<![CDATA[Students’ difficulties in solving mathematical problems are often closely related to weaknesses in metacognitive regulation, particularly in planning, monitoring, and evaluating problem-solving processes. One systematic framework to identify these difficulties is Newman’s Error Analysis, which classifies students’ errors into sequential stages of problem solving. This study aims to describe students’ metacognitive errors based on Newman’s error types in mathematics learning using a Deep Learning approach. This research employed a mixed methods approach with a sequential explanatory design, focusing on qualitative descriptive analysis. The participants consisted of three ninth-grade students from SMP Islamiyah Ciawi in the 2025/2026 academic year, selected purposively to represent high, moderate, and low levels of metacognitive ability. Data were collected through open-ended problem-solving tests on solid figures, a metacognitive questionnaire using a Likert scale, and semi-structured interviews. Data analysis was conducted by identifying students’ errors at each stage of Newman’s procedure—reading, comprehension, transformation, process skills, and encoding—and relating them to metacognitive indicators. Methodological triangulation was applied to enhance the credibility of the findings. The results indicate that students with high metacognitive ability tend to exhibit minimal errors, mainly at the encoding stage. In contrast, students with moderate and low metacognitive abilities demonstrate dominant errors at the transformation and process skills stages, with low-metacognitive students also experiencing reading and comprehension errors. These findings suggest that metacognitive regulation significantly influences the type and stage of students’ errors. In conclusion, integrating explicit metacognitive scaffolding within Deep Learning practices is essential to reduce students’ errors and enhance mathematical problem-solving performance.]]>
Copyrights © 2026
