Hypothesis

For a procedural document whose purpose is immediate task execution, unassisted correct completion is the primary outcome. Task-invalidating or critical errors and assistance are guardrails; time, workload, satisfaction, and confidence are secondary diagnostics. A delayed no-document repeat or novel transfer task is required only for durable-learning claims.

Falsification criteria

Reject or revise this vector if task analysis cannot produce reliable goal states and error codes, if independent raters cannot score it consistently, if the instrument has floor/ceiling effects, or if representative users show that the selected task purpose is not immediate execution.

Evidence assessment

Standards and official human-factors guidance support multidimensional, context-bound outcomes and separate unassisted completion, assistance, time, and errors [EV-COMM-2026-0002, EV-COMM-2026-0007, EV-COMM-2026-0008]. Observed weak subjective/objective agreement makes satisfaction unsafe as a primary proxy [EV-COMM-2026-0009]. Learning research narrows the retention requirement to claims of durable learning [EV-COMM-2026-0010].

Unknowns

Task-specific scoring reliability, minimum detectable difference, participant burden, accessibility accommodations, acceptable floor/ceiling thresholds, and whether binary or partial-credit completion is more sensitive.

Implications

The measurement architecture is ready for task-specific feasibility design, but not for an effectiveness claim or a universal composite score.