GSE/RES Development of Gender-Based Science Performance Models

University Of California-Los Angeles, Los Angeles CA

Investigators

Abstract

A project team led by University of California Los Angeles is developing models documenting how males and females from middle school to the university develop and stabilize strategies during scientific problem solving. They are modeling: 1) The gender-related development of learning strategies in multiple sub-domains of chemistry, 2) How knowledge of solving problems in chemistry is differentially retained over time, and how learning re-occurs when needed, and, 3) How online collaborative activities / environments can be organized to maximize the strengths of the different problem-solving characteristics of female and male students. Intellectual Merit While the literature provides evidence for behavioral/cognitive gender differences in various STEM activities, there is less information on how these differences contribute to strategy development and use in scientific problem solving. The project will use artificial neural network and Hidden Markov Modeling of student performance and progress on complex science problem solving tasks to develop the gender models. There are commitments from two school districts and two universities to involve thousands of students across grade levels spanning middle school through university and in quasi-experimental groups with all female, all male and mixed gender groupings. Supporting data will include pre/post tests of content knowledge, science attitudes and teacher and student technology use, as well as overall academic performance including state standardized test data. The project team includes researchers, educators and students in California (UCLA, Placentia-Yorba Linda Unified School District), South Carolina (Clemson), Kentucky, and Italy (IRST), a psycho-metrician with experience in gender research and expertise in item response theory (ETS), and advisory members from the gender research community. Findings will be disseminated to teachers, the gender research community, basic science educators, and members of the intelligent tutoring and collaborative learning communities. The problem solving tasks will be available online to other teachers and researchers world wide along with the resulting neural network and Markov progress and performance models. Broader Impacts The long-term goal is to use this information to help institutional systems better understand how gender differences are likely to influence performance and participation of males and females in complex STEM learning/testing environments and to provide teacher guidelines on how to maximize learning for all students. The features of the models that are developed will potentially inform instructional practices from grades 6-16. While the study focuses on chemistry content, the modeling approach is applicable to other science domains.