Formative Feedback¶

Strategy · All strategies
Evidence · 14 claims (9 for, 2 mixed, 3 against) · 26 studies (9 quant-synthesis, 8 review, 4 causal, 4 associational, 1 theoretical), q2–q4 · 8 of 26 report an effect size · 2 claims rest on one study

Description¶

Formative feedback is information given to a learner about their work, at a point where they can still change the work or their approach to the next one. What makes it formative is not its timing alone but its content and its consequence: it addresses the task or the process rather than the person, it names a specific next action, and it is followed by an opportunity to act. A comment on a returned final paper that no one will revise is summative regardless of how detailed it is.

Design Implications¶

Feedback is among the highest-variance interventions in education — large average effects concealing a substantial minority of studies where feedback made performance worse Feedback improves learning by a medium amount on average, but its effect varies widely with the information it carries, and a substantial share of feedback effects are negative [+S]. Kluger and DeNisi's meta-analysis located the mechanism: feedback that draws attention to the self rather than the task diverts effort into managing self-image instead of improving work, and this is where negative effects concentrate Praise for intelligence, rather than for effort, after success undermines children's motivation and performance after a later failure [-M]. Person-directed praise and bare grades both do this.

The usable design rule comes from Hattie and Timperley: effective feedback answers three questions — where am I going, how am I doing, and where to next — and operates at the task, process, or self-regulation level rather than the self level Feedback that answers three questions (Where am I going? How am I doing? Where to next?) improves learning. [+M] Feedback Most Effective At Task And Process Levels [+S]. The second rule is that feedback only works if it is used: information the learner reads and files changes nothing Feedback improves learning, with a medium average effect that varies widely by feedback type [+M]. Most of the design work in practice is engineering the uptake, not improving the wording.

Context¶

Requirements¶

• Criteria the learner already understands, so "where am I going" has a referent (Criteria Development, Learning Goals)
• A subsequent opportunity to act — a revision, a re-attempt, or a closely related next task. Without it the feedback cannot be formative
• Time protected for learners to read, interpret, and respond, treated as classwork rather than homework
• Feedback specific enough to act on: naming what to change and how, not only that something is wrong
• Separation from grades, since a mark attached to a comment reliably captures the learner's attention and suppresses engagement with the comment Praise for intelligence, rather than for effort, after success undermines children's motivation and performance after a later failure [-M]
• Clear success criteria shared with learners before work begins, so feedback has a reference point
• Timing that lets learners revise or retry — feedback delivered after the opportunity to act is wasted
• A climate of psychological safety; learners must see feedback as information, not judgment
• Follow-up structures: revision cycles, Practice, or Coaching that convert feedback into action
• A clear target: learners must know what good performance looks like (criteria, exemplars, rubrics)
• Low-stakes opportunities to perform and receive feedback (Practice, drafts, Check-Ins)
• Time and structure for learners to act on the feedback — revision, retry, or self-explanation
• Feedback that specifies the gap and a next step, not just a grade

Constraints¶

• Feedback directed at the person — praise, ability attributions, discouragement — moves attention to the self and can depress subsequent performance Praise for intelligence, rather than for effort, after success undermines children's motivation and performance after a later failure [-S]
• Excessive feedback overwhelms: marking every error on a piece of work produces a document the learner cannot act on and often does not read Cognitive Overload Degrades Learning [-M]
• Feedback at the self-regulation level presumes learners who can act on it; novices given "think about your strategy" without a concrete task-level correction have nothing to do [~M]
• Learners without sufficient prior knowledge cannot judge whether feedback applies to their work, and may implement it mechanically or not at all Prior Knowledge Needed For Accurate Self Assessment [~M]
• Delayed feedback on a task the learner has moved on from produces no revision and little learning [-M]
• Peer-generated feedback varies in accuracy with the peer's own expertise, so it needs criteria and moderation Peer feedback accuracy depends on expertise. [~M]
• Overly detailed feedback can overwhelm novices; too little leaves experts unmoved — the expertise-reversal effect applies to feedback granularity [~M]
• In high-stakes evaluative climates, learners may optimize for the grade rather than the improvement the feedback describes [~M]
• Feedback without an opportunity to apply it produces little learning gain Formative assessment practices improve achievement. [-M] — the "formative" label is meaningless unless the loop closes
• Grades or scores attached to feedback can absorb attention and erase the effect of comments; learners focus on the number, not the advice [~S]
• Overly frequent or vague praise reduces effectiveness and can create dependence on external evaluation [~M]
• Feedback that highlights ability rather than strategy can harm motivation for struggling learners [~M]

