Formative Feedback

Abstract

Formative feedback, which MeSH classifies under learning, is information given to a learner with the aim of changing their thinking or behaviour to improve learning, rather than to grade or rank what has already been done. Its defining question is not whether feedback helps but when and why: the same message raises performance in most studies and lowers it in a sizable minority, and explaining that split has been the field's central achievement. This article traces the explanation through five linked ideas — Sadler's gap, the locus of attention a message creates, the levels feedback can address, the contrast between verification and elaboration, and the learner capacities that make it work. Three interactive demonstrations let the reader close a performance gap, move the locus of feedback between task and self, and compare verification against elaborated feedback.

Keywords: formative feedback, formative assessment, feedback intervention theory, self-regulated learning, feedback literacy

What Formative Feedback Is

Formative feedback is information communicated to a learner that is intended to modify their thinking or behaviour for the purpose of improving learning (#ref-shute-2008). The word that carries the definition is formative: the feedback forms what comes next. This separates it cleanly from summative evaluation, which measures and reports attainment after the fact. A mark at the foot of a returned essay the course has finished with is summative; the same remark offered while the essay can still be revised is formative. The distinction is not in the content of the message but in whether the learner has the occasion, and the means, to act on it.

Three consequences follow from defining feedback by purpose rather than form, and they organize the rest of this article. First, feedback is relational: it only exists against a standard and a current state, so a theory of feedback is first a theory of the gap between them (#ref-sadler-1989). Second, feedback is not uniformly helpful — the way a message is framed determines whether it aids learning or damages it (#ref-kluger-denisi-1996). Third, because the learner must close the loop, feedback that is never understood or acted on is not formative regardless of how expertly it was written (#ref-carless-boud-2018). The programme of research that established formative assessment as a genuine lever on attainment (#ref-black-wiliam-1998) rests on getting all three right at once.

The Feedback Gap

The foundational account is D. Royce Sadler's. For feedback to improve learning, Sadler argued, three conditions must all be met: the learner must possess a concept of the standard (or goal) being aimed for, must be able to compare their current performance with that standard, and must engage in appropriate action to close the gap between the two (#ref-sadler-1989). Feedback, on this view, is not the teacher's comment at all — it is the whole loop, and the comment is useful only insofar as it enables the learner to run it.

The argument has a sharp implication that distinguishes it from a transmission picture of feedback. If the learner does not hold the standard, telling them they have fallen short of it conveys no actionable information; they cannot locate the gap because they cannot see the target. This is why handing students a rubric, or worked exemplars, is so often more powerful than handing them comments: it builds the concept of the standard that makes every later comparison legible. Sadler's conditions also explain the field's recurring finding that students frequently cannot use the feedback they are given — the loop fails not at the message but at the comparison or the action. The first demonstration makes the gap and its closure concrete.

Feedback Intervention Theory

The most consequential empirical correction to an optimistic view of feedback came from Avraham Kluger and Angelo DeNisi's meta-analysis of more than 600 effect sizes. The headline average was positive but modest — a mean effect of roughly d = 0.41 — yet the striking result lay in the distribution rather than its centre: in over a third of the cases the feedback intervention reduced performance relative to no feedback at all (#ref-kluger-denisi-1996). Feedback, in other words, is not a reliably benign act; sometimes telling a person how they are doing makes them do worse.

Feedback Intervention Theory (FIT) explains the split by the locus of attention a message induces. Any task is attended to on a hierarchy running from the details of the task, up through the processes used to perform it, to the self that performs it. Feedback directs attention somewhere on that hierarchy, and the direction matters more than the valence. Messages that keep attention on the task and on how to do it better tend to help; messages that push attention up to the self — praise, grades, comparison with others, anything that prompts the question how am I doing as a person? — tend to divert cognitive resources away from the task and onto self-regulation of affect, and it is these that most often depress performance. The second demonstration lets the reader move the locus and watch the predicted effect.

