Practical Spatial and Fit Reasoning Assessment for Trades, Technical and Warehouse WorkA piece can be turned round. It can never be turned over - and that is where most people lose the figure.

Methodology: Thirty-four original exercises across five formats: twelve drawn assembly figures, six keyed questions, six binary claims, four select-every-that-applies exercises and six certainty probes. The declared response instruction is knowledge and applied reasoning throughout - which figure do these pieces build, and what does this drawing fix - and one instruction covers the whole form. Construct statement: it measures whether somebody can turn a shape in their head and keep it, whether they notice handedness when a mirror image is offered, whether they can tell in advance that a thing will not go through a gap, and what they take a drawing to have fixed. It does not measure manual skill, speed of work, mechanical knowledge, mathematics or intelligence, and it predicts nothing about any particular job. Every assembly exercise is verified by construction rather than by eye: the pieces are represented as cell sets, the search covers all four quarter turns and every translation, and the generator refuses an exercise unless exactly one of the four figures can be tiled by the pieces and the other three cannot. All four figures carry the same number of cells, which removes counting as a route to the answer. On the four mirror exercises the distractor is the keyed figure reflected, and the piece set is checked for chirality first, because a mirror trap built from a symmetrical piece is not a trap. Scoring is a band ladder with honest measurement error. Every exercise is corrected against the authored answer prior on its own options rather than against a uniform guess, the corrected total is placed on a five-rung ladder whose cut scores are printed on the report, and a rung is claimed only when the whole ninety-five per cent band lies inside it. Otherwise the report names the two rungs the band spans and says so in words. Omega is estimated from the item count and a stated assumed inter-item correlation, in the open, because no live sample exists; the standard error is computed from it and carried onto the ladder. Each of the four parts rests on fewer than eight exercises, so each carries a three-way placement and never a number or a percentile. The six certainty probes are balanced in direction - three ask how sure you were and three how often you were guessing - so agreeing with everything produces no certainty at all, and they are reported beside the ladder rather than inside it. Accessibility is stated rather than implied. The assembly exercises are visual and have no non-visual equivalent, because the thing being measured is visual; the drawings use shape and position only, never colour, and the remaining twenty-two exercises are written and carry the whole of the other three parts. Constructs and sources: mental rotation and the finding that response time and error rise with the angle of rotation (Shepard and Metzler; Vandenberg and Kuse); the separation of intrinsic from extrinsic and static from dynamic spatial ability (Uttal and colleagues' meta-analysis, which also establishes that spatial skill is trainable and durable); spatial visualisation as a predictor of attainment and entry into technical and skilled trades work (Wai, Lubinski and Benbow's longitudinal work; Shea, Lubinski and Benbow); the handedness or chirality distinction between rotation and reflection, which is the single most common error in mental-rotation tasks and is why reflected distractors are standard in the literature; the classical orthographic projection result that one view underdetermines a solid; the moving-sofa and rod-round-a-corner family of clearance problems, where the binding constraint is the diagonal rather than the sum of the widths; Haladyna, Downing and Rodriguez on plausible distractors and cue control; Ebel and Frisbie on discrimination; and the standard-setting and measurement-error practice behind a band ladder (modified Angoff cuts with a reported standard error of measurement, and the rule that a band is assigned only when it lies wholly inside one rung). All exercises are original works; no commercial instrument's items, figures or name is reproduced, and no affiliation is implied.