A supervised learning approach to detect gender stereotype in online educational technologies

• Ano de defesa: 2019
• Autor(a) principal: Silva, Josmário de Albuquerque
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Alagoas; Brasil; Programa de Pós-Graduação em Modelagem Computacional de Conhecimento; UFAL
• Palavras-chave em Português: Tecnologia educacional; Identidade de gênero na educação; Estereótipos (Psicologia); Educational technology; Gender identity in education; Stereotypes (Psychology); CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: http://www.repositorio.ufal.br/handle/riufal/5878

Educational Technology (Edtech) has impacted the way humans learn and teach, e.g., improving students’ engagement, bolstering collaboration, increasing learning retention, and assisting teachers in creating and delivering new contents. However, researchers have highlighted that issues related to gender equality like gender stereotypes need to be addressed in order to promote plural and inclusive learning settings. In fact, recent findings show stereotypes have impacted several aspects in the learning process, e.g., performance, engagement, confidence, self-image, and anxiety. However, to address those issues, we require in advance to find out whether a given educational technology is stereotyped. Given that scenario, we propose an approach based on supervised learning classifiers to detect gender stereotypes in online educational environments. The method consisted of gathering situational cues of stereotype threat, i.e., textual contents and color schemes from web pages to develop and validate a machine learning predictive model. In addition, in order to validate the problem and gather more information about the impact of gender stereotypes in such settings, we primarily performed a systematic review to highlight evidence in the field and summarized the findings among different types of educational technologies and their implications in the last 20 years. The review also shows methodological approaches used along with those years and the limitations of such studies. Regarding predictive models, our approach showed a high precision on predicting gender stereotype threat in online settings. We also implemented the approach and applied it towards top-ranked universities’ web pages and the results suggest the presence of male bias in such settings.We discuss those findings and present a research agenda to underline research points that should be concerned when investigating gender stereotypes and educational technologies.