Measuring Racial and Gender Representation in YouTube Instructional Videos using Natural Language Processing (NLP)

• In Event: Quantitative and Mixed Methods Research in Social Studies Education

Abstract

YouTube instructional videos are becoming increasingly common in classrooms, with several channels now widely used and even formally endorsed by educational institutions. Like any curriculum, these resources shape whose histories are taught and valued. While prior research on racial and gender representation has focused on textbooks, little attention has been given to the instructional videos teachers now use. This study addresses that gap by using natural language processing to examine the racial and gender representation of historical figures in US history and civics videos. Findings reveal a significant underrepresentation of figures racialized as Black and Brown—especially in content from PragerU Kids and Khan Academy. These results have significant implications as digital resources become increasingly integrated into formal education.

Authors

• Cody C. Pritchard, University of Tennessee

• Hadi Bhidya, University of Tennessee

• Joshua Rosenberg, University of Tennessee

• Gregory D. Peterson, University of Tennessee