← Every exhibit
How should a team review an indicator?
| Pattern in the review | First question | Do not conclude yet |
|---|---|---|
| A familiar label has one convenient field | Which facets of the target domain are actually observed? | The field is the construct |
| The metric is precise but has no validation path | What independent observation could challenge the reading? | Precision proves validity |
| The indicator rises after a workflow change | Did the process generate more target behavior or more recording? | The underlying construct improved |
| The field is edited after the outcome | Was the indicator available at the stated decision time? | The historical value was known ex ante |
| Missing records are coded as zero | Does absence mean non-occurrence, non-observation, or ineligibility? | Zero is the true value |
| The team assumes classical error | What evidence supports independence and the stated error distribution? | The reliability-ratio correction applies |
| A proxy is useful for triage | Is the narrower use and limitation written beside the score? | The proxy supports every downstream decision |
| A different construct has the same label | Did the target definition change between periods or teams? | The trend is comparable |
Swipe or scroll horizontally if the table is wider than your screen.
Reference & Evidence
Source: Table from this essay. Sources and interpretation are given in the article.
Related exhibits
-
What are the construct, domain, indicator, and observation?
From the essay What is measurement error? The gap between a construct and its indicator
-
The measurement-error indicator audit
From the essay What is measurement error? The gap between a construct and its indicator
-
Which mechanisms can create the gap?
From the essay What is measurement error? The gap between a construct and its indicator