Gendered socialization: The impact of an interface agent on middle-grade females’ affect in learning introductory algebra

I examine how a pedagogical agent (an animated human-like character) could create social context in a computer-based algebra-learning environment to help address equity issues in mathematics education. The study was framed in the perspectives of equity in mathematics education and the social and cultural context of mathematics learning for diverse groups of students in gender and ethnicity. First, classroom-based experiments were conducted with 120 9th-graders enrolled in required introductory algebra (53% males and 47% females; 51% Caucasian and 49% Hispanic). The results revealed that student gender and ethnicity were significant determining factors for the impact of the agent on learner affect: a 2-way ANOVA indicated that females evaluated their agent as affable significantly more than did males (p \u3c .005, η2 = .07). Hispanic students evaluated their agent as affable significantly more than did Caucasian students, (p \u3c .001, η2 = .17). For mathematics attitudes, a 2-way MANCOVA revealed a significant main effect of student gender on mathematics attitudes: females showed significantly more positive attitudes toward learning mathematics in the agent-based environment than did males (p \u3c .05, η2 = .04). For mathematics self-efficacy, a 2-way MANCOVA resulted in no significant effect. To identify any increase from pre to posttest self-efficacy, a 2-way repeated ANOVA was further conducted, which resulted in a significant main effect of gender: females significantly increased their math self-efficacy from pretest to posttest whereas males did not show this increase (p \u3c .01, η2 = .06). For mathematics learning, a 2-way repeated ANOVA revealed no main or interaction effect of student gender and student ethnicity; rather, the participants overall significantly increased their algebra learning from pre to posttest (p \u3c .001, η2 = .45). To better understand the results, in-depth interviews were conducted with 23 students. The interviews revealed, first, that females clearly developed a more human-like relationship with their agent throughout the time, compared to Caucasian males. Second, all females expressed a highly positive view about the quality and relevance of explanations provided by their agent, which presumably made them more receptive to lengthy explanations. Males in general valued their agent only for receiving quick feedback, but tended to skip the more lengthy explanations. Third, students’ views of and attitudes toward their agent as well as the quality of their learning experience were largely shaped by their everyday classroom experience. Females, especially Hispanic females, explicitly reported positive learning experiences after working with an agent. The findings imply that students’ learning experience in real and virtual spaces are closely interconnected