What We Teach About Race and Gender: Representation in Images and Text of Children's Books.

This article focuses on the representation of skin color, race, gender, and age in influential children's books in the United States over the past century, using novel computational methods from computer vision and natural language processing to systematically analyze both images and text. It compares two main collections of award-winning books: the "Mainstream" collection (Newbery and Caldecott awards) and the "Diversity" collection (awards highlighting marginalized identities), finding persistent underrepresentation of Black and Latinx characters relative to their population shares, greater symbolic than substantive inclusion of females, and a tendency for children to be depicted with lighter skin than adults. The study also examines economic factors influencing these patterns, showing that consumers tend to purchase books reflecting their own identities and political beliefs, while books centering nondominant identities are priced higher and less available in predominantly White communities. These findings suggest that both supply- and demand-side market forces contribute to the persistent overrepresentation of historically dominant identities in children's literature, which may influence the intergenerational transmission of social beliefs.