Peer Review in the Classroom: Helping Students Evaluate Each Other's Science

This isn't a new idea for this classroom. In 2008, The American Biology Teacher published “Science Seminar: Science Capstone Research Projects as a Class in High School” (Schwebach, 2008), describing a course at The Beacon High School in New York City where every senior, not just the ones already inclined toward research, completed an original science project as a graduation requirement. Communicating the work and assessing it were two of that course's four organizing themes — peer review was built into the structure, not added as an afterthought. What follows scales that same idea down to a single lab period and up to a full research paper, and connects forward to Battlefield Science, the student-edited research journal this school runs today — the largest-scale version of everything below, covered fully in Article #10 of this Teaching tab.

One discipline holds all three scales together: peer feedback works best formative, and gets weaker the moment it's also the grade. Black and Wiliam's landmark review of over 250 formative-assessment studies found effect sizes of 0.4–0.7 when students are given feedback aimed at improvement rather than a score, with the largest gains for the students who most need them (Black & Wiliam, 1998). Peer review that's scored the moment it's given teaches students to perform politeness, not evaluation.

Three scales, one underlying discipline

In the lab: corroborating data across groups

After a lab, before anyone writes a conclusion, put every group's raw results where the whole class can see them — the board, a shared doc, whatever's fastest. Then work the room through three questions together: where do groups agree, where do they diverge, and what would explain the divergence — measurement error, a procedural difference, or something real? This does two things at once. It's informal peer review of methodology (a group whose numbers are wildly out of line usually finds out why, from their classmates, before it ever reaches a grade). And it hands every student a genuine, pooled evidence base — larger and more reliable than their own group's data alone — to actually cite in their own lab write-up's CER paragraph.

In the unit: the formative-only gallery walk

For students who don't yet have a semester-long project, displaying work is the most efficient way to get real peer feedback into a single class period. Post work-in-progress — a data poster, a partial draft, a diagram — around the room. Students circulate in pairs with a simple feedback slip at each stop:

The feedback slips go to the project's owner, not to the teacher, and nothing about the gallery walk itself is graded — only the final version is. That single design choice is what keeps this formative: students give and receive real critique because nothing is at stake in the moment except the work getting better. Left ungraded is also what keeps feedback from collapsing into surface-level praise; research on peer-assessed gallery walks has found students default to commenting on surface polish unless a structure like the three prompts above steers them toward substance.

In the project: visiting-peer review, scored against a shared rubric

For a longer research project or paper in progress, bring in “visiting peers” — students from another lab group or another class period who haven't been immersed in this specific project — to review a work-in-progress draft. A visitor who wasn't in the room for every decision reads more like a real reviewer and less like a groupmate returning a favor. Score the review itself against the same claim-evidence-reasoning lens this site's CER article already builds (Article #5), so a student's peer-review skill and their own writing skill are trained by the same framework instead of two unrelated ones:

The three-way review: student, AI, and you

This is the newest layer, and it changes what a teacher's own role in peer review looks like. Encourage students to use AI while drafting, but require them to bring back, in their own words, what the AI actually suggested before moving forward — the reporting step is where the thinking happens, not the AI conversation itself. From there, the teacher's job shifts from grading a finished draft to reviewing the collaboration: what did the AI contribute, what did the student actually understand and use, and what's the next specific push that gets the work somewhere neither the student alone nor the AI alone would have reached. That's peer review too — the teacher as a collaborator interrogating the instrument, in exactly the posture Ch. 14's CER framework and Ch. 15's instrument-versus-reasoner argument both describe (see this site's CER article, Article #5, and the instrument-vs-reasoner protocol, Article #9). The AI's draft becomes the text under joint review, not a substitute for either the student's thinking or the teacher's.

Schwebach, J. Reid, “Science Seminar: Science Capstone Research Projects as a Class in High School,” The American Biology Teacher, Vol. 70, No. 8 (October 2008), pp. 488–497, National Association of Biology Teachers — the author's own prior publication; the four organizing themes cited here (hypothesis building, supporting research, communicating/organizing, and assessment) are drawn from the published abstract, not the full text — pull the original article for exact mechanisms before citing them more specifically. Paul Black and Dylan Wiliam, “Inside the Black Box: Raising Standards Through Classroom Assessment,” Phi Delta Kappan, 1998 (effect sizes, formative vs. summative peer assessment). Gallery-walk-as-formative-protocol and the surface-feedback tendency without structured prompts: general peer-assessed gallery walk research in K-12 STEM education literature — verify current sources before publishing. NGSS: SEP-3 (planning and carrying out investigations), SEP-7 (engaging in argument from evidence), SEP-8 (obtaining, evaluating, and communicating information).