IS.2:4.3 - Decide whether another observation is worth obtaining

First ask whether existing grounds already support a response. If all live explanations support repairing the same unavailable input, the current decision may not need a unique cause. If one explanation requires training and another requires changing sanctions, a discriminating observation can prevent substantial wasted work.

Describe the possible answer before commissioning the inquiry. Name how it would change the action, who can obtain it, the conditions required and when it can arrive. Compare its whole burden with an adequate current response, a small reversible start, narrowing the promise or stopping. C.11.DUA supplies that comparison.

A low-burden contrast can be useful: try the same lookup through an authorized working route; inspect one task with and without the existing aid; ask the receiver to show what made the result unusable. Preserve the relevant conditions and the limitation of the contrast. A successful assisted attempt can show what the combined performance obtained; it leaves independent capability and the assistance’s isolated effect unsettled.

When the proposed explanation and causal reliance require more than the local contrast can establish, use A.3.1.MI:4.5, C.28 and the appropriate RMP design. Do not promise an effect estimate from a convenient before/after count. Conversely, unresolved causality does not prevent every useful local repair.