A Common Rubric for Per-Stage Assessment Across the K-12 Informatics Pipeline

The Vietnamese K-12 informatics pipeline that leads from first-encounter programming to IOI-feeder competitive readiness crosses four pedagogically distinct stages: Foundation (first contact, age-appropriate informal programming), Consolidation (Python fluency and decomposition), Specialisation (algorithm-cluster vocabulary), and Competitive (contest-format spine-cluster deployment). Existing assessment practice equips each stage with a different rubric, designed in isolation, often by a different author or institution, with different criterion names, different band labels, and different scoring conventions. The consequence is that a learner's progression across the pipeline is not legible as a single trajectory: a Foundation "proficient" rating does not commensurate with a Specialisation "proficient" rating, and the placement-diagnostic and audit loop must repeatedly re-derive what each rating means. This paper proposes a common rubric — seven criteria scored on a 4-band scale — that is constant in criterion identity across all four stages, but stage-calibrated in band-meaning so that the rubric's difficulty surface tracks the pipeline's progression. The seven criteria are C1 conceptual understanding (notional-machine fidelity), C2 procedural fluency (write-from-spec under stage-appropriate constraints), C3 problem decomposition (subgoal labelling, functionlevel abstraction, cluster-strategy retrieval), C4 debugging-andrevision (error-localisation and corrective revision against named feedback), C5 generalisation and transfer (out-of-distribution probe performance), C6 communication (explanation, code-asdocument, post-mortem narration), and C7 metacognition (selfregulated learning, self-assessment calibration). Each criterion is anchored on a published rubric-design tradition (Andrade and Du 2005; Allen and Tanner 2006; Wiggins and McTighe 2005) or programming-assessment tradition (Lister 2004; Sentance et al. 2019; ACM CC2020) and is stated as a falsifiable observable indicator at each band. The rubric is consumed by the program audit loop as the per-stage scorecard input, crosswalks to the E1.1 seven-band pipeline system, supplies the rubric-row content of the E4.4 two-axis assessment matrix, and acts as the construct-map calibration anchor for the E1.5 placement diagnostic instrument. The contribution is not novel criterion content — the criteria are inherited from the canon — but the cross-stage coherence rule that holds criterion identity constant while shifting band-meaning, and the stage-calibration tables that make the rule operational.