Designing Assessment as a Learning Process
Student self-assessment and peer-assessment become educationally powerful when they do more than assign a judgment: they give learners evidence they can interpret and a next move they can make. The final question is therefore not whether students can produce a score that matches a teacher’s score, but whether the assessment process helps them notice quality, explain a gap, revise their work or strategy, and monitor what happens next. The model below brings together the shared mechanism, the different roles of self- and peer-assessment, and the limits of the evidence supporting them.
This conclusion is deliberately conditional. The research supports a positive pattern, but not a universal protocol. It supports criteria, feedback, reflection, preparation, and action as important parts of the process, but it does not show that adding a rubric, a peer-review round, or a self-rating will automatically produce durable learning. The difference between a useful assessment and a ceremonial one lies in the links among its parts.
The unit of analysis is a closed learning loop
The smallest meaningful unit is not the moment of judgment. It is the cycle that begins with a formative purpose and ends with a better-informed next action. A teacher who wants students to improve an explanation might state the learning goal, make the relevant qualities visible through criteria or a model, ask students to inspect a draft, and create time for revision. A student who only reports a rating has completed one act. A student who uses the rating or explanation to change the draft has participated in the loop.
The loop has a practical order. First, the learner needs to know what the work is trying to accomplish. Second, the criteria must make that purpose inspectable. Third, the learner needs evidence from a product or process. Fourth, the learner makes a judgment about how the evidence relates to the standard. Fifth, the judgment must be interpreted: what explains the gap, and which part is most worth changing? Sixth, the learner revises, changes a strategy, seeks help, reallocates effort, or chooses a different next task. Finally, the learner monitors the new attempt. That last step matters because the next piece of evidence tests whether the adjustment had the intended effect.
This order is not a mandatory form with seven boxes. It is a way to diagnose missing causal links. If the purpose is summative, the learner may have no reason to change the work. If the criteria are vague, the learner cannot locate evidence. If the evidence is available but the judgment is not explained, the learner may know that something is wrong without knowing what to do. If feedback arrives but no revision or next step follows, the process stops at information. The loop is closed only when assessment changes subsequent activity.
Self-assessment and peer-assessment enter this loop through different evidence sources. Self-assessment gives the learner a direct route to their own process, product, confidence, and uncertainty. It can support ownership and internal monitoring, especially when the judgment is task-specific and criteria-referenced. Peer assessment adds another person’s reading of the work. The reviewer practises applying and explaining criteria; the recipient compares an external perspective with their own interpretation. A coordinated process can use both, but combining them does not erase the difference between internal evidence and social evidence.
The role of a rating becomes clearer in this model. A number can summarize a judgment, make progress visible, or prompt a conversation. It cannot by itself reveal whether the problem is missing evidence, weak reasoning, poor organization, or an irrelevant response. Qualitative explanation is not automatically better either; a comment detached from the criteria or impossible to act on can be as unhelpful as a score. The question is whether the representation gives the learner enough information to choose a defensible next move.