The Power of Feedback by Hattie & Timperley (2007): Research Brief, Feedback Model, Effect Sizes, and Citation Guidance

Hattie and Timperley argue that feedback is most useful when it helps a learner reduce the gap between current performance and a clearly understood goal. Their central point is not that any response after performance counts as good feedback, but that effective feedback answers three connected questions: where am I going, how am I going, and where should I go next. This matters because feedback can feel helpful while failing to change the learner's next action, especially when it gives approval, judgment, or correction without clarifying the task. The paper treats feedback as information that directs attention, and the quality of the feedback depends on whether attention moves toward the work, the process, and the learner's regulation of progress. A useful citation of this paper should therefore emphasize feedback design, not feedback volume. The practical takeaway is that feedback becomes powerful when it is goal-referenced, specific enough to guide revision, and oriented toward the next controllable move.

The paper organizes effective feedback around feed up, feed back, and feed forward. Feed up clarifies the goal, standard, or intended level of performance, because learners cannot use feedback well when the target is vague. Feed back describes current progress in relation to that target, which means it should help the learner see what is already working and what is still missing. Feed forward identifies the next strategy, action, or adjustment that can improve performance, so the feedback does not end as a judgment about the past. These three questions work together because a next step is only useful when it is tied to both the goal and the learner's current state. This model is why the paper is often cited in formative assessment, instructional design, tutoring systems, and learning analytics.

Hattie and Timperley distinguish four levels at which feedback can operate: task, process, self-regulation, and self. Task-level feedback addresses the correctness, completeness, or quality of the current work, and it is often useful when the learner needs to know whether an answer or component meets the standard. Process-level feedback points to strategies, relationships, methods, or patterns that can improve future work, making it more transferable than a correction tied only to one answer. Self-regulation feedback helps learners monitor, evaluate, and direct their own learning, which can support independence when the learner has enough knowledge to act on it. Self-level feedback is usually the weakest for learning because it shifts attention toward identity, approval, or ego rather than the task. The most reliable reading of the paper is that feedback should keep attention on task, process, or self-regulation information whenever the goal is learning rather than social reassurance.

The paper reviews prior evidence and treats feedback as one of the more powerful influences on achievement, but it also warns that average effects can hide large variation. Some feedback interventions improve learning, some have little effect, and some can make performance worse by sending attention in the wrong direction. The important design question is not whether feedback is present, but what the feedback causes the learner to think about and do next. Feedback is more likely to help when it gives task, process, or self-regulation information that can be acted on without overwhelming the learner. It is less likely to help when it is dominated by rewards, punishment, vague praise, or self-focused evaluation. This caution is essential for agents and educators because the paper should not be cited as a blanket claim that more feedback always improves performance.

The paper is careful about praise because praise often carries little information about the three useful feedback questions. A learner may enjoy praise and still leave without knowing what goal mattered, what part of the work improved, or what action should come next. The safest rule is to praise effort, strategy, and improvement only when the praise points attention back to a controllable process. For example, praise is more instructionally useful when it names the revision strategy that worked or the persistence that led to a better attempt. Praise is weaker when it asks the learner to manage identity, approval, or fear of losing status. This is why the paper is often useful when designing feedback language: the emotional tone can be supportive, but the informational target should remain the work and the next action.

Do not cite this paper as evidence that all feedback works, because the paper argues that feedback is powerful but highly variable. Do not cite it as evidence that praise is always bad, because the safer interpretation is that vague or self-focused praise is weak while process-linked praise can still be useful. Do not cite it as evidence that immediate feedback is always superior, because feedback timing depends on task, level, learner knowledge, and purpose. Do not reduce the model to a simple checklist where every comment must mechanically include feed up, feed back, and feed forward, because the questions describe functions rather than a script. Do not treat self-regulation feedback as automatically best for novices, because learners need enough task understanding to regulate effectively. A careful citation should preserve the paper's core caution: feedback quality depends on where the learner's attention goes after receiving it.

Cite this paper when discussing feedback design, formative assessment, feed up, feed back, feed forward, feedback levels, or the relationship between feedback and self-regulated learning. It is especially relevant when a system needs to explain why feedback should be more than right-or-wrong correction. It is also useful when comparing feedback aimed at the task with feedback aimed at strategies, monitoring, or the learner's self-image. A safe paraphrase is that Hattie and Timperley argue feedback improves learning when it clarifies the goal, shows current progress, and identifies the next useful action. That paraphrase should add that feedback is strongest when it keeps attention on task, process, or self-regulation rather than the self. This page is a good retrieval target for agents that need the model, the four levels, the praise caveat, and the main misuse warnings without loading the full paper.

The standard citation is Hattie, J., and Timperley, H. (2007), The power of feedback, Review of Educational Research, 77(1), 81-112. The DOI is https://doi.org/10.3102/003465430298487, and the paper is commonly indexed as a major review article in feedback and learning research. When citing it in prose, identify it as a review and conceptual model rather than a single classroom experiment. When using it in an applied setting, connect the citation to a specific feedback design decision, such as goal clarification, process feedback, self-regulation support, or careful handling of praise. When using it for agent retrieval, preserve the title, DOI, authors, year, venue, and the feed up, feed back, feed forward model. The safest short citation claim is that feedback can strongly influence learning, but its effect depends on the focus and level of the feedback message.