Feedback: the highest-leverage move in teaching, and the easiest to fumble

Evidence grade: STRONG that feedback is high-leverage; STRONG that its effects are wildly variable — including harmful. The landmark meta-analytic finding isn't "feedback works"; it's that feedback is among the most powerful influences on learning and that a substantial minority of feedback interventions measurably lower performance. The design details are the whole game.

Ask what improves learning most and "feedback" tops every list — correctly. But the field's foundational meta-analysis (Kluger & DeNisi, 1996, spanning a century of studies) contained the number that should hang over every marking desk: in roughly a third of studies, feedback interventions reduced performance. Feedback is not a substance that helps in proportion to dose; it's an intervention whose design decides its sign.

What separates helping from harming

The level it aims at. Hattie and Timperley's influential model (2007) distinguishes feedback about the task (what's right/wrong here), the process (what strategy would work), self-regulation (how to check your own work), and the self ("good boy," "you're clever"). The evidence pattern: task and process feedback carry the gains; self-level feedback — including praise — carries the reversals. Kluger and DeNisi's account: feedback that draws attention to the self rather than the task spends cognition on ego management (exactly the ability-praise finding, and the controlling-reward finding, converging from three literatures).

Grades crowd out comments. Butler's classic study (1988): students receiving comments improved; students receiving grades didn't; and — the finding that stings — students receiving grades plus comments did no better than grades alone. The number eats the sentence: once the grade is visible, the comment goes unread. This one result justifies comment-only marking policies for formative work, and it's why "8/10, good effort" is the most common way excellent feedback gets neutralized.

Feedback must land where work continues. Black and Wiliam's formative-assessment synthesis (1998) locates the leverage precisely: information that changes what the learner does next. Feedback on finished, closed work — however precise — is an autopsy. The operational test for any feedback practice: what does the learner do differently in the next ten minutes because of it?

Timing, honestly. The folklore says "immediate is always better"; the literature is more textured. For procedural skill acquisition and error correction, immediacy has strong support — errors practiced are errors automatized, and coaching-at-the-moment is where simulation-based practice shines. For some conceptual and retention outcomes, slightly delayed feedback sometimes matches or beats immediate (the delay adds a spaced re-encounter). Classroom translation: correct the keystroke now; the essay can wait a day — and the retrieval attempt should precede the answer either way (struggle first, help second).

What the evidence doesn't say

It doesn't say praise is forbidden — it says praise is not feedback: warmth belongs in classrooms; information belongs in feedback, and confusing the two produces the self-level effects.
It doesn't say all grades are harmful — summative judgment has legitimate jobs; the finding is about mixing it into formative moments where the comment was the point.
It doesn't crown a universal timing rule — match timing to the outcome (procedural: now; reflective: soon), and hold the strong claim loosely.

In the classroom

Aim at task and process; starve the self-level. "The second paragraph lost the argument's thread — try stating its claim first" beats "7/10" and annihilates "you're such a good writer."
Go comment-only on formative work — Butler's finding, applied. Grades where grades are due; never as garnish.
Build the next-ten-minutes loop: feedback → immediate re-attempt on a parallel item. Feedback without a re-do is a speech.
For skills, correct at the keystroke — and let errors trigger help, not precede it: attempt, then feedback, in that order.

How Wiz Kids applies this

Every task gives task-level feedback at the moment of action (the golem bonks at the wall, the formula recalculates on entry, the wrong scenario choice coaches on the click — with why, then a retry: the next-ten-minutes loop mechanized). There are no grades anywhere — stars are mastery information ("this skill, this level"), never a judgment mixed into coaching — and the fiction's warmth is deliberately kept separate from the information channel: the mascot celebrates you; the feedback talks about the work.

References

Kluger, A. N., & DeNisi, A. (1996). The effects of feedback interventions on performance: A historical review, a meta-analysis, and a preliminary feedback intervention theory. Psychological Bulletin, 119(2).
Hattie, J., & Timperley, H. (2007). The power of feedback. Review of Educational Research, 77(1).
Butler, R. (1988). Enhancing and undermining intrinsic motivation: The effects of task-involving and ego-involving evaluation on interest and performance. British Journal of Educational Psychology, 58(1).
Black, P., & Wiliam, D. (1998). Assessment and classroom learning. Assessment in Education, 5(1).
Shute, V. J. (2008). Focus on formative feedback. Review of Educational Research, 78(1) — the design-guidelines synthesis, including the timing nuances.

© Glu IO Pty. Ltd. — Wiz Kids (wiz.kids). Link freely; republication requires permission — see terms. Found an error in our reading of the research? We correct fast: tell any teacher piloting Wiz Kids.