Self-Assessment: From Self-Rating to Self-Regulation

The subject here is student self-assessment: the different ways learners judge their own competence, process, or work, and the conditions under which that judgment changes what they do next. The key distinction is between a self-rating that closes with a number and a self-assessment that produces usable feedback. By the end, the reader should be able to classify a self-assessment practice by its purpose and object, recognize why task-specific criteria matter, and interpret claims about “accuracy” without confusing agreement with learning.

Self-assessment is sometimes treated as a test of whether students can grade themselves correctly. That is one possible research question, but it is not the whole educational problem. A learner may judge a draft imperfectly and still notice a weakness, seek information, and revise. Another learner may guess a score that happens to match a teacher’s mark and do nothing with it. The first judgment may be more educationally valuable even though it is less precise as a prediction.

Self-assessment is a family of questions, not a single score

A broad definition is useful because students assess more than finished products. They may judge their competence to perform a task, monitor the process by which they are learning, or evaluate a product such as an essay, presentation, solution, or clinical performance. They may signal whether they understand a concept, test themselves, write a reflective entry, use a checklist, or assign a grade. These practices overlap, but their consequences depend on what the learner is being asked to notice and what the judgment is for.

The first distinction is the object. A competence judgment asks, in effect, “Can I do this?” A process judgment asks, “Is the way I am working helping me reach the goal?” A product judgment asks, “Does this piece of work meet the standard?” A task-specific competence judgment can guide a decision about practice or help-seeking. A global judgment—“I am good at mathematics” or “I am not a language person”—is much less actionable because it does not identify a task, a feature, or a next step.

The second distinction is purpose. Formative self-assessment is intended to generate feedback that informs correction, relearning, or improvement. Summative self-assessment records a judgment after the learning episode, often as a final rating or self-grade. The two can use similar words while asking the learner to behave differently. “What mark should I receive?” invites a claim about an outcome. “Which part of the work does not yet meet the criteria, and how will I improve it?” invites an investigation.

This difference is not merely philosophical. When a self-assigned mark affects a final grade, students have a reason to present the most favorable evaluation they can defend. The critical review of self-assessment research reports that self-ratings became more inflated in such circumstances and that the relationship with professors’ assessments disappeared in the group whose self-assigned grade counted. The result is not evidence that students are inherently dishonest. It is evidence that the assessment’s incentive and purpose shape the judgment it elicits.

A formative purpose also imposes a timing condition. A judgment made after every opportunity to change has disappeared may still satisfy curiosity or document a perception, but it has little leverage over learning. The formative value of self-assessment rises when the learner can use the information to revise a draft, select a practice task, seek help, adjust effort, or change a strategy. Without such an opening, “self-assess” can become a decorative pause between instruction and grading.

The third distinction is the reference point. Some self-assessments are informal and rely on a learner’s sense of understanding. Others ask learners to compare their work with a model, rubric, checklist, or set of performance standards. The reference point does not need to eliminate judgment; it makes the judgment inspectable. Instead of asking whether the learner feels confident, it gives them a way to examine why confidence is or is not warranted in this task.

A useful reference point must be understandable at the level of the work. “Show deep understanding” may describe an aspiration but leave the learner unsure what evidence to find. “Explain the central cause, connect it to two relevant observations, and distinguish it from a similar process” gives the learner something to locate. Criteria that are concrete, task-specific, and graduated can make small differences visible and support incremental adaptation. The critical review summarizes evidence that performance-based criteria outperformed vaguer competence descriptions in one task and that graduated standards helped learners detect progress in a motor-learning task.

This is why a rubric is not automatically formative. A rubric can be used as a scoring grid after the work is finished, or as a reasoning tool while the work is still changeable. In the second use, the learner looks for evidence of each criterion, records what is present or missing, and makes a plan for revision. The document has not taught anything by itself. The learner’s comparison and response have made it part of a learning process.

The same point applies to reflection. A prompt such as “How did you feel about your assignment?” may elicit an honest response without revealing what should change. A more focused prompt—“Which explanation was hardest to support, what evidence did you use, and what would you try before writing the next paragraph?”—connects feeling and judgment to a decision. Reflection becomes instructionally consequential when it helps the learner choose a better action, not simply when it produces more words.

How a self-judgment becomes a revision activity

Consider a first draft of an argument. The learner begins with a goal: make a defensible claim for a reader who may disagree. The criteria make that goal visible through features such as a clear claim, relevant evidence, and an explanation of how the evidence supports the claim. The learner then inspects the draft, not to discover a single true number, but to locate evidence for each feature. They may underline the claim, mark the supporting passage, and notice that the connection between them is assumed rather than explained.

That observation is still not feedback in the full sense. It becomes feedback when the learner interprets the discrepancy and chooses what to do. Perhaps the missing step is a sentence explaining why the evidence matters. Perhaps the source is not relevant enough and needs replacing. Perhaps the learner understands the reasoning but has written it unclearly. Each diagnosis leads to a different revision. A self-assessment that stops at “evidence: 2/4” conceals these possibilities; a criteria-linked explanation opens them.