The Power of Feedback: Levels and Questions

Figure 1

The Hattie and Timperley Model: Three Questions across Four Levels

John Hattie and Helen Timperley synthesized the evidence into a model that has become the field's common vocabulary. Effective feedback, they argued, answers three questions: Where am I going? (the goals, which they call feed up), How am I going? (progress toward them, feed back), and Where to next? (the activities needed to make better progress, feed forward) (#ref-hattie-timperley-2007). Each question can be posed at four levels: the task (how well a task is performed), the process (the strategy behind it), self-regulation (the learner's monitoring and direction of their own effort), and the self (personal evaluation, typically as praise).

The model's enduring contribution is its ranking of the levels, which converges with FIT from a different starting point. Feedback about the process and about self-regulation is the most powerful for deep learning, because it gives the learner something transferable to do. Task-level feedback helps when the information is correct and specific but does not generalize beyond the instance. Feedback at the self level — “well done, very clever” — is the least effective, because it carries almost no information about the task and, exactly as FIT predicts, draws attention to the self. Hattie's wider syntheses place feedback among the strongest single influences on achievement, but this internal structure is the reason the average conceals so much variation.

Verification versus Elaborated Feedback

If the locus of feedback explains much of when it helps, its information content explains much of how much. The basic contrast is between verification — a signal of whether an answer is right or wrong, including simple knowledge of the correct result — and elaborated feedback, which adds information: why the answer is wrong, what the relevant principle is, what to try next. An early meta-analysis of feedback in test-like events found that feedback improved learning only when it did more than announce the verdict; when students could already see whether they were right, bare confirmation added nothing, and feedback that merely revealed the correct answer was weaker than feedback that explained it (#ref-bangert-drowns-1991).

Table 1 sets out the three feedback types this literature distinguishes, from the barest signal to the most informative, against what each tells the learner and the effect each typically produces.

Valerie Shute's review consolidated this into the working definition of formative feedback and a set of design guidelines: elaborated feedback generally produces more learning than verification alone, but more is not always better, since overly long or complex feedback can overwhelm the learner and is often ignored (#ref-shute-2008). The pattern reappears in digital settings. A meta-analysis of computer-based learning environments found that elaborated feedback outperformed both simple verification and knowledge-of-correct-response feedback, with the advantage largest for higher-order learning outcomes rather than rote recall (#ref-van-der-kleij-2015). The third demonstration contrasts the learning curves these feedback types produce.

Feedback and Self-Regulation

The thread running under Sadler, FIT, and Hattie-Timperley alike is that the learner, not the teacher, is the agent who makes feedback work. Deborah Butler and Philip Winne's theoretical synthesis made the mechanism explicit: feedback is the engine of self-regulated learning. A self-regulating learner sets goals, monitors progress against them, and adjusts strategy — and external feedback enters that cycle as information the learner uses to calibrate their internal monitoring (#ref-butler-winne-1995). On this account feedback does not act on the learner directly; it acts through the learner's own regulatory processes, which is why the same message can be transformative for one student and inert for another.

David Nicol and Debra Macfarlane-Dick carried the idea into practical principles for higher education. If good feedback is feedback that develops self-regulation, then its design should be judged by whether it clarifies what good performance is, facilitates self-assessment, delivers high-quality information about the learning rather than the learner, encourages dialogue, and provides opportunities to close the gap — their seven principles of good feedback practice (#ref-nicol-macfarlane-dick-2006). The reframing moves the teacher's effort upstream, from writing better comments to building the conditions under which students can generate and use feedback themselves.

Feedback Literacy

The most recent turn names what Sadler's conditions quietly required all along. David Carless and David Boud define student feedback literacy as the understandings, capacities, and dispositions a learner needs to make productive use of feedback: appreciating what feedback is and their role in it, making evaluative judgements about quality, managing the affect that critical feedback provokes, and taking action to improve (#ref-carless-boud-2018). The concept completes the shift the field has been making for thirty years — from feedback as a message the teacher transmits to feedback as a process the learner must be able to complete.

Feedback literacy also reconciles the apparent paradox that feedback is both one of the strongest influences on achievement and one of the most variable. If uptake depends on learner capacities that are themselves unevenly developed, then the variation is expected, and the remedy is not only better comments but deliberate instruction in how to read and act on them. This is the practical descendant of Sadler's first condition: a learner who cannot yet judge quality must be taught to, because without that judgement the comparison at the heart of every feedback loop cannot be made.

