List, Array, and Dictionary: A Teaching-Order Synthesis for Minimum Novice Confusion

The introductory Python curriculum has to introduce three collection abstractions — the built-in list, the NumPy ndarray (the cultural referent for "array" in Python-pedagogy contexts), and the built-in dict — and the order in which they are introduced determines a large share of the misconceptions the curriculum produces. The pedagogical literature has converged on a partial answer for sequences before mappings in the abstractdata-type tradition, but the empirical work has not adjudicated the three-way ordering at the level Python-first curricula require: where in the sequence does dict belong (immediately after list, deferred until after function-passing is taught, or interleaved with the sequence concept), and where does the typed contiguous ndarray belong (early as a numeric alternative to list, deferred until after list and dict are stable, or scoped out of the introductory band entirely). This paper synthesises Sajaniemi's roles-ofvariables programme, Sorva's notional-machine work, Caceffo et al.'s 2016 misconception inventory, Kohn's Python-specific name-binding and error-message work, du Boulay's notionalmachine analysis, the Robins-Rountree-Rountree 2003 review and the Qian-Lehman 2017 update, the McCracken et al. 2001 multi-institutional assessment, the Lister 2004 / Lopez-Whalley 2008 tracing-explaining-writing studies, Guo's Python Tutor visualisation work, the Brown-Wilson 2018 ten-quick-tips, the NumPy array-programming literature, and the cognitive-loadtheory updates. The paper delivers three contributions. First, a five-criterion methodology for adjudicating data-structure teaching order: cognitive-load profile, transfer from prior knowledge, Python-idiom fit, misconception density, and prerequisite chain. Second, a per-candidate-ordering misconception inventory (listfirst, dict-first, array-first, list-then-array-then-dict, list-thendict-then-array) drawing on the existing literature. Third, a recommendation: list (introducing sequence and mutation), then dict (introducing key-value mapping abstraction), with NumPy ndarray deferred until a specific numeric or scientific-computing use case justifies it, supported by a falsifiable rationale, a designrule set, and an explicit mapping onto the PP and FS2P Vol 2–3 chapter windows. Per the elix-researches no-cohort constraint, no learner data is collected; deployment validation is identified as a downstream cohort responsibility and is out of scope here.