This note examines a specific decision-quality failure: a decision package presented with high apparent confidence, resting on evidence that would not, on close inspection, support that level of confidence. The evidence itself may be entirely legitimate — the problem is not fabrication, it is overstatement of what the evidence can bear.

The mechanism. Decision research documents persistent problems of overconfidence and miscalibration. Organizational settings can add incentives for decisive language, but the relevant governance question here is narrower: whether the confidence stated in the decision package exceeds what the evidence supports. This is a distinct issue from evidence quality — the underlying calibration mechanics of confidence versus evidence reliability are examined in depth elsewhere in this archive (see Evidence Sufficiency). Here, the concern is narrower and decision-specific: does the confidence language used in the decision package match what a careful reader of the underlying evidence would conclude, or does it exceed it.

Why this is a decision-quality problem rather than an evidence problem. The same evidence can be presented as “strongly supports” or “is consistent with, though not conclusive,” and a board will treat these framings very differently even though the underlying facts have not changed. The decision quality question is whether the choice of confidence language was itself deliberated and defensible, or whether it was simply the language that made the recommendation easiest to approve.

Governance consequence. When a decision relies heavily on stated confidence without separately examining what the evidence supports, the confidence language can begin doing judgment work that the underlying evidence has not earned.

Evidence boundary. The general tendency toward confidence inflation under organizational pressure is well documented in decision-science and behavioral literature, including findings that this tendency persists even among people who are actively trying to correct for it. The specific application here — that decision packages should be reviewed for confidence-language calibration as a distinct step from evidence review — is Obraval’s interpretation.

The executive question. Read the confidence language in the recommendation separately from the evidence itself. Does the evidence, read cold, support that language — or does the language do work the evidence cannot?