of people of color among top-charting musicians.

The authors assembled a dataset of 4,921 records (4,895 individuals; some appear in multiple domains), then, after applying consensus thresholds and removing unresolved cases, analyzed 4,747 records: 2,092 contemporary artists; 866 fashion designers; 1,569 film principals; and 220 popular musicians. Data sources and scope: Contemporary art uses prior data of artists (born 1940+) from 18 major U.S. museum collections. Fashion covers designers from 2,608 shows across the Big Four fashion weeks (2018–2019 seasons/lines), deduplicated to 889 unique designers before inference. Film uses IMDB’s principal cast/crew (10 per film) for the top 100 grossing films of 2018 and 2019, deduplicated to 1,580 unique principals. Music uses Billboard year-end charts (Hot 100, Top 200 Albums, Top 100 Artists for 2018–2019), arriving at 221 unique primary artists. Crowdsourced demographic inference: For fashion, film, and music, MTurk workers (US-based, ≥1000 prior HITs, ≥99% approval) researched each individual’s gender and primary race/ethnicity using public sources, providing confidence ratings. At least four independent workers assessed each person. Confidence-weighted consensus required ≥6 points (out of 12) to accept a category; ties or insufficient consensus were set to missing (NA). Small remaining NA cases were assessed by team members walled off from analysis. Race/ethnicity response options followed broad ACS-aligned categories; Middle East/North Africa was ultimately grouped into “Not Listed” to align with ACS comparisons. Final categories: AIAN, Asian, Black, Latinx, NHPI, White, Not Listed; gender options: woman, man, AGNN (agender, gender nonconforming, nonbinary), or unable to determine. Records missing either gender or race/ethnicity after consensus (172; 3.5%) were dropped. Data reliability and validation: Multiple quality controls were used (worker qualifications, examples/instructions, multi-rater with confidence weighting). A 3% validation subsample (n=142) evaluated by an author (not involved in initial inference) fully agreed with the crowd consensus. Benchmarks: U.S. 2019 ACS five-year PUMS provided comparison values; sex used as proxy for gender (noting the limitation and erasure of gender minorities). Ethics: Public-figure, public-source, inferred demographics; names redacted in public data release to mitigate harm; unredacted data available to researchers on request.

Across domains, women and racial/ethnic minorities are underrepresented relative to U.S. population benchmarks (ACS). Gender: Women comprise 51% of the U.S. population but only 28% of contemporary artists, 45% of high fashion designers, 27% of box office film principals, and 17% of popular musicians. Race/ethnicity: Non-white groups make up 39% of the U.S. population but only 22% in art, 22% in fashion, and 19% in film; music is an exception at 48% (driven largely by Black artists). Intersections: White men dominate each domain—approximately 55% in art, 42% in fashion (with white women at 36%), 61% in film, and 42% in music (with Black men at 34%). White men are overrepresented versus the U.S. population by factors of about 1.4 (fashion, music), 1.8 (art), and 2.0 (film). Latinx representation is particularly low, notably about 2.4% in fashion despite an estimated 17% of the U.S. population. Sample sizes analyzed: art N=2,092; fashion N=866; film N=1,569; music N=220.

The findings quantitatively substantiate substantial under-representation of women and marginalized racial/ethnic groups among influential contributors in art, fashion, film, and music, addressing the study’s goal to identify loci of exclusion. While Black men are highly represented in music relative to population share, over-representation does not imply equity or freedom from marginalization. Across domains, dominance by white men indicates persistent structural dynamics. The authors emphasize that over-/under-representation is distinct from majoritization/minoritization, and most marginalized groups remain numerically minoritized within domains. These results can inform targeted efforts by institutions and stakeholders to address disparities and monitor change over time.

Women are underrepresented across all four creative domains relative to U.S. benchmarks, and aggregated marginalized racial/ethnic groups are underrepresented by roughly a factor of two in art, fashion, and film. Intersecting identities reveal white men’s overrepresentation across domains (factors ~1.4–2.0). The study offers a transparent, repeatable approach to track diversity over time but is constrained to socially inferred identities and convenience-sampled proxies of influence. Future work should: gather and use self-identified demographic data to reduce misclassification and erasure; expand to additional identity axes (e.g., nonbinary, transgender, LGBTQ+, disability, multiracial) and domains; examine mechanisms driving disparities and interventions’ effectiveness; and consider alternative data sources and international comparisons.

Key limitations include: reliance on socially inferred demographics from public sources rather than self-identified data; ACS benchmark uses binary sex, erasing many gender identities and limiting comparability; broad race/ethnicity categories and lack of multiracial classification reduce fidelity; convenience sampling and domain-specific proxies (museum collections, Big Four fashion weeks, IMDB principals, Billboard charts) may not fully represent each field; comparison to U.S. population is imperfect given international contributors; small portion of records (3.5%) with unresolved demographics were dropped; the approach does not study mechanisms of under-representation; and ethical constraints necessitated redaction of names in public data.