Implementation Variability¶

• Comment-only marking — grades withheld until after revision, so the comment is the only available signal (Effective Feedback)
• Whole-class feedback — the teacher reads a set, identifies the three common issues, and teaches into them rather than writing individual comments
• Audio or video feedback — spoken commentary conveys more nuance per minute and reads as less punitive (Audio Feedback)
• Feed-forward on the next task — feedback framed entirely as instructions for the upcoming piece, which guarantees an opportunity to act
• Peer feedback against criteria — distributes the load and benefits the giver as much as the receiver (Peer Feedback)
• Drafting cycles — feedback embedded in a required draft-revise sequence (Drafting and Feedback Cycles)
• Teacher-to-learner feedback: the default mode; strongest when tied to explicit criteria
• Peer feedback: builds evaluative judgment; requires training and structured protocols (e.g., two-stars-and-a-wish, rubric-guided critique)
• Self-assessment: develops self-regulation as learners internalize the criteria Self-monitoring improves self-regulation. [+M]
• Automated feedback: scalable and immediate, but limited to well-specified task domains
• The same formative logic applies to teacher professional growth: performance management framed as developmental rather than evaluative, with the teacher participating in setting goals, yields feedback that is more meaningful and applicable to daily practice
• Whole-class feedback: common errors addressed collectively, saving time and reducing individual marking load
• Peer feedback: structured protocols (e.g., two-stars-and-a-wish, gallery critique) build evaluative judgment while distributing the feedback load
• Automated feedback: hints and correctness signals in platforms like Khan Academy or Codecademy deliver immediate, on-demand feedback at scale
• Self-assessment against exemplars: learners compare their work to models, building the internal standards needed for independent performance

Target Learners¶

• Learners working on anything revisable — writing, design, code, performance, problem sets with re-attempts
• Novices, who benefit most from task-level correctives and least from abstract process commentary
• Learners who currently receive grades without commentary, for whom almost any actionable specific is an improvement
• Weaker fit where the work is genuinely final and no related task follows; there the honest description is summative assessment
• Requires care with learners for whom critique carries high social or emotional cost, where framing and privacy matter as much as content
• High-achieving learners often need less corrective feedback and more challenge; excessive guidance can reduce their engagement [~M]
• Learners with low self-efficacy need early, achievable success feedback to sustain persistence Self-efficacy predicts academic persistence. [+M]
• Advanced learners benefit from feedback that raises challenge rather than confirms success [~W]
• Younger learners need feedback delivered immediately and concretely; older learners can act on delayed, more abstract feedback [~M]

Target Learning Goals¶

• Improved performance on the specific task and on structurally similar later ones
• Understanding of quality criteria — what "good" consists of in this domain Self Assessment Against Criteria Supports Self Regulated Learning [+M]
• Self-regulation: internalizing the feedback question set so learners eventually ask it of their own work
• Correction of specific misconceptions, particularly confidently held ones High-confidence errors are more likely than low-confidence errors to be corrected on a later test after feedback. [+M]
• Skill refinement: closing specific gaps between current and target performance
• Self-regulated learning: building learners' capacity to judge and adjust their own work
• Conceptual understanding: feedback that surfaces misconceptions for revision Erroneous examples build conceptual knowledge. [+M]
• Skill refinement: correcting procedures and strategies during acquisition
• Conceptual understanding: surfacing and addressing misconceptions before they stabilize
• Self-regulation: building learners' capacity to judge and adjust their own work

