A Self-Assessment Tool Design for Self-Learning K-12 Informatics Students

A self-learner studying Vietnamese K-12 informatics from the elix-researches program's textbook pipeline — From Scratch to Python (FS2P, Layer A), Python Pathway (PP, Layer B), Competitive Programming Pathway (CPP, Layer C) — operates without a teacher in the loop. There is no instructor giving feedback on a weekly problem set, no chuyên-Tin classroom culture providing peer-pressure-driven calibration, no Fridayafternoon quiz to interrupt the illusion of understanding. The self-learner's only feedback channels are the volume itself (chapter-end exercises with answer keys), online resources (forum threads, Stack Overflow, AI chat assistants), and the learner's own internal monitoring. The third channel — internal monitoring — is the load-bearing one, and it is also the channel most documented as systematically miscalibrated: the KrugerDunning literature establishes that the least-skilled performers most overestimate their performance, the Stankov calibration literature establishes that confidence-accuracy decoupling is a stable individual-difference trait, the Loksa-Ko programming self-regulation literature establishes that novice programmers consistently mis-estimate trace accuracy and debugging time. A self-learner relying on internal monitoring alone is therefore the worst-case feedback configuration in the entire pipeline. This paper specifies a self-assessment tool that supplies the missing feedback channel: a weekly ten-minute-or-less instrument with five components — a skill-checklist anchored to the current volume's chapter-level learning objectives and to the E4.1 per-stage rubric (approximately ten items), a retrieval-practice quiz of approximately five items exploiting the Roediger-Karpicke testing effect for durable encoding, a single code-trace task probing notional-machine fluency in the Lister-Sorva tradition, a selfprediction-and-calibration meta-task that compares the learner's predicted score against the actual score to surface DunningKruger-class miscalibration, and a short progress-journal prompt that scaffolds the Zimmerman self-regulated-learning cycle. Scoring is rubric-band based and consistent with the E4.1 per-stage rubric so a learner's weekly tool output is directly interpretable in pipeline-band terms. The output is a three-part report: a current-week scorecard, a calibration-offset diagnostic, and a named-scaffolding recommendation drawn from a fixed catalogue. The tool is engineered against five design criteria — low-overhead use (under ten minutes weekly), actionable feedback (subscore plus named scaffolding, not a single grade), calibration-correction (explicit comparison of predicted versus actual to disrupt Dunning-Kruger overconfidence), motivation preservation (non-punitive language, framed as a check-in rather than a grade, with progress-narrative scaffolding consistent with Zimmerman SRL), and Vietnamese-language deployment (template content rendered in Vietnamese, age-appropriate phrasing per pipeline series, no English-only technical-jargon load). The deliverable is the tool specification — components, scoring