How good is your rubric? From design to practice

Rubrics are among the most widely used assessment tools in higher education. When well designed and properly implemented, they can improve grading consistency, reduce student anxiety, support self-regulation, and help students develop evaluative judgement. A broad literature documents the characteristics of effective rubrics, and university teaching centres have produced useful practitioner guides. Yet this knowledge remains scattered: individual studies address isolated design features, practitioner documents are rarely grounded in the full evidence base, and no single instrument brings it all together.  

What is a rubric? 

An analytic rubric is a table with criteria (rows) that define what is being assessed, performance levels (columns) that set the possible achievement levels, and descriptors (cells) that define what work looks like at each level for each criterion.  

The 12 criteria 

In my recent working paper (Santos, 2026), I propose a 12-criterion Rubric Quality Framework organised into four dimensions, each grounded in academic research. 
 
Rubric structure 

1.1 Number of criteria. Three to six named criteria capture the key dimensions of a task without cognitive overload. 

1.2 Performance levels. Three to five levels, applied consistently across all criteria, balance discrimination with usability. 

1.3 Aggregation rule. An explicit rule for combining criterion scores into a final grade. 

Criteria design 

2.1 Observable traits. Criteria should target qualities of the work that any trained assessor can observe and evaluate consistently. 

2.2 Criterion independence. Each criterion should assess a distinct aspect of the work. 

Performance Descriptors 

3.1 Descriptor quality. Descriptors should be specific enough to assess against real work, avoiding vague terms. 

3.2 Qualitative differentiation. Adjacent levels should describe meaningful differences in what the work does, not merely vary adverbial intensifiers (e.g. “clearly identifies,” “identifies,” “partly identifies”). 

3.3 Attribute consistency. The same attribute must be assessed at every level within a criterion. 

3.4 Descriptor framing. Descriptions should use positive, actionable language rather than deficit-focused phrasing (e.g. “the work lacks”). 

3.5 Traits per cell. Each cell should address only as many traits as can be assessed as a coherent whole. 

Language and Equity 

4.1 Language accessibility. Rubric language must be clear and accessible to the students using it. 

4.2 Cultural and linguistic neutrality. Criteria and descriptors must not embed dominant cultural or linguistic norms as universal standards of quality.  

Where rubrics fall short 

Applying the framework to 79 publicly available rubrics yielded a mean score of 9.2 out of 12, but the criterion-level results reveal important gaps (see Figure 1). The three lowest-scoring criteria are: descriptor framing (3.4), qualitative differentiation (3.2), and aggregation rule (1.3). 

Figure 1: Percentage of rubrics meeting each quality criterion, grouped by framework section. 

An implementation guide 

Good rubric design is necessary but not sufficient for effective use. So I complement the analysis with a 13-step implementation guide across three stages.  

Pre-assessment 

A.1 Clarify purpose. Decide whether the rubric is primarily formative, summative, or both. 

A.2 Map intended learning outcomes. Each criterion should trace back to at least one ILO, covering all assessed outcomes without gaps. 

A.3 Apply the Rubric Quality Framework. Check the rubric against all 12 criteria, revising until each is met. 

A.4 Seek expert review. Ask a colleague with disciplinary expertise whether the criteria reflect appropriate standards. 

A.5 Pilot before use. Apply the rubric to past student work with a colleague and refine where disagreements arise. 

Assessment period 

B.1 Share early. Distribute the rubric when the assessment is introduced. 

B.2 Teach the rubric actively. Use a brief in-class exercise or pre-recorded video explaining what each descriptor means and what it looks like in practice. 

B.3 Provide annotated exemplars. Pair the rubric with anonymised examples illustrating what different performance levels look like (ideally a mid-range piece). 

B.4 Support peer feedback. Structured peer feedback guided by the rubric improves both the work reviewed and the reviewer’s own work. 

B.5 Invite self-assessment before submission. Rubric-guided self-assessment supports the development of self-regulated learning. 

Post-assessment 

C.1 Write rubric-referenced feedback. Explain what the work achieves at the assigned level and what would need to change to reach the next. 

C.2 Calibrate before grading. Have assessors independently grade a small sample, then compare scores to reach a shared interpretation. 

C.3 Review and refine. After each cycle, identify what needs revision.  

Whether you are building a new rubric or revisiting an existing one, the 12 criteria provide a structured way to identify what might be improved. Combined with the implementation guide, they offer a path from a stronger document to more effective use in practice. 

References 

Santos, Ana (2026). What Makes a Good Rubric? A Quality Framework and Diagnostic Analysis. Working Paper.