Standing Instructions and Follow-Through Assessment for Detail-Critical WorkEverybody remembers the rule. The question is whether you remember it at the moment it applies.

Methodology: Forty original exercises across six formats: sixteen delayed-cue exercises split evenly between genuine triggers and near misses, two late recall checks, five select-every-that-applies exercises, six keyed claims, four matching exercises, three ordering exercises and four grid memory spans. Construct statement: it measures whether a standing instruction is applied at the moment it applies, when nothing on the screen mentions it, and whether the respondent can tell a genuine trigger from something that resembles one. It does not measure intelligence, does not measure clinical memory, is not a diagnosis of anything, and says nothing about any condition. Declared response instruction: knowledge and rule application, one instruction for the whole instrument. Scoring is signal detection (C6). Acting on the rule on a trigger exercise is a hit; acting on it on a near-miss exercise is a false alarm; d-prime is the discrimination and the criterion is where the respondent sets their threshold, both computed with the log-linear correction of half a trial per cell so an extreme rate cannot make either infinite. Reporting the criterion is not decoration: two people with the same discrimination, one of whom over-fires and one of whom under-fires, are different people to manage and a single accuracy percentage hides it. The two strands that are not signal detection - the memory spans and the exercises on building external checks - are chance-corrected against each item's own authored answer priors and reported apart from the headline, never averaged into it. Every competency carries a three-way placement rather than a number, because none of the four reaches the eight exercises a number would need. McDonald's omega is reported rather than alpha, estimated before live data from the exercise count and a stated assumed inter-item correlation. The trial set is balanced eight and eight by construction and the balance is audited empirically rather than asserted. Constructs and sources: prospective memory and its event-based and time-based forms (Einstein and McDaniel, multiprocess framework; Dismukes 2010 on prospective memory in laboratory, workplace and everyday settings; Loft, Dismukes and Grundgeiger on safety-critical work); implementation intentions and if-then planning (Gollwitzer; Gollwitzer and Sheeran meta-analysis); goal activation and place-keeping after interruption (Altmann and Trafton, memory for goals); signal detection theory and the separation of sensitivity from criterion (Green and Swets; Macmillan and Creelman; Hautus on the log-linear correction); checklist and external-memory design in aviation and surgery (Degani and Wiener; Gawande); the psychology of everyday slips and lapses (Reason, human error taxonomy; Norman on action slips); and item-writing guidance from Haladyna, Downing and Rodriguez. All items are original works written for this instrument. No commercial instrument is reproduced, no trademarked scale name is used, and no affiliation with any test publisher is implied or exists. The self-report questionnaires that exist in this area measure what a person believes about their own forgetting; this measures what they do, which is a different question with a different answer.