The Self/Non-Self Collaborative Methodology: Research Proposal for Formative Assessment in Learning Management Systems using the SNCM v2.0

Abstract

Formative assessment is an ongoing, low-stakes informal assessment method used by teachers to monitor student understanding and provide feedback, allowing for adjustment to instruction. Formative assessment embedded within Learning Management Systems (LMS) typically features: automated quizzes, completion trackers, and rubric-scored submissions that evaluate the correctness of an artefact rather than the trajectory of a learner's reasoning. This article proposes three contributions. Part I traces the history and empirical record of formative assessment from Scriven's original formative/summative distinction through Black and Wiliam's evidence synthesis to the mixed, more modest results of large, adequately powered trials, situating the specific evidentiary gap that motivates this work: LMS-based formative assessment is well-instrumented for behavioural and outcome data but could be better instrumented for the process of self-directed reasoning. Part II proposes the Self/Non-Self Collaborative Methodology v2.0 (SNCM) as a candidate method for closing that gap: a structured human-AI dialogic protocol in which a human agent (Self, S) directs an AI interlocutor (Non-Self, N) through bounded phases of ideation, structured debate, and iterative refinement, producing an auditable record of the learner's own evaluative judgments reinforcing learning and retention. This article does not claim SNCM is validated; instead, it specifies SNCM's constructs, scoring instruments, and algorithms in falsifiable, measurable terms: replacing prior black-box functions and an unjustified similarity threshold with an independent-rater scoring protocol, a standard-setting-derived cut score, and formally analysed termination properties. And positions the method's constructs against real, published human-AI co-writing frameworks. Part III supplies a complete, ready-to-enact research protocol: design, sample size, rater training, standard-setting procedure, pre-registered hypotheses, falsification criteria, and statistical analysis plan. So that SNCM's claims can be empirically tested by independent research. SNCM's full algorithmic specification is deliberately confined to appendices; the main text is reserved for the theoretical argument, the operational definitions, and the protocol needed to empirically test.

Keywords