Decision Quality¶
Decision quality is evaluated using information available when the decision was made, not with knowledge of the outcome. Outcome matters, but answers what happened after action.
Established foundation and candidate synthesis¶
The Society of Decision Professionals describes six connected elements: frame, alternatives, information, values and trade-offs, reasoning, and commitment to action. Its chain metaphor warns that the weakest link limits the process. The CDI-BoK uses this established professional frame; the candidate contribution is a measurement surface adding traceability for Human–AI conversation, execution, risk and PULSE learning.
Why assess before outcome¶
Baron and Hershey experimentally showed that outcome knowledge can change retrospective evaluation of the producing decision—primary evidence of outcome bias, not a universal magnitude estimate. Preserve two snapshots: the pre-action decision snapshot and the later review snapshot, never overwriting the first.
Ex-ante profile¶
| Dimension | Review question | Material blocker |
|---|---|---|
| Frame and authority | Are choice, owner, deadline and affected people clear? | No real choice or action authority |
| Alternatives | Are materially different options considered, including no action? | Preferred option is the only one analyzed |
| Evidence and uncertainty | Is information fit for purpose and are unknowns and counterevidence visible? | Decisive premise lacks provenance or uses false certainty |
| Values and trade-offs | Are criteria, priorities and distributional consequences explicit? | Real criterion is hidden or excludes those bearing harm |
| Reasoning and challenge | Is the option linked to consequences and tested against a strong objection? | Recommendation lacks mechanism, comparison or stress test |
| Governance and commitment | Do rights, permission, escalation, stop conditions and action have owners? | Execution can occur outside limits or accountability |
| Expectation and learning | Are baseline, range, horizon, guardrails and review date fixed? | Expectation will be defined after outcome |
Do not average automatically.
Review scale¶
- Unknown: insufficient information.
- Weak: content exists but cannot support the decision or leaves a blocker.
- Adequate: sufficient for current stakes and reversibility, with stated limits.
- Strong: triangulated, challenged and ready for the specific decision.
Strong does not guarantee a good outcome. Adequate is not mediocre when reversible, low-cost action does not justify more research.
Proportionality¶
Review burden rises with potential harm, third-party reach, irreversibility, correction cost, material uncertainty, delegated action and hard-to-observe effects. Frequent reversible decisions may use a minimum record; critical decisions require independent review, sensitivity analysis and affected-party evidence.
What the profile must not do¶
It must not compare teams through averages hiding blockers; reward documentation volume; replace outcome with process compliance; evaluate after the fact as if later information were known; or equate committee agreement with quality.
Strongest counterargument¶
A qualitative rubric can become bureaucratic and still depends on judgment. Correct. This version therefore preserves evidence by dimension, blockers and proportionality and does not claim psychometric validation. Field evidence showing redundant dimensions, low inter-rater agreement or a less gameable predictive alternative would change this architecture.