A new Rutgers University study analyzing the differences in images used for male and female biographies on Wikipedia shows that significant gender bias exists in the visual aspects of the biographies and provides guidelines for future management of these content asymmetries.
The paper, “Visual Gender Biases in Wikipedia: A Systematic Evaluation across the Ten Most Spoken Languages,” co-authored by Associate Professor of Library and Information Science Vivek Singh at the Rutgers School of Communication and Information, will appear in the Proceedings of the ICWSM (16th International Conference on Web and Social Media), and the pre-print is now published in SocArXiv.
Singh and his coauthors analyzed all 6.2M biographies available on Wikipedia across 300 occupations in the ten most widely spoken languages in the world (English, Mandarin Chinese, Hindi, Spanish, French, Standard Arabic, Bengali, Russian, Portuguese, and Indonesian), and undertook quantitative and qualitative analysis of gender differences in the written and visual content. Of the biographies analyzed, they found less than 20% portraying women.
Continued exposure to such biased content could foster and fortify gender prejudices and stereotypes globally and widen the gender gap in Wikipedia audiences, the authors wrote. “Wikipedia, a collaborative crowd-sourced program, is the largest, most visited online encyclopedia,” Singh said. “Since it spreads information freely in more than 300 languages, many users, tools, and dashboards rely on its content. Hence, there is a need to maintain its fairness and completeness. Previous research had suggested differences in the textual content for male and female biographies. Much less was known about the visual content.”
The study’s cross-lingual results indicate, Singh said, that much of the male bias in content arises when editors select which personalities should have a Wikipedia page (in none of the ten languages analyzed do female biographies represent more than 23% of the record); the trends in written and visual content are quite dissimilar; biographies of men tend to have more images across languages; and biographies of women average better visual quality.
Singh co-authored this article with Pablo Beytia (Humboldt University of Berlin), Pushkal Agarwal (King’s College London), and Miriam Redi (King’s College London and Wikimedia Foundation).
The authors wrote, “Despite the crucial role of images for gender roles and stereotype diffusion, they have been mostly overlooked in research on gender asymmetries on Wikipedia. The few existing studies analyzing visual disparities across genders are relatively small scale in terms of quantity of biographies analyzed (1,000 articles at most), languages considered (English or German only), and range of metrics. This study aims to overcome these limitations to advance the development of an overall, systematic perspective on visual gender biases in Wikipedia. For the first time, we have conducted a macro cross-lingual exploration of gender gaps in Wikipedia biographies from a multimodal perspective.”
Acknowledging that Wikidata enables researchers to identify people with 36 possible genders, the authors said they decided to focus on comparing males and females. Unfortunately, the authors said, “only 1,223 biographies (0.01%) do not belong to those categories, which are further separated into multiple identities (e.g., intersex, transgender male, transgender female, agender, hermaphroditism, etc.). This high dissection did not allow us to perform, for those genders, the granular analysis we intended while observing trends in different languages and types of occupations. We consider this to be a limitation of the current work.”
Singh said their findings “add to an understanding of the phenomena and are a call to action to ensure fairer representation on Wikipedia. It is important for individuals to see women role models and a fair gender representation in different professions. A fairer representation can allow more girls to aspire for different professions and become noteworthy leaders in their fields.”
Discover more about the Library and Information Science Department at the Rutgers School of Communication and Information on the website.