Good Manufacturing Practice and Documentation Assessment for Production, Quality and Packing TeamsTwo ways to be wrong about a batch record, and only one of them reaches a patient.

Methodology: Thirty-eight original exercises across five formats: twelve single-choice decisions, eight select-every-that-applies exercises, eight binary claims, five matching exercises and five ordering exercises, interleaved so that no format runs for more than three items. One response instruction is declared for the whole instrument and it is a KNOWLEDGE instruction: every exercise asks what good manufacturing practice requires, never what the respondent would do. Construct statement: this instrument measures whether a person knows what current good manufacturing and documentation practice requires of a record, an event or a batch, and whether they can tell a record that must stop a release from one that should not. It does not measure honesty, diligence, attention to detail on a real line, behaviour under production pressure, knowledge of any one country's regulations, or fitness for any role, and it is not a qualification and confers nothing. Scoring is a decision-cost matrix. Every decision exercise is a release-or-hold call, and every wrong response on one is classified as one of two errors: a LET-THROUGH, where a defective record or product moves forward, priced at six units; or an OVER-HOLD, where an acceptable record or product is stopped, repeated or rejected, priced at three units. The right response costs nothing. The matrix is a value judgement rather than a measurement, which is why it is printed in the seed and on the report: it says that product reaching a patient or a consumer costs twice what an unnecessary investigation costs. The two error counts are reported as a pair and never averaged into one figure; the cost-weighted figure that the matrix makes of them is printed beneath the pair and labelled as what it is. Each count is corrected against an answer prior authored on every option, so that zero in each column is what a respondent answering the way people typically answer these exercises would score, rather than a uniform draw. For select-all exercises the prior is the share of respondents expected to select each option. The two fixed habits, holding everything and releasing everything, are priced on the same form and printed beside the respondent's own figures, and the over-hold price was set so that neither habit beats the typical respondent: it began at two units, at which holding everything scored better than the typical respondent, and was raised to three. The matching and ordering exercises test procedure knowledge, are scored by position and chance-corrected against a random arrangement, and are reported apart from the cost figures. Reliability is McDonald's omega estimated from item count and a stated assumed inter-item correlation; a part with fewer than eight exercises or an omega below .70 carries a three-way placement and no number, and that refusal is printed. Every reported figure carries its standard error band. No percentile appears, because no norm group exists yet. Keying is to internationally convergent principles rather than to any country's regulation: the sources are the World Health Organization's Guideline on data integrity (Technical Report Series 1033, Annex 4, 2021) and its Good manufacturing practices for pharmaceutical products: main principles (TRS 986, Annex 2, 2014); the PIC/S Guidance on good practices for data management and integrity in regulated GMP/GDP environments (PI 041-1, 2021); the PIC/S Guide to Good Manufacturing Practice for Medicinal Products (PE 009); the ICH Q7 Good Manufacturing Practice Guide for Active Pharmaceutical Ingredients, ICH Q9(R1) Quality Risk Management and ICH Q10 Pharmaceutical Quality System; the ISPE GAMP Guide: Records and Data Integrity (2017); the ISO 9000:2015 definitions of correction, corrective action and preventive action; the ISO 22716:2007 cosmetics good manufacturing practice standard; the ISO 13485:2016 requirements for medical device quality management systems; the Codex Alimentarius General Principles of Food Hygiene (CXC 1-1969, revised 2020); the European Commission's EudraLex Volume 4 Chapter 4 on documentation and Chapter 5 on production, cited as one published statement of the same principles rather than as the key; and, for the decision-cost scoring and the item-writing rules, Haladyna, Downing and Rodriguez (2002) on multiple-choice item writing and Swets, Dawes and Monahan (2000) on decision thresholds under unequal costs. Every exercise names a principle, never a clause number, a national form or a jurisdiction-specific threshold, and no software product, vendor or certification mark is named anywhere. The ALCOA+ attribute names are a public, regulator-published set. All items are original works. No training provider's course material, certificate scheme, instrument or report layout is reproduced or implied, and no affiliation with, accreditation by or endorsement from any regulator, standards body, training provider or certification scheme exists or is claimed.