{
  "schema": "vedokrok.public-item.v1",
  "release_id": "MHC-RPUB-20260920-75ad787a",
  "url": "/knowledge/name-the-smallest-change-worth-acting-on",
  "id": "MHC-D-RESEARCH-0266",
  "version": "0.1.0",
  "title": "Name the smallest change worth acting on",
  "summary": "Detectable is not the same as worth the migration.",
  "kind": "template",
  "body": "Define a practically meaningful change before the result arrives. Include implementation costs, consequences and relevant tradeoffs. Then compare the estimated effect and its uncertainty with that threshold, rather than treating statistical significance as an automatic instruction to act.",
  "limits": [
    "Some safety or rights requirements are not appropriately traded away for an average performance gain."
  ],
  "topics": [
    "union-statistical-judgment"
  ],
  "intents": [],
  "source_ids": [
    "RS-535ADDC5A7E803D5"
  ],
  "evidence": [
    {
      "claim": "A statistically detectable difference can be too small to matter practically, while a consequential difference can remain uncertain in a small study.",
      "source_id": "RS-535ADDC5A7E803D5",
      "role": "supports",
      "note": "The practical threshold must come from the decision context, not from the p-value.",
      "locator": "Practical Versus Statistical Significance"
    }
  ],
  "use_when": [
    "A study or experiment may detect differences too small to justify a real decision."
  ],
  "avoid_when": [
    "Some safety or rights requirements are not appropriately traded away for an average performance gain."
  ],
  "example": "A tiny speed gain may not justify replacing a reliable workflow that would require weeks of retraining.",
  "check": "The decision threshold is explicit and was not chosen solely to flatter the observed result.",
  "template": "A change would justify [decision] if it improves [outcome] by at least [meaningful amount], without worsening [important constraint].",
  "sources": [
    {
      "id": "RS-535ADDC5A7E803D5",
      "title": "Quantitative Techniques",
      "url": "https://www.itl.nist.gov/div898/handbook/eda/section3/eda35.htm"
    }
  ],
  "relations": [
    {
      "from": "MHC-D-RESEARCH-0266",
      "to": "MHC-D-RESEARCH-0267",
      "type": "useful_with",
      "url": "/knowledge/ask-what-a-nonsignificant-study-could-have-missed"
    }
  ],
  "collections": [
    {
      "id": "RC-D4FD0B28897E9517",
      "title": "Check what a study can actually tell you",
      "url": "/collections/check-what-a-study-can-actually-tell-you"
    }
  ]
}