Peer assessment has a reputation problem. Ask students to review each other's work and you often get back "good job, maybe add more detail." That's not because students can't give feedback. It's because the task is too open.
Narrow the focus
Ask peers to look at one or two criteria, not the whole rubric. "Find the thesis and underline it. Does every paragraph connect back to it?" is a task students can do well. "Give feedback on this essay" isn't.
Use a protocol
A protocol is a short, fixed sequence. Two that work well:
Praise–Question–Polish
- One specific thing that works (point to the line).
- One question a reader would have.
- One concrete suggestion.
Glow and grow with evidence
- Glow: one criterion the work meets, with the line that shows it.
- Grow: one criterion it doesn't yet meet, and a suggestion.
Model it first
Show an anonymous sample, and have the class give feedback together. Then show weak feedback ("it's good") next to strong feedback ("your second paragraph has evidence but doesn't explain it; add a sentence after the quote"). Students quickly see the difference.
Make feedback quality count
If peer scores feed directly into grades, students are either too generous to friends or too harsh to rivals. A better approach: grade students on the quality of the feedback they give. A two-point check (specific? actionable?) is enough.
Close the loop
Peer feedback only matters if writers use it. Have each student note which piece of feedback they used and what they changed. That reflection, more than the feedback itself, is where a lot of the learning happens.
Pair it with your rubric
Use the same criteria as your rubric, ideally a single-point rubric, so peer feedback, your feedback and the grade all point in the same direction. For your own comments, a comment bank helps keep things consistent.