In China’s comparative-education scholarship, the long-dominant “translation–reportage” style facilitates knowledge transfer but falls short of explaining why an effect occurs, for whom, and where it fails. This paper proposes a mechanism-comparison framework grounded in constructivist grounded theory that integrates mechanism–context–boundary across concept calibration, coding, and cross-case constant comparison, and standardizes outputs via a three-tables-plus-one-figure layout and a reporting checklist. Methodologically, we implement dual-layer grounding (within-case → cross-case) + constant comparison + negative-case sampling: we first extract the A→M→Y (trigger–mediator–outcome) chain within a focal context; then, across contexts, we identify stable links and contextual modifiers and use negative cases to locate thresholds and scope conditions. Findings are compressed into middle-range propositions and coupled to QCA (set calibration and recipe testing) and process tracing (time-stamped evidentiary tests) to form a mutual-corroboration pipeline. The contribution is threefold: shifting the unit of comparison from country vignettes to process mechanisms; rendering “context” as a measurable modifier and applicability domain; and formulating boundary propositions that state “where the chain no longer holds.” We provide a replicable workflow and presentation templates that serve both usable explanations and auditable evidence. The framework is well-suited to policy-transfer/lesson-drawing questions and small-to-medium qualitative projects that can triangulate with texts/administrative data. We conclude with templates and a minimum-compliance checklist, discuss scope conditions and common pitfalls, position the approach as complementary to MSSD/MDSD, and outline future directions in computation-assisted constant comparison and cross-team corroboration.
Liao et al. (Tue,) studied this question.