In recent years, the importance of self-regulated learning (SRL) has become increasingly prominent in STEM education, encompassing science, technology, engineering, and mathematics. Self-regulated learning is regarded as a key competency for students to engage in autonomous learning and achieve sustainable development, particularly in complex and dynamic learning environments. However, there remains a lack of systematic and comprehensive research on how to foster self-regulated learning in STEM education. Following the PRISMA guidelines, this study conducts a systematic literature review to explore self-regulated learning in STEM education and learning contexts, rigorously selecting and analyzing relevant studies published between 2016 and 2025, and ultimately including 23 studies in the review. This review focuses on the primary pathways through which self-regulated learning is promoted in STEM education, including the technological tools that support and scaffold SRL, the instructional approaches to foster SRL, and the cognitive, affective, and behavioral outcomes associated with SRL and SRL-related interventions. As the study results show, a variety of modern technological tools support the development of self-regulated learning in STEM, including AI-powered learning systems, digital learning platforms, computer simulation technology, etc. Furthermore, the instructional approaches, such as innovative modules, game-based instructional approaches, and instructional methods, were found to contribute to the development of SRL. In addition, the review reveals that SRL and SRL-related interventions were associated with a range of cognitive, affective, and behavioral outcomes, with particularly notable associations reported for academic achievement and strategy use.
Creative thinking has emerged as a critical competency in twenty-first century engineering practice, yet pedagogical strategies that systematically cultivate this capacity within STEM-oriented curricula remain underexplored. This study investigates how the purposeful integration of STEM technologies—encompassing digital fabrication tools, simulation environments, collaborative coding platforms, and project-based learning frameworks—influences the development of creative thinking skills among undergraduate engineering students. Employing a convergent mixed-methods design, the study collected quantitative data through the Torrance Tests of Creative Thinking (TTCT) administered to 124 engineering students across two cohorts at a Central Asian technical university, alongside qualitative data derived from semi-structured interviews, classroom observations, and artefact analysis conducted over one academic semester. Findings indicate statistically significant gains in creative fluency (p < .001, d = 0.74), originality (p < .001, d = 0.61), and elaboration (p < .001, d = 0.58) among students in the technology-integrated cohort relative to the comparison group. Qualitative analysis identified three enabling mechanisms: structured ambiguity in design briefs, iterative maker-cycle feedback, and cross-disciplinary technological exposure. The study contributes a theoretically grounded instructional model—the STEM Creative Cycle (SCC)—and offers practical implications for curriculum designers, faculty developers, and educational policy-makers aiming to embed creativity in engineering education.
The growing demand for STEM professionals highlights the importance of effective STEM education. However, a challenge remains in how STEM education is implemented in practice, as translating STEM principles into meaningful classroom activities continues to be difficult. Integrated STEM approaches address these challenges by linking learning to authentic, interdisciplinary contexts. In this context, the STEMOOV model was developed to operationalise a STEM vision, offering a practical tool to improve STEM education and develop essential skills. This study employed a three-round Delphi method to achieve expert consensus on the STEMOOV model. A panel of academics, educators, and researchers from various countries and education levels evaluated the model, iteratively revising their opinions on its domains. The study examined which STEM skills are fostered by the model, how it supports students and teachers, its strengths and weaknesses, and its overall relevance for STEM education. The experts agreed that the STEMOOV model effectively fosters skills like collaboration, problem-solving, and critical thinking. Its adaptability to diverse contexts was seen as a strength, though flexible curricula, resources, and teacher training are critical for successful implementation. Overall, the STEMOOV model is a promising tool for integrated STEM education, with potential to engage students and support teachers.
This two-class quasi-experimental pilot documented class-level score patterns before and immediately after a four-week recycling and upcycling STEM-PBL unit in three self-reported components of environmental literacy among fourth-grade students in an Arabic-speaking school in northern Israel. Two intact classes (N = 62), with one class per condition, received either STEM-PBL instruction (n = 32) or teacher-led comparison instruction (n = 30). The intervention, conducted during regular science lessons, engaged small groups in relevant design challenges integrating measurement, materials, tool use, and engineering constraints. STEM was treated as an instructional design approach rather than an achievement outcome, and implementation fidelity was not independently assessed. Outcomes were measured using an age-appropriate three-point questionnaire assessing positive environmental behavioral tendencies, awareness and perceived knowledge, and negative environmental behavioral tendencies. Because the same students completed pre- and post-unit questionnaires, but anonymous responses could not be linked, observations were repeated but unpaired; therefore, inferential pre-posttests and standardized effect-size estimates were not used. Descriptively, the experimental class showed more favorable post-unit means and larger raw mean differences than the comparison class across all three outcomes. These patterns do not establish individual change, causal effectiveness, or improvement in environmental literacy as a whole. Interpretation is further limited by one class per condition, immediate self-report measurement, marked ceiling and floor effects, possible content overlap and demand characteristics, and lack of fidelity evidence. The unit should therefore be viewed as a documented instructional example warranting replication across classrooms with delayed assessments, objective knowledge measures, behavioral indicators, and implementation-fidelity measures.
This study examined the relationship between Computational Thinking Skills (CTS) and Higher-Order Thinking Skills (HOTS) among high school students majoring in biology in Jakarta, Indonesia, using a quantitative, correlational, and cross-sectional research design. Data were collected from 72 students (36 male, 36 female) selected from a population of 144 students through cluster random sampling across all classes, with a stratified procedure within the sampled clusters to achieve a balanced gender composition of 36:36 as an intentional design feature. The CTS test, adapted from the Bebras Computational Thinking Challenge (Blokhuis, 2016), consists of 18 items covering four components abstraction, algorithm design, decomposition, and evaluation while the HOTS test, based on the revised Bloom’s taxonomy (Anderson & Krathwohl, 2016), consists of 39 multiple-choice questions targeting analysis (C4), evaluation (C5), and creation (C6). Internal consistency in this sample was Cronbach’s α = 0.766 for the CTS and α = 0.540 for the HOTS; the lower HOTS coefficient is discussed as a measurement limitation in Section 5. Overall proficiency was low (CTS = 28.3%; HOTS = 42.2%). Pearson correlations and simple linear regressions showed a significant positive relationship between CTS and HOTS in both subgroups (males: R = 0.626, R² = 0.391; females: R = 0.709, R² = 0.503; both p < 0.001). A direct moderation test found that the CTS × gender interaction was not significant (b = −0.054, t = −0.711, p = 0.480; ΔR² = 0.004), indicating that the difference in R² values does not reflect a significant difference in slope based on gender. Given the cross-sectional nature of this study, the results are interpreted as associative and predictive, not causal; intervention and longitudinal studies are needed.
