Crop Advisory and Agricultural Input Recommendation Assessment for Field Officers and Input DealersThe product was right for the cause. The cause was wrong for the field.

Methodology: Thirty-nine original exercises across six formats, authored as twelve chains. Each chain opens with a prerequisite exercise (what does this field pattern or symptom indicate; which cause fits) and continues with one to four dependent exercises on the same field (what to recommend, what dose, what the label requires, what to tell the farmer and what to record). Fifteen single-choice calls, seven true-or-false claims, five select-every-that-applies exercises, four match-the-following exercises, four ordering exercises and four dose-and-dilution sliders. One response instruction is declared for the whole instrument and it is a KNOWLEDGE instruction: what is the right reading and the right advice here, never what the respondent would do on a busy market day. Construct statement: this measures crop-advisory judgment in an input-dealer or field-officer setting, across five parts: reading the field and the symptom, identifying the cause, choosing the input and the dose, safe handling and the waiting period, and what to tell the farmer and what to record. It is a competence assessment and not a certification, it is not agronomic advice for any real field, it is not a licence to sell or apply any product, and it is not a substitute for the product label or for a qualified agronomist; it says nothing about how a person performs at a real counter with a queue and a farmer who wants a product. The five parts are organised as a content map on the openly published qualification-pack unit structure for an agri-input dealer and extension services provider under the National Skills Qualifications Framework, as listed on the public National Qualifications Register by the National Skill Development Corporation and the Agriculture Skill Council of India. That structure is named for attribution only. This instrument is not affiliated with, endorsed by, or derived from that framework, its publishers, any sector skill council, any certification, any regulator, any registration list or any manufacturer, and every exercise is an original work written for this instrument. No product brand is named anywhere; actives, classes and formulations are described generically. No item depends on any one country's registration list or any one state's scheme: where a rule matters, the exercise states it in one sentence and tests whether the respondent applies it. Nothing in this instrument keys a higher dose, a shorter waiting period or a tank mixture the label does not allow; where a key touches dose, timing or mixing, the keyed answer is the label-respecting one. Farmers and fields are described by what was done and what the field shows, never by community, land size, sex or literacy. Scoring design, C23, prerequisite-conditioned accuracy with a refused conditional. Every keyed exercise carries a declared answer prior, the share of ordinary respondents expected to choose each option, each match, each position or each slider value, authored per exercise. Each exercise scores a quality between 0 and 1 (select-every exercises as balanced accuracy, keyed options selected and unkeyed options left, so that selecting everything and selecting nothing score the same), and the mean of the prior's expected quality on the same exercises is the chance term. Three figures are chance-corrected against that null so that zero is a typical respondent: BASE, accuracy over the prerequisite exercises; UNCONDITIONAL, accuracy over all the dependent exercises; and CONDITIONAL, accuracy over only those dependent exercises whose own chain's prerequisite the respondent answered correctly (quality of at least 0.75, which on a single-choice or true-or-false exercise means correct). The gap, conditional minus unconditional, is the quantity the design exists for: it separates advice that went wrong because the reading before it went wrong from advice that went wrong after a correct reading. Two rules make it a design rather than an accuracy percentage. First, the conditional figure is REFUSED, not reported as zero, when fewer than eight dependent exercises followed a correct reading; a respondent whose readings were mostly wrong has no conditional figure, and the report prints the refusal where the figure would have gone, in the same size type, because printing zero there would say you advise badly when what happened is that the instrument could not ask you. Eight is the smallest count whose assumed reliability clears the .70 floor unrounded at the assumed inter-item correlation. Second, the conditioning shrinks the denominator, so the conditional and the unconditional figures are not measured on the same number of exercises; both counts are printed beside both figures, each carries a standard-error band from its own count, and no direction is drawn between them unless the two 95 per cent bands fail to overlap; where they overlap, the report says in words that the two are not distinguishable at this length. An unanswered exercise leaves the numerator, the denominator and the chance term together; it never scores zero. An empty sitting scores exactly zero. A sitting with fewer than twenty-two answered exercises is refused a report rather than given one on too little. Reliability is McDonald's omega assumed from an inter-item correlation of .24, decided unrounded; a part with fewer than eight answered exercises or omega under .70 carries a three-way placement and no number, no band and no percentile, and the refusal is printed. The standard deviations used for the bands are taken from a modelled reference sitting of mixed ability built from the declared priors and run through this instrument's own scorer; they are assumptions stated in the open, and observed data will replace them and the priors together. No percentile appears anywhere, because there is no norm group. Three fixed habits were pushed through the real scorer before release, always the strongest input, always a spray, always the same option position, and each lands at or below zero; the figures are recorded in the seed. Sources drawn on: Agrios, Plant Pathology (5th ed., 2005) on symptom reading, spread patterns and the disease triangle; Schumann and D'Arcy, Essential Plant Pathology (2nd ed., 2010) on distinguishing living from non-living causes by pattern; Marschner's Mineral Nutrition of Higher Plants (3rd ed., 2012, ed. P. Marschner) on nutrient mobility and where a deficiency shows first; Pedigo and Rice, Entomology and Pest Management (6th ed., 2009) and Stern, Smith, van den Bosch and Hagen, The integrated control concept (1959) on economic thresholds and acting only when a count warrants it; Matthews, Bateman and Miller, Pesticide Application Methods (4th ed., 2014) on dose, dilution, volume and nozzles; the FAO and WHO International Code of Conduct on Pesticide Management (2014) on labels, protective equipment, containers, re-entry and pre-harvest intervals; Damalas and Eleftherohorinos, Pesticide exposure, safety issues, and risk assessment indicators (2011) on handling behaviour; Rasmussen, Skills, rules and knowledge (1983) and Klein, Sources of Power (1998) on diagnosis as hypothesis-driven work; Croskerry, The importance of cognitive errors in diagnosis and strategies to minimize them (2003) on premature closure, the reading that stops before the evidence does; Haladyna, Downing and Rodriguez, A review of multiple-choice item-writing guidelines (2002) on plausible distractors and cue control; McDonald, Test Theory: A Unified Treatment (1999) on omega; Gollwitzer and Sheeran, Implementation intentions and goal achievement (2006) on the if-then form of the one action; Jacobson and Truax, Clinical significance (1991) on the reliable-change threshold used for the re-measurement note.