Fact, Process, Accuracy: A Distractor Design Framework for Diagnostic Multiple-Choice Items in School Science

A conventionally scored multiple-choice item discards most of the information a student's response carries: a wrong answer is recorded as zero and the reason for it is lost. This report specifies Fact-Process-Accuracy (FPA), a distractor design framework in which every distractor of a single-tier item is constructed to carry exactly one diagnosable fault, located either in the declarative knowledge the item depends on (Fact), in the procedure applied to that knowledge (Process), or in the final form of the answer (Accuracy), while the key is correct on all three. We give definitions of the three dimensions, a six-step item-writing protocol built around a single-fault constraint, a partial-credit scheme with a response-to-feedback mapping, and three fully worked items with derivations drawn from NCERT Class 10 physics, chemistry and biology. We position FPA against misconception-driven distractors, two-tier diagnostic tests and Ordered Multiple-Choice items, and argue that its distinguishing property is that faults are defined over the solution procedure rather than over a construct-specific learning progression, which makes items cheaper to author and tractable as a target for automated generation. We report no empirical validation here, and the framework's psychometric assumptions, in particular the ordering of the three dimensions, remain untested; we state them explicitly as design assumptions rather than findings, and describe the LLM-based implementation that subsequently operationalised the framework.