Developing students' STEM identities and career awareness within a science discipline can be challenging for teachers as STEM identity extends beyond specific content expertise and traditional pedagogical skills. This study investigated how teachers' STEM identities along with their shared understanding of the goals for including STEM identity activities in a geoscience curriculum unit contributed to the design of classroom activities intended to develop their students' STEM identities. Data for this study includes video recordings from a teacher professional development workshop in which six middle school teachers from a large district in central California explored their own STEM identities, survey data on students' STEM identities, and artifacts from co-designed activities tailored to teachers' classes. Results reveal that teachers' STEM identities significantly influence how they support the development of students' STEM identities. In addition, providing teachers with the opportunity to describe their own STEM identity led to meaningful focus for the co-design process. Finally, while the focus of this workshop was a geoscience unit, this study argues that the co-design process as applied to STEM identity activities can apply to other STEM topics.
Educators are being encouraged to teach with "big" datasets that have more cases and attributes than are typically used in the classroom. When introduced carefully, these types of datasets can allow students to engage in complex and self-directed reasoning, develop data management and inquiry skills, and experience data analysis in a way that is more authentic to professional practice. However, "big data" is also often unwieldy. It can overwhelm students, overload their software tools, and interfere with planful analysis. How can we make such datasets manageable? This paper presents pedagogical strategies and technical methods for educators, educational designers, and young investigators to reduce the number of cases or attributes in a large dataset in ways that do not unduly compromise their analyses. We illustrate these strategies with interactive demonstrations in two freely-available open source tools: the Choosy plugin in CODAP, and a Python Jupyter notebook.
Responding to the need for interdisciplinary, real-world STEM-integrated learning, we examine the relationship between teachers’ adaptive expertise in bioinformatics (a STEM field that integrates biology, data literacy, and computational approaches) and students’ classroom experiences. We draw on data from an earlier study, in which we conducted a summer professional development workshop and subsequent school-year implementation of a bioinformatics curriculum with five biology teachers. This study uses a mixed methods approach combining student pre/post suerveys, factor and regression analyses, and previously collected teacher adaptive expertise data to investigate how teachers’ abilities to adapt the STEM-integrated curriculum influenced their students’ classroom experiences. Using an adaptive expertise lens, we reveal challenges teachers faced in adapting STEM-integrated curricula in their practice due to a need for greater competence and confidence in content knowledge. Students’ classroom experiences showed significant growth in three of the five factors related to learning about bioinformatics with small to medium effect sizes. Notably, the two factors related to interests in working with real-world data and opportunities to develop literacy skills did not show significant growth. A regression analysis demonstrated statistically significant effects of teachers’ adaptive expertise on the student experience outcomes measured. We suggest that high-quality professional development should support teachers to become adaptive experts in STEM-integrated curricula, with a particular focus on improving teachers’ content knowledge. Furthermore, we contend that more research that can highlight specific knowledge needs of teachers (e.g., more adaptive expertise) is necessary if we are to fully realize the enormous potential for STEM-integrated learning in K–12 classrooms.
IntroductionArtificial intelligence (AI) and machine learning (ML) increasingly shape decision-making in high-impact domains such as healthcare, yet most secondary students lack opportunities to understand how these systems function or reproduce inequities. Advancing AI literacy in K–12 education requires context-rich curricula paired with valid assessments that measure students’ understanding of foundational ML concepts. This study developed and provided initial validation evidence for an assessment of foundational ML understanding while examining student learning within a context-based healthcare AI curriculum.MethodsWe conducted a multi-year study in which high school students engaged in an 18-lesson curriculum integrating data science, AI, and ML through authentic healthcare applications. Students analyzed real-world datasets, built and evaluated ML models, and participated in a collaborative datathon with healthcare and data science professionals. A curriculum-aligned assessment measuring foundational ML concepts was administered before and after instruction, and psychometric analyses, as well as multiple linear and multilevel regression models were used to evaluate assessment quality and changes in student learning.ResultsStudents entered the curriculum with limited but non-random knowledge of ML concepts and demonstrated statistically significant gains following instruction. The assessment also demonstrated improved reliability and item discrimination following instruction, suggesting that students developed more coherent conceptual knowledge of ML. At the same time, the findings identified several issues with the assessment and some of its items that warrant further refinement.ConclusionThis study provides initial validation evidence for one of the few assessments of foundational ML concepts developed for secondary students and provides evidence that a context-based healthcare curriculum can support foundational ML learning. The findings also highlight the importance of addressing both authentic contexts and continued assessment development to support the technical and societal dimensions of improving AI literacy.
Learning about natural hazards and risks through science practices entails considerations of uncertainty. We examined ways in which students expressed their epistemic (un)certainty about claims they made based on their inscription-based science practices. We analysed the level and reasoning of epistemic (un)certainty articulated by students in grades 6-13 (n = 266) as they interpreted cone of uncertainty maps to forecast hurricane risks, used a computer simulation to understand how pressure systems influence a hurricane's path, and analysed graphs to identify the relationship between ocean surface temperature and hurricane frequency. We delineated students' epistemic uncertainty in four dimensions of science practices: conceptual, material, epistemological, and social. Results showed that (1) students exhibited the highest levels of certainty when analysing graphs and the lowest when using a simulation, (2) their reasoning predominantly reflected conceptual and epistemological dimensions, (3) the social dimension was most prevalent when considering hurricane risks for communities and practices of the science community, and (4) significant positive relationships existed among students' claims, their levels of certainty, and the extent of epistemic engagement in their reasoning. We discuss theoretical, research, and pedagogical implications of these findings.
