Which practice supports reliability and validity when designing rubrics for performance tasks?

• A. Use vague criteria and skip pilot testing
• B. Rely solely on numeric scores
• C. Define criteria with observable descriptors, pilot test, and provide exemplar anchors
• D. Avoid any criteria for reliability

Answer: (C)

When designing rubrics for performance tasks, you want scoring to be both consistent across raters and truly reflect the intended performance. The best practice is to define criteria with observable descriptors, test the rubric in a real setting to see how it works, and provide exemplar anchors that show exactly what each level looks like.

Observable descriptors make the criteria measurable. If a criterion describes specific behaviors or outcomes that can be seen or produced, different raters can identify those exact cues and score similarly, strengthening reliability. Pilot testing the rubric on actual tasks lets you spot ambiguities, misunderstood language, or gaps in the levels, so you can revise the rubric before it’s used widely. Exemplar anchors—concrete examples for each level—give both students and scorers a shared reference point, clarifying what performance at a given level looks like and supporting valid interpretations of the rubric.

Without clear, observable criteria, or without testing and anchors, reliability suffers because raters interpret the rubric differently. Relying only on numeric scores without descriptive criteria can obscure what is being measured, and avoiding criteria for reliability eliminates a mechanism for ensuring consistent judgments.

So, combining observable descriptors, pilot testing, and exemplar anchors is the strongest approach to promote both reliability and validity in rubrics.