Teaching Students to Mark Their Own Work: Self-Assessment and Peer Assessment That Improves Learning

Teaching Students to Mark Their Own Work: What the Research Actually Supports, and Where a Famous Statistic Gets Misused

A contributing voice from educational research | Specialist in curriculum, pedagogy, and school leadership

Students who regularly assess their own work against explicit criteria develop metacognitive skills that improve performance, and this claim is genuinely well supported, though one of the most frequently cited statistics used to support it deserves more caution than it usually gets. John Hattie’s influential meta-analysis found an unusually large effect size for something called “self-reported grades,” a figure widely repeated as proof that teaching self-assessment dramatically boosts learning. That statistic has been the subject of genuine methodological debate about what it actually measures, and the honest, more defensible version of the claim rests on a different, more specific body of evidence: metacognitive strategy instruction, including but not limited to self-assessment against explicit criteria.

Key facts

• John Hattie‘s widely cited meta-analysis of education research found an unusually large effect size for “self-reported grades”, often summarised in popular education writing as evidence that teaching self-assessment dramatically improves learning. This specific statistic has been the subject of real methodological critique: some researchers argue it primarily reflects the correlation between students’ predicted grades and their actual grades, a different and more limited finding than “teaching self-assessment as a skill produces large learning gains.”
• Separate from that contested statistic, the Education Endowment Foundation‘s evidence-based guidance ranks metacognition and self-regulation strategies among the highest-impact, lowest-cost categories of intervention it reviews, a more specifically evidenced and less contested claim than the single large effect-size figure often quoted.
• The specific mechanism connecting self-assessment to genuine learning gains is tied to explicit criteria: a student comparing their own work against a clear, specific standard, rather than a vague sense of “is this good,” is doing something closer to what the formative assessment research (including Sadler’s three-condition framework, familiar from formative assessment research generally) identifies as effective, knowing where you are relative to a defined goal, not simply forming a general impression.
• Peer assessment, a related but distinct practice, has its own separate research base, associated with researchers including Keith Topping, generally finding real benefits when structured around explicit criteria, similar to the self-assessment mechanism, rather than open-ended, unstructured peer feedback.

The honest version of the claim survives scrutiny better than the famous statistic

A single, very large effect-size number is easy to repeat and hard to verify without reading the methodological debate behind it. The more careful claim, that metacognitive strategy instruction, including structured self-assessment against explicit criteria, is a genuinely high-impact, low-cost intervention, is less dramatic but considerably more defensible, and it’s the version worth actually building a classroom practice around.

This distinction matters for how a teacher justifies time spent on self-assessment to sceptical colleagues or school leadership. Citing Hattie’s largest-ever effect size invites a well-informed challenge about what that specific statistic measures. Citing the more specific, more carefully evidenced finding, that structured metacognitive practice, including self-assessment against explicit criteria, is genuinely high-impact and low-cost, is a claim that holds up better under scrutiny.

Five principles for teaching genuine self- and peer-assessment

1. Diagnose whether students are assessing against explicit, specific criteria or forming a vague general impression of their own work. The mechanism the research actually supports depends on clear, defined standards, not simply asking a student “how do you think you did,” which is a meaningfully weaker version of the practice.
2. Build the criteria before asking students to assess against them. A rubric or explicit standard needs to exist and be understood by the student before self-assessment can function as the research describes; skipping this step and asking for self-assessment without a clear standard is a different, less-evidenced activity.
3. Treat peer assessment as a related but separate skill requiring its own explicit criteria. The same structural requirement, explicit standards rather than open-ended impression, applies to peer feedback, and the two practices, while related, are not interchangeable.
4. Use the relationship and trust already established in the classroom to make honest self-assessment safe. A student is more likely to genuinely and accurately assess their own weaknesses against a rubric with a teacher they trust not to penalise honesty, connecting this practice to the broader, well-established finding about relationship quality moderating instructional effectiveness.
5. Start with one specific piece of work and one explicit rubric, rather than a whole-class self-assessment initiative. Building genuine, criteria-based self-assessment practice around a single, well-defined task is more likely to demonstrate the actual mechanism than a broad, less structured rollout.

Frequently asked questions

Is the famous statistic showing self-assessment as one of the single biggest factors in student achievement actually reliable? It deserves more caution than it typically receives. Hattie’s large effect size for “self-reported grades” has been the subject of genuine methodological debate, with some researchers arguing it primarily reflects the correlation between predicted and actual grades rather than directly measuring the effect of teaching self-assessment as a skill. The more specific, less contested evidence for metacognitive strategy instruction generally, including the Education Endowment Foundation’s high-impact, low-cost ranking, is a more defensible basis for classroom practice.

Does self-assessment actually improve learning, or is the whole claim overstated given the statistic controversy? The underlying claim survives the controversy over that one specific statistic: metacognitive strategy instruction, including structured self-assessment against explicit criteria, has real, separately evidenced support, particularly through research reviewed by bodies like the Education Endowment Foundation, even setting the contested Hattie figure aside.

What’s the difference between genuine self-assessment and just asking a student how they think they did? The mechanism the research supports specifically depends on explicit, defined criteria, a rubric or clear standard a student compares their own work against, not a vague general impression. Asking “how do you think you did” without that structure is a meaningfully different and less evidenced activity.

Is peer assessment the same intervention as self-assessment, with the same evidence? Related but distinct: peer assessment has its own separate research base, associated with researchers including Keith Topping, and generally requires the same structural ingredient, explicit criteria, to produce genuine benefit rather than open-ended, unstructured peer commentary.

What to do:

1. Name one student in your class who is most affected by a lack of explicit criteria for self-assessing their own work, and write down one specific rubric or standard you will introduce for them this week, not a general improvement, but a named action for a named child.
2. Search DIKSHA or ask your Block Resource Coordinator for any resource specifically on metacognitive strategy instruction or criteria-based self-assessment in your state context. If nothing exists, note that gap; the absence of a resource is itself useful information.
3. Spend five minutes this week observing whether students in your class assess their own work against a clear standard or a vague general impression, not teaching, just watching and noting.
4. Tell one colleague about the methodological debate over Hattie’s self-reported grades statistic, and discuss which version of the self-assessment claim, the famous number or the more specific evidence, your school’s current practice is actually built on.
5. At the end of the week, ask yourself what you would try differently next week, and what you would look for, such as whether a student’s self-assessment against a rubric matches your own assessment more closely than before, to know whether it is working.

Note on sourcing: web search was unavailable during this session, so current, specific citations could not be verified live. The core research described here, Hattie’s self-reported grades statistic and the methodological debate around it, the Education Endowment Foundation’s metacognition and self-regulation guidance, and peer assessment research associated with Keith Topping, reflects well-established, broadly known findings and debates in educational research, but a reader preparing this for publication should verify specific citations before treating them as fully sourced.

Support resources: NIPUN Bharat FLN assessment tools (via your state SCERT or DIET); NCERT teacher guides; Azim Premji Foundation open-access research; DIKSHA platform content in 36 languages; Tele-MANAS mental health helpline: 14416 or 1800-891-4416; CHILDLINE: 1098.