What is an assessment framework?
An assessment framework is the governed structure that holds an organisation's rubrics and the weightings applied to them. It establishes who may change a criterion and what approval that change requires, and it records which version of the rubric applied to which cohort. A rubric describes a single assessment, where the framework governs every assessment an organisation performs.
Also called: Evaluation framework
Why thresholds need an owner before they need a number
The moment a framework starts to matter is the second cycle. In the first cycle the rubric is fresh in everyone's mind, and the people who agreed the weightings are still in the room. By the fourth cycle the weightings have been adjusted twice with nobody able to say on whose authority, and the organisation cannot reproduce the standard it applied two cycles ago. A framework is what makes a rubric an institutional asset rather than a document somebody owns.
- A framework fixes authorship, so every criterion has someone accountable for it.
- Versioning lets an organisation reproduce the standard applied to a past cohort.
- Changes pass through approval rather than arriving as an edit to a spreadsheet.
- A single framework makes separate assessments comparable to one another.
How an assessment framework works
- The dimensions available to any rubric are defined as a controlled vocabulary.
- Each rubric declares its dimensions and the weighted scale in use.
- Thresholds are set with the justification for their position recorded alongside.
- A change to any element creates a new version with an approver and a date.
- Each decision references the framework version under which it was taken.
- Superseded versions are retained for the full retention period of the decisions.
How an assessment framework is evaluated
A framework is assessed on whether it can answer questions about past decisions. Test the following against cycles already closed:
- ability to reproduce the exact rubric version applied to a given cohort
- proportion of weighting changes with a recorded approver
- time required to answer which threshold applied on a given date
- consistency of dimension definitions across separate rubrics
- count of criteria with no named owner
Limitations, risks and common failure modes
- Frameworks are frequently documented once and then maintained in parallel spreadsheets.
- A weighting adjusted mid-cycle produces two incomparable halves of one cohort.
- Dimension names are reused across rubrics while meaning something different in each.
- Retention policy is set by the system vendor rather than by the appeal period.