The introduction of a new national standard for general secondary education in Latvia in 2019, developed within the competence-based education reform project Skola2030, substantially reshaped upper secondary physics education. As part of the reform, physics was reorganized into two courses: Physics I and the advanced Physics II course. Physics II, taught in Grade 12, introduced new and conceptually demanding topics, such as the equilibrium and rotation of rigid bodies, and placed increased emphasis on deep conceptual understanding and problem-solving skills.

Despite the reform’s ambitious objectives, its initial implementation revealed significant challenges. Teachers and students lacked access to instructional materials fully aligned with the updated curriculum, particularly for Physics II. In addition, physics teachers in Latvia constitute a highly heterogeneous group in terms of subject-matter expertise and pedagogical preparedness, partly due to a long-standing shortage of qualified physics teachers. These factors have created a strong need for instructional approaches that can support both teachers and students in addressing advanced physics content under reform conditions.

In this context, blended learning emerges as a promising pedagogical approach. Blended learning integrates face-to-face instruction with digital and online learning activities in a purposeful manner, enabling flexibility, personalization, and learner autonomy (Batista-Toledo & Gavilan, 2022; Liu et al., 2023; Ye et al., 2024). Empirical research indicates that blended learning can enhance conceptual understanding, student engagement, and academic achievement across diverse educational contexts, including physics education (Wang et al., 2022; Mangi & Verma, 2025). Importantly, blended learning aligns closely with the competence-based reform’s emphasis on transversal skills such as self-directed learning, learning to learn, and responsibility for one’s own learning.

A key element of blended learning in science education is the use of high-quality instructional videos. Platforms such as Khan Academy offer carefully structured explanations, visualizations of abstract concepts, and opportunities for self-paced learning. Prior research shows that instructional videos are particularly effective when integrated into flipped or blended classroom models, where students engage with video content before class and classroom time is dedicated to problem-solving, discussion, and higher-order cognitive activities (Bhaw et al., 2024; Qomara et al., 2024; Putri, 2021). At the same time, research cautions against uncritical use of online videos, highlighting risks such as passive viewing, misconceptions, and the “illusion of understanding” (Kulgemeyer & Wittwer, 2023). These findings underline the importance of expert selection, pedagogically sound integration, and the use of strategies that promote active engagement.

While blended learning has been widely examined in higher education and general science education, empirical evidence from advanced upper secondary physics courses remains limited, particularly in authentic classroom settings during large-scale curriculum reform. Moreover, few studies compare different blended learning models implemented across multiple schools.

Against this backdrop, the present study examines the implementation of different blended learning approaches, including the use of Khan Academy instructional videos, in the Physics II course in Latvian upper secondary schools. The central research question is whether, and to what extent, blended learning-supported instruction using Khan Academy videos leads to greater learning gains in advanced physics topics compared to traditional instruction.

By examining blended learning in the context of an ongoing curriculum reform and a demanding upper secondary physics course, the study provides evidence from a rarely studied educational setting and offers insights relevant for teachers, curriculum developers, and education stakeholders across Europe and beyond. Since Khan Academy materials are available primarily in English, the experience of translating and adapting instructional videos into Latvian using artificial intelligence also has broader European relevance, particularly for education systems operating in small-language contexts with limited access to high-quality, curriculum-aligned STEM resources.

Khan Academy videos aligned with the Physics II curriculum were selected from the khanacademy.org collections College Physics 1, College Physics 2, High School Physics, and Middle School Physics. The videos were translated and transcribed into Latvian using artificial intelligence tools trained to recognize physics concepts and terminology. The translated videos were made available to teachers and students via the project website (luka.lu.lv/topics). In total, more than 80 videos covering key Physics II topics (including mechanics, thermal physics, electricity and magnetism, wave optics, and modern physics) were provided. The platform also included selected materials relevant to Physics I, a glossary of physics terms, and national examination indicators aligned with the curriculum. Three teachers implemented blended learning approaches when designing and delivering instructional units. These included flipped classroom, project-based learning, and flexible blended learning models, which allow varying levels of learner autonomy, flexible pacing, and the integration of digital resources into classroom activities. In contrast, teachers in the comparison schools employed traditional instructional approaches, primarily teacher-led classroom instruction, without systematic integration of blended learning models. The study employed a quasi-experimental pre–post design with a comparison group. Students’ physics knowledge was measured before and after instruction in the Physics II topic “Equilibrium and Rotation of Rigid Bodies.” Participants included more than 100 upper secondary students (aged approximately 17–19) from seven Latvian secondary schools. The final analysis included 124 students who completed both the pre-test and post-test, providing paired measurements. 48 of them learned in a blended learning classroom and 76 in a traditional classroom. The assessment instrument consisted of an eight-item multiple-choice test designed to measure conceptual understanding of key learning outcomes in the selected mechanics topic. Each item had five answer options. The test was administered within a standardized 15-minute time frame. Classroom teachers were not informed of the test items, and testing was conducted by individuals other than the class teachers to minimize instructional bias. Test results were expressed as raw scores (0–8) and percentages (0–100%). Data analysis was conducted at both the student and group levels. Pre- and post-test scores, absolute gains, and relative gains were calculated for each student. Group-level descriptive statistics were computed, and paired-samples t-tests were used to assess pre–post differences. Effect sizes were calculated using Cohen’s d_z. All analyses were performed using Microsoft Excel. Ethical considerations included informed consent, voluntary participation, and full data anonymization.

