Beyond Correctness: Tracking Reasoning Sequences to Investigate Conceptual Instability in Student Learning
Correct answers are often treated as evidence of conceptual understanding, yet students may arrive at correct answers through memorization, guessing, or conceptually flawed reasoning. This study investigates whether tracking sequences of reasoning states across related problems can reveal emerging conceptual instability that individual answers and overall accuracy scores may fail to capture. The research examines whether correct answers supported by unsound conceptual reasoning may precede errors when related concepts are tested in structurally different ways. It also investigates whether the occurrence of a conceptual error provides information about the reasoning state that follows. The study draws on an analysis of student assessment work and self-reported examination experiences, alongside a separate analysis of the researcher's own entrance-examination practice data. These datasets are examined independently to explore reasoning quality, error patterns, and performance across related problems. The findings indicate that answer correctness and conceptual reasoning are distinct aspects of student performance. The analysis identified different types of reasoning and error states that may be obscured by conventional correctness-based assessment. A small number of comparable follow-up cases in the separate JEE dataset showed that an initial conceptual error was not invariably followed by another conceptual error. However, the available transition evidence was insufficient to establish a reliable predictive relationship between earlier reasoning states and subsequent performance. This study explores reasoning-sequence analysis as a complement to conventional accuracy-based assessment. By shifting attention from answer correctness alone to relationships between reasoning states across related problems, it investigates a more detailed approach to identifying conceptual difficulties and understanding how they emerge during student problem-solving.
Authors
- S S Haya
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-10-26
- DOI
- https://doi.org/10.5281/zenodo.23656110
- Primary Topic
- Science Education and Pedagogy
- Type
- preprint