The sequence can be represented as an interaction between the learner, the work, and the standard. External feedback may enter the process, but the learner remains responsible for interpreting it and deciding how to respond.

sequenceDiagram participant L as Learner participant T as Task or draft participant C as Criteria or model participant F as Teacher or external feedback L->>T: Produce or inspect work L->>C: Compare evidence with standards C-->>L: Make features visible L->>L: Form a self-judgment opt External information available F-->>L: Add feedback or another reading end L->>T: Revise or choose next step L->>L: Monitor the next attempt

The sequence clarifies several points that are easy to lose in a discussion of self-assessment. First, the learner needs an object to inspect. A global opinion about ability has no visible surface on which to work. Second, the criteria must be connected to the task; otherwise, the learner is comparing the work with an abstraction. Third, a self-judgment is an intermediate state. Its value depends on the revision, practice, help-seeking, or strategy change that follows. Fourth, outside information can refine the process without canceling the learner’s agency. A teacher, peer, model, or other evaluator may reveal a blind spot, but the learner still has to understand and use the information.

This chain connects self-assessment to self-regulated learning. Self-regulated learning involves setting goals and monitoring and managing thoughts, feelings, actions, resources, and context in pursuit of them. Criteria-referenced self-assessment supplies a structured occasion for that work. The learner enters with a goal, monitors performance against a reference point, generates internal feedback about the gap, and regulates the next attempt. If the learner changes the amount of practice, searches for help, reorganizes the work environment, or chooses a more suitable task, the assessment has affected regulation beyond the product itself.

The critical review reports studies in which rubric- or checklist-referenced self-assessment was associated with self-regulation across the phases of forethought, performance, and reflection. It also summarizes positive evidence for achievement: among the reviewed studies, formative self-assessment research and two meta-analyses showed positive associations, including an average effect of 0.62 on writing quality in one synthesis and g = 0.34 on subsequent tests after self-grading in another. Those figures support the possibility that formative self-assessment contributes to learning. They do not show that every self-grade produces a later gain or identify one component as the cause.

Learner perceptions help explain why process matters. The review found that younger students often had a less sophisticated understanding of what self-assessment was for, a condition that can encourage shallow compliance. In higher-education settings, students more often described formative self-assessment as useful for guiding evaluation and revision, taking responsibility, thinking critically, and regulating learning. Positive perceptions were typically linked to active work with criteria and opportunities to revise. The pattern is consistent with a simple principle: learners are more likely to value self-assessment when they can see what it is helping them do.

Accuracy is not the same as usefulness

The language of accuracy creates a trap. In many studies, a student’s self-rating is compared with a teacher’s grade or another external judgment. The closer the numbers, the more “accurate” the self-assessment is said to be. The critical review recommends the more cautious term consistency because the external judgment may itself be imperfect, and because alignment does not establish what the learner did with the information.

Consistency is still relevant. A learner who systematically misses a central feature of a task may need help seeing the standard or interpreting evidence. The review summarizes evidence that guidelines, feedback, standards, modeling, training, and experience can improve the alignment between student and external judgments. But the educational reason to improve alignment is not to produce a better number for its own sake. It is to help the learner make more suitable choices about revision and further study.

That distinction becomes clearer when comparing two situations. In the first, a student predicts a test score, receives the grade, and learns that the prediction was close. Nothing about the student’s study strategy necessarily changes. In the second, a student underestimates a draft, receives feedback, identifies evidence that was already present, and revises with greater confidence. The first is more consistent with the external result; the second may involve more learning. Neither outcome should be inferred automatically, but they show why consistency is a measurement question while useful action is a learning question.

The review also cautions against transferring results from artificial calibration tasks directly to classroom learning. Predicting whether one remembers a pair of unrelated words is not the same as deciding how to revise a complex explanation or which part of a mathematical model requires more practice. In some laboratory studies, the computer—not the learner—controls which items are restudied. That design can reveal something about monitoring, but it does not reproduce the choices students make in authentic classrooms and homework.

Nor does the evidence justify dismissing calibration altogether. A badly misread criterion can send a learner toward an unhelpful task. The better conclusion is conditional: self-assessment should be supported so that learners can compare performance with meaningful standards, obtain corrective information when needed, and use the result to select a next step. Research still needs to test the missing link directly—whether the quality of a formative self-judgment predicts the quality of subsequent study, revision, and learning.

Self-assessment, then, is not an isolated act of looking inward. It is a structured way for learners to generate and interpret information about their own work. It is strongest when the object is specific, the criteria are usable, the purpose is formative, and the learner can act. It can include a rating, but the rating is only one representation of a more important judgment. When another person’s work and interpretation enter the process, the reference point changes from an internal inspection to a social exchange. That is what makes peer-assessment a distinct route through the same learning loop.

References