Training Facilitation and Group Handling Assessment for Trainers and FacilitatorsThe room tells you what it needs. Almost none of it is said out loud.

Methodology: Forty-four original exercises in two strands and six formats: sixteen keyed room reads on a six-point scale, fourteen situations with graded partial credit, four select-every-that-applies exercises, four keyed claims, three matching exercises and three most-and-least forced choices. The two strands carry two different response instructions and are therefore never combined into one figure: the reads ask what is happening in the room, which is a judgement about evidence, and the situations ask what the respondent would most likely do, which is a behavioural tendency. Construct statement: it measures the accuracy of somebody's read of a group from what the group actually did, and separately the quality of what they would do about it. It does not measure presentation skill, subject knowledge, likeability or confidence in front of a room, it is not a certification to train anybody, and it does not predict how a particular session will go. Read accuracy is scored as a profile-match correlation between the respondent's sixteen ratings and the keyed sixteen. A correlation is invariant to a linear shift, so a person who uses the top of the scale throughout and a person who uses the middle are compared on the shape of their reading rather than on where they sit, and neither elevation nor spread can earn or lose a point. Elevation and spread are still computed and reported beside the correlation, because they are findings of their own about how somebody rates. The flatness check runs on the raw presses rather than on any reversed or recentred value, because a respondent pressing the same button throughout otherwise acquires a shape that belongs to the authoring rather than to them. The eight keyed profiles were written to have genuinely different shapes, because a correlation against sixteen similar keys measures nothing. Half the read items are worded in the opposite direction so that agreeing with everything cannot produce an accurate profile. The response strand is chance-corrected against each item's own authored answer priors and reported with its own error band. Constructs and sources: adult learning and what a group brings into a room (Knowles on andragogy); psychological safety as the condition for a group speaking up, and silence as its absence (Edmondson); the distinction between observation and inference, and why a facilitator who cannot separate them cannot be corrected (the ladder of inference in the action science tradition; Argyris and Schon); group participation dynamics and the effect of structure on who speaks (nominal group and think-pair-share findings on turn distribution); interpersonal accuracy measured against a criterion rather than self-reported (Funder's realistic accuracy model; Cronbach on the components of accuracy in judging others, which is why elevation and differential accuracy are separated here); profile similarity by correlation rather than by distance, and its invariance to elevation; the post-prandial dip in alertness and the effect of a change of activity; and item-writing guidance from Haladyna, Downing and Rodriguez. All items are original works written for this instrument. No training provider's course, model, certification or material is reproduced, named or implied, and no accreditation or endorsement by any body exists or is suggested.