Instructions¶

1. Make the target visible first. Share criteria or exemplars before the work begins, so feedback has something to point at (Criteria Development).
2. Limit yourself to two or three points. Choose the ones that would most improve the next attempt, and leave the rest unmarked (Cognitive Load Management).
3. Write at the task or process level. Describe what the work does and what to change; avoid comment on the learner's ability, effort, or character.
4. Answer "where to next" concretely. End each point with a specific action — a sentence to rewrite, a step to redo, a check to run.
5. Withhold the grade until revision is done. If a mark must be recorded, record it privately and release it after the learner has responded.
6. Schedule the response in class. Give protected time for learners to read the feedback, mark what they will change, and make the change (Feedback).
7. Require evidence of uptake. Ask learners to indicate where they acted on each point, which makes non-use visible to both of you Feedback improves learning, with a medium average effect that varies widely by feedback type [+M].
8. Check whether it landed. Look at whether the named issue actually changed; if it did not, the problem is usually clarity or opportunity, not motivation.

Related Strategies¶

• Formative Assessment Cycles — the surrounding loop that generates the evidence this feedback responds to
• Effective Feedback — the general characteristics of feedback that works, across formative and other uses
• Drafting and Feedback Cycles — the structural guarantee that feedback has something to act on
• Peer Feedback — distributes the giving, and benefits the giver The quality of feedback students give peers predicts gains in their own work, but no recorded study isolates a causal benefit of the assessor role [+M]
• Audio Feedback — a delivery variation that raises nuance per minute of teacher time
• Exemplar-Based Feedback — using annotated examples to communicate the standard rather than describing it
• Assessment for Learning — the assessment system that generates formative feedback
• Activating Prior Knowledge — knowing the starting point makes the feedback gap meaningful
• Self-Monitoring — the internalized endpoint of formative feedback
• Use Worked Examples — exemplars set the quality standard that feedback references
• Use Retrieval Practice — retrieval tasks generate the evidence that formative feedback acts on
• Use Mastery Learning — formative feedback is the mechanism by which mastery thresholds are reached
• Ten recommendations for teachers on effectively using formative assessment in the classroom

Examples¶

Comment-only marking in secondary science: Work is returned with two specific comments and no mark; learners spend the first ten minutes of the next lesson responding in a different colour, and grades are released only afterwards.

Whole-class feedback sheets: Rather than annotating thirty scripts, the teacher records the three most common errors, two strong examples, and one whole-class next step, then teaches directly into those before returning the work.

Code review as formative feedback: Reviewers comment on the code and the approach, never on the developer; each comment names a concrete change, and the pull request cannot merge until the author has responded to every one — uptake is structurally enforced.

Writing conferences: A three-minute one-to-one on a draft, ending with the learner stating the single change they will make next — the "where to next" question answered by the learner rather than the teacher. - MyWritingLab / Turnitin Feedback Studio — inline instructor and automated comments on drafts before final submission, making revision the point of the feedback cycle. - Khan Academy — immediate correctness feedback and hint scaffolds on every practice item, so learners act on feedback within seconds. - Peer observation programs in schools — teachers observe colleagues using a structured framework and exchange developmental (non-evaluative) feedback, applying the same formative logic to educator growth; this works best when teachers help set the observation goals and feedback is decoupled from formal evaluation. - Dylan Wiliam's "embedded formative assessment" — engineering classroom discussions, hinge questions, and comment-only marking (no grades) so students must read and use the feedback. - Dylan Wiliam's "embedded formative assessment" techniques — hinge questions, comment-only marking, and traffic-light self-assessment — operationalize feedback loops in everyday classroom routines. - Khan Academy (https://www.khanacademy.org) — immediate correctness feedback and step-level hints on practice exercises, with mastery tracking that lets learners retry until proficient. - Draft-and-revise writing workshops — teacher and peer comments on drafts, with revision required before grading, so feedback is consequential.

Key Sources¶

• Hattie, J., & Timperley, H. (2007). The power of feedback. Review of Educational Research, 77(1), 81–112. doi:10.3102/003465430298487
• Shute, V. J. (2008). Focus on formative feedback. Review of Educational Research, 78(1), 153–189. doi:10.3102/0034654307313795
• 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), 254–284. doi:10.1037/0033-2909.119.2.254
• Black, P., & Wiliam, D. (1998). Assessment and classroom learning. Assessment in Education: Principles, Policy & Practice, 5(1), 7–74. doi:10.1080/0969595980050102
• Wiliam, D. (2011). Embedded formative assessment. Solution Tree Press.