Competency Framework¶

Reference artifact — instantiates Competence Calibration Feedback

A leveled map of what capability looks like at each stage of a domain, giving the calibration loop a fixed reference to measure self-assessment and evidence against.

Competency Framework is the artifact the rest of the loop reads from — not itself a comparison or a conversation, but the fixed ruler they all use. Its defining move is to make "competent" observable and leveled: it draws a boundary around a named domain and lays out, stage by stage, what performance looks like from novice through expert, in behavioral terms rather than felt ones. Because the levels are written down and shared, every other mechanism — the assessment, the feedback, the decision-rights rule — is measuring against the same thing rather than against each person's private notion of "good." It is a map, and like any map it does nothing until someone travels by it.

Example¶

A hospital's nursing staff keep self-assessing all over the place: some new grads call themselves "advanced," some veterans undersell genuine expertise. The unit adopts a competency framework — a clinical ladder that names, for each level, what a nurse at that stage can actually be observed doing: a novice follows protocols step-by-step and needs the rules made explicit; a proficient nurse reads a deteriorating patient holistically and reprioritizes without being told. The ladder draws its stages from a recognizable model of skill acquisition that runs from rule-following novice to intuitive expert.^[1] Now when a nurse rates herself "proficient," the claim is checkable — proficient means something specific and observable — and the charge nurse, the preceptor, and the nurse herself are all pointing at the same rung. The framework didn't calibrate anyone by itself; it gave every calibration in the unit a common language.

How it works¶

What distinguishes it is that it is deliberately static and descriptive. It fixes the domain boundary so feedback can't drift into a vague judgment of "general ability," and it specifies each level in observable behaviors so that a claim like "I'm at level 3" has a truth condition. It carries no evidence, captures no self-assessment, and runs no comparison — it is the shared referent those operations consume. Its whole value is being written down, agreed, and stable enough that many people can calibrate against one standard.

Tuning parameters¶

Level granularity — a few broad bands versus many fine steps. Fine steps show progress vividly but invite box-ticking; broad bands are robust but blunt.
Behavioral concreteness — abstract traits ("shows leadership") versus observable behaviors ("reprioritizes a caseload when a patient destabilizes"). Only the observable form is actually calibratable.
Domain breadth — one narrow competence versus a whole role's map. Broader is convenient but blurs the boundaries that keep feedback specific.
Descriptor authority and currency — who authored the levels (expert panel, regulator, local team) and how the framework is kept up to date as the work changes.

When it helps, and when it misleads¶

Its strength is that it gives the whole loop a shared, stable language, so self-assessment and evidence are measured against the same yardstick instead of against each participant's private standard. Well-built stage models make the difference between levels concrete enough to see.^[1]

It misleads most when it is mistaken for the loop itself — "we have a competency framework, so we're calibrated" — while nothing actually compares anyone's evidence to it; a ruler on the shelf measures nothing. Leveled ladders can ossify into bureaucratic box-ticking, where collecting sign-offs replaces demonstrating the competence. And a poorly built framework encodes a bad or biased standard that every downstream calibration then faithfully enforces. The discipline is to treat it as an input to feedback and assessment rather than a substitute for them, and to validate and refresh the descriptors so the ruler stays true.

How it implements the components¶

Competency Framework fills only the reference-standard slice — the ruler, not the measuring:

competence_domain — draws the boundary of the capability being calibrated, so feedback is about a specific competence, not general worth.
performance_benchmark — supplies the leveled reference standard that assessments, feedback, and decision-rights rules all measure against.

It does not capture self-assessment (self_assessment, Confidence Rating Scale), gather evidence (performance_evidence_set, Skills Assessment), or run any comparison (calibration_gap_map, Calibration Exercise); the framework is the ruler others measure with.

Instantiates: Competence Calibration Feedback — supplies the fixed, leveled standard the rest of the loop calibrates against.
Sibling mechanisms: Skills Assessment · Decision Rights by Competence · Benchmarked Feedback · Exemplar Comparison · Confidence Rating Scale · Calibration Exercise · Calibration Conversation · Peer Review · Reflective Error Log · Simulation or Case Test · Supervised Practice

Notes¶

The most common failure is confusing the artifact with the archetype. Owning a competency framework feels like calibration but isn't — the loop only exists when evidence and self-assessment are actually compared against the framework and acted on. Kept as a static reference that nothing reads, it is documentation, not feedback.

References¶

[1] The Dreyfus model of skill acquisition describes a progression from rule-following novice through competent and proficient to intuitive expert; Patricia Benner applied this arc to nursing in From Novice to Expert. Such stage models are useful precisely because they turn "getting better" into a sequence of observable, nameable levels a person can be placed on. ↩