BCSER: Using Computational Language Processing Techniques to Determine if Parental STEM Language Varies By Child Gender

This STEM education research capacity building project combines a research project that investigates whether parents transmit gender stereotypes about STEM to pre-school children in everyday conversations and a detailed professional development plan that enables the principal investigator to develop competencies in computational research methodologies, skill in mentoring junior colleagues, and knowledge of research on the underrepresentation of women in STEM. The research will examine whether parents use more STEM words and topics with boys than girls, providing evidence that parental language may contribute to girls’ underrepresentation in STEM.

The project also will develop and test a new research method by using a novel quantitative methodology, computational text analysis. The research project and professional development plan will support a research trajectory aimed at using computational text analysis to examine how variations in parents’ speech impact children’s downstream outcomes, such as scientific thinking and STEM participation using longitudinal research designs.

The project will position the investigator to explore potential malleable factors associated with gender differences in STEM outcomes, with the goal of proposing targeted interventions to broaden representation in STEM for all learners.

The project will be guided by the sociocultural theory of learning and development, gender schema theory to contextualize why parents may communicate about STEM differently with boys versus girls, and expectancy theory to explain how gender differences in parents’ beliefs and behaviors may impact children’s future STEM motivation and persistence. The investigator will examine three research questions: (1) Are there differences in parents’ use of STEM language based on whether the parent is interacting with their son or daughter? (2) What family characteristics moderate associations between parents’ STEM language and child gender? and (3) How reliable is computational text analysis compared to manual coding approaches for identifying parental STEM language?

Videotapes and survey data of demographic constructs, science attitudes, and gender schemas will be collected for 80 families. Transcripts from both sources will be compiled into one corpus for analysis. Computational text analysis will be used to determine whether parents’ use of STEM-related words and topics differ when speaking to their sons versus daughters.

The investigator will use the open-source R statistical modeling language to perform the entire text analysis workflow. The research findings will advance core knowledge about the processes underlying how STEM stereotypes are transmitted from parents to young children.

This project is supported by the ECR: Building Capacity in STEM Education Research competition of the EHR Core Research (ECR) Program. ECR funds fundamental research focused on STEM learning and learning environments, building capacity in STEM fields, and STEM workforce development.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.