The findings indicate that students in both blended learning and traditional classes demonstrated learning gains from pre-test to post-test in the selected physics topic. However, the magnitude of these gains varied substantially across schools and instructional models. Overall, comparing the pre-test and post-test results, combining the results of the two groups, a moderately large effect was found (Cohen's d = 0.60), indicating a significant increase in students' knowledge after a series of physics lessons taught using blended learning, while the effect was smaller after traditional lessons (Cohen's d = 0.39). At the same time some blended learning implementations were associated with large and statistically significant improvements in conceptual understanding, while others showed only modest or non-significant gains. Importantly, the results suggest that blended learning is not inherently effective by default; rather, its impact depends on how it is pedagogically designed and implemented. Differences observed across blended learning models highlight the importance of instructional coherence, purposeful integration of digital resources, and alignment with learning objectives. The variability in outcomes also reflects differences in teacher preparedness, class size, and instructional context. From a broader perspective, the study demonstrates that blended learning can support learning in advanced upper secondary physics under conditions of curriculum reform, but only when accompanied by appropriate teacher support and high-quality learning materials. The results contribute to the international literature by providing empirical evidence from an advanced secondary-level physics course in a national reform context. The findings have practical implications for policymakers and curriculum developers by underscoring the need to invest not only in digital resources but also in sustained professional development and pedagogical guidance. For teachers, the study highlights the potential of blended learning to support instruction of challenging content, while cautioning against assuming uniform effectiveness across implementations.

1. Ahmad, M., Hussain, S., Fakhar-Ul-Zaman, & Qahar, A. (2023). Learning outcomes by integrating blended learning flipped classroom model: An experiment on secondary school students. International Research Journal of Management and Social Sciences, 4(3), 566–578. https://doi.org/10.5281/zenodo.10725219 2. Batista-Toledo, S., & Gavilan, D. (2022). Implementation of Blended Learning during COVID-19. Encyclopedia, 2(4), 1763-1772. https://doi.org/10.3390/encyclopedia2040121 3. Bhaw, N., Hungwe, R., & Kriek, J. (2024). A study on the impact of Khan Academy videos: Enhancing Grade 11 thermodynamics learning in a rural high school. Science Education International, 35(2), 163–172. https://doi.org/10.33828/sei.v35.i2.10 4. Kulgemeyer, C., & Wittwer, J. (2023). Misconceptions in Physics Explainer Videos and the Illusion of Understanding: an Experimental Study. International Journal of Science and Mathematics Education, 21(2), 417–437. https://doi.org/10.1007/s10763-022-10265-7 5. Ye, Y., Zhang, G., Si, H., Xu, L., Hu, S., Li, Y., Zhang, X., Hu, K., & Ye, F. (2024). A Hierarchy-Based Analysis Approach for Blended Learning: A Case Study with Chinese Students. In X. Song, R. Feng, Y. Chen, J. Li, & G. Min (Eds.), Web and Big Data. APWeb-WAIM 2023. Lecture Notes in Computer Science, 14332 (pp. 89–102). Springer, Singapore. https://doi.org/10.1007/978-981-97-2390-4_7 6. Liu, Q., Chen, L., Feng, X., Bai, X., & Ma, Z. (2024). Supporting Students and Instructors in Blended Learning. In: Li, M., Han, X., & Cheng, J. (Eds.), Handbook of Educational Reform Through Blended Learning. Springer, Singapore. https://doi.org/10.1007/978-981-99-6269-3_5 7. Mangi, A. G., & Verma, C. (2025). Statistical Perspectives on the Effectiveness of Blended Learning in Education: A Review. In Z. Illés, C. Verma, P. J. S. Gonçalves, & Y. Singh (Eds.), Proceedings of International Conference on Recent Innovations in Computing. ICRIC 2024. Lecture Notes in Electrical Engineering, vol 1421. Springer, Singapore. https://doi.org./10.1007/978-981-96-6034-6_18 8. Putri, A. A. (2021). The effectiveness of Khan Academy as a science learning support to improve students’ mastery of skills: A literature review. Journal of Environmental and Science Education, 1(2), 52–56. https://doi.org/10.15294/jese.v1i2.50370 9. Qomara, A. L., Siswati, B. H., & Wahono, B. (2024). Flipped classroom based on Khan Academy as a student’s problem-solving abilities and cognitive learning outcomes booster. JPBI (Jurnal Pendidikan Biologi Indonesia), 10(2), 467–475. https:/doi.org/10.22219/jpbi.v10i2.33942 10. Wang, D., Zhou, J., Wu, Q., Sheng, G., Li, X., Lu, H., & Tian, J. (2022). Enhancement of Medical Students' Performance and Motivation in Pathophysiology Courses: Shifting From Traditional Instruction to Blended Learning. Frontiers Public Health, 9, Article 813577. https://doi.org/10.3389/fpubh.2021.813577