The Other Half: What You Bring to a Pair, and What You Assume Somebody Else Will CatchThe Other Half: What You Bring to a Pair, and What You Assume Somebody Else Will Catch. The work that goes wrong between two people is almost never work either of them dropped. It is work neither of them thought was theirs.

Methodology: Measures what one person contributes to one close working partnership, through original situational, allocation and self-report items keyed to published constructs. Construct statement: It measures what one person brings to one close working partnership: whether they put the division of labour into words, whether they pick up the work that sits between the two roles, whether they say a small thing while it is still small, and whether they carry the other person's side when that person is not in the room. It is a one-sided instrument about a two-sided thing. It did not hear from the other person, it did not watch the two of you work, and it says nothing at all about them, about the partnership as a whole, or about whether the two of you are well matched. Declared response instruction: behavioural tendency throughout - every situational stem asks what the respondent is most likely to do rather than what a person should do, because instructed-knowledge framing is the more fakeable of the two, and every allocation stem asks which parts the respondent would take on rather than which parts ought to be taken on. Item format: sixteen situations set inside a standing two-person working relationship, each with four courses of action carrying a graded partial-credit weight from zero to three and an authored base rate; plus eight fixed-budget ownership allocations, each breaking one piece of shared work into seven concrete parts and asking which of them the respondent would take on. Every allocation carries parts that are clearly the respondent's, parts that are clearly the other person's, and two SEAM parts that neither job description covers and that both people are likely to leave; sixteen of the fifty-six parts are seams. Each situational option also carries a LOAD - how much of the shared work the move puts on the respondent, from taking the most on to leaving the most to the other person - and across the sixteen situations the mean score at each of the four load levels is exactly one and a half out of three, so no fixed habit on that axis pays. Ten balanced-keyed self-check statements complete the form, half of them worded so that agreement is not the flattering answer. Scoring design: the allocation block is scored as an expert-vector allocation match. The respondent's choices across every part of every allocation are assembled into one vector, one where they took a part on and zero where they left it, and correlated against the authored ownership vector on a five-point scale from clearly yours to clearly theirs. A correlation is invariant to linear transformation, so it scores the SHAPE of the allocation and lets the LEVEL - how much of the work was taken on at all - be reported separately as its own finding with its own fix. Both directions of the level error are real and are named in the instrument's own words: taking on what was theirs, and leaving what was yours. The signed and the absolute figure are both reported and the absolute one decides the verdict. A respondent whose allocation vector has no variation in it, having taken every part or none, has no shape to match: no shape figure is reported at all, the report says so in plain words, the level is reported as the only reading available, and the sitting is routed to the careless-responding composite. The share of SEAM parts claimed is reported as a third reading in its own right, because leaving every seam is the specific failure this instrument exists to find and it is invisible inside an overall correlation. Both the situational block and the allocation shape are chance-corrected against the declared per-option base rates rather than against uniform random, the allocation null being the Monte-Carlo mean correlation of a respondent drawn from those declared marginals, computed deterministically. The declared rates are published on the report as a range and a mean, and observed marginals will replace them. Only answered items enter both sides of every ratio; below sixty per cent coverage the report leads with that and prints no headline figure. The composite carries its standard error of measurement at both the sixty-eight and the ninety-five per cent spans, printed in words as well as drawn, and the reliability behind those spans is a stated assumption rather than a measurement. The keyed allocation has NOT been rated by a panel: an expert-vector design properly needs five or more raters with a reported intraclass correlation, this one has none, and the report carries a panel saying so and saying what would change when one exists. Each of the four competencies rests on six scored exercises and is reported as a three-way classification rather than a number, and the reliability table that justifies that refusal is printed on the report. Construct areas drawn on: role clarity and role ambiguity in work roles; task interdependence and coordination requirements in small work units; shared mental models and the convergence of two people's picture of the same job; transactive memory systems, in which who knows what is itself a shared resource; implicit coordination and the cost of relying on it when the work changes; psychological safety and the willingness to raise a small problem before it is a large one; employee voice and silence, and the decision to speak or withhold; boundary-spanning and handover work, and the research on tasks that fall between defined roles; diffusion of responsibility, in which a job everybody could do is a job nobody does; backup and backing behaviour in interdependent work; relationship maintenance and repair in continuing working relationships; complementarity as distinct from similarity in dyadic performance; equity and contribution perception in shared work; and the escalation of small unspoken grievances in long-running partnerships. All items are original works, no trademarked instrument or branded methodology is named or reproduced, and no affiliation with any source is claimed. AssessAll original design.