Worked Example

Consider feedback as an intervention with an average effect on performance. Kluger and DeNisi's meta-analysis reported a mean standardized effect of roughly d = 0.41 across their corpus (#ref-kluger-denisi-1996). A standardized effect of that size can be read as a shift in the standard-normal distribution: a learner who would otherwise perform at the population median (the 50th percentile) is moved to the percentile given by Φ(0.41), where Φ is the cumulative standard-normal function.

Computing it, Φ(0.41) ≈ 0.6591, so the median learner is lifted to roughly the 66th percentile — a gain of about 15.9 percentile points. That single number is the case for feedback. The case against complacency is in the same study's spread: because more than a third of individual interventions had negative effects, the 0.41 average is the net of large gains and real losses, not a uniform nudge upward. If one assumes, illustratively, that the harmful third average d = −0.20 while the beneficial two-thirds average d = +0.72, the weighted mean is (1/3)(−0.20) + (2/3)(0.72) ≈ 0.41 — reproducing the headline figure while making its composition visible. The arithmetic makes the field's central lesson quantitative: the expected value of feedback is positive and worth pursuing, but the variance is large and has a genuinely negative tail, so which feedback is given decides whether a given learner lands in the gaining majority or the losing minority.

Discussion

Four of the ideas here converge on one claim from different directions. Sadler derives it conceptually: feedback is a loop the learner closes, not a message the teacher sends (#ref-sadler-1989). Kluger and DeNisi derive it empirically: feedback that directs attention to the self rather than the task often backfires, so the message must keep the learner working on the work (#ref-kluger-denisi-1996). Hattie and Timperley derive it structurally: the process and self-regulation levels, which give the learner something to do, carry the effect, while self-level praise does not (#ref-hattie-timperley-2007). Butler and Winne derive it mechanistically: feedback operates through the learner's own self-regulation (#ref-butler-winne-1995). The practical programmes — Black and Wiliam's embedding of assessment in everyday teaching (#ref-black-wiliam-1998), Nicol and Macfarlane-Dick's seven principles (#ref-nicol-macfarlane-dick-2006), and Carless and Boud's feedback literacy (#ref-carless-boud-2018) — are all attempts to engineer the conditions under which that loop actually runs.

What the evidence does not support is the folk theory that more feedback, or more detailed feedback, is automatically better. Bangert-Drowns and colleagues showed that verification without elaboration can add nothing (#ref-bangert-drowns-1991); Shute showed that excessive elaboration can overwhelm (#ref-shute-2008); and the negative tail in the Kluger-DeNisi distribution shows that feedback can actively harm. The mature position treats feedback as a designed intervention with a known failure mode, to be shaped by where it points the learner's attention and whether the learner can act on it — not as an unconditional good.

Current Directions

The most active front updates the quantitative base and widens the lens from the message to the learner. A large re-analysis of the educational feedback literature confirmed that high-information feedback produces the largest effects and that feedback effects vary widely by type, task, and outcome — a modern restatement, on a bigger corpus, of the lesson that the average conceals the mechanism (#ref-wisniewski-2020). In parallel, a comprehensive review comparing more than a dozen feedback models and theories across half a century has begun to consolidate the field's fragmented vocabulary, mapping where the models agree, where they merely rename one another, and where genuine theoretical disagreement remains (#ref-lipnevich-panadero-2021). The open questions are increasingly about process: how learners actually decode and act on feedback over time, how feedback literacy can be taught rather than assumed, and how dialogic and technology-mediated feedback change the loop that Sadler first described. The direction of travel is away from asking whether feedback works and toward specifying the conditions under which a particular learner will use a particular message.

Key Researchers

Paul J. Black

(1930–2026). British physicist and science educator at King's College London who, with Dylan Wiliam, produced the Assessment and Classroom Learning review and the Inside the Black Box programme that established formative assessment as a robust lever on attainment and launched the modern assessment-for-learning movement. Wikipedia · Google Scholar

David Boud

(Alfred Deakin Professor, Deakin University). A leading theorist of sustainable assessment and, with David Carless, of student feedback literacy — the argument that feedback must build the learner's lasting capacity to judge their own work rather than merely correct the work in hand. ORCID · Faculty

David Carless

(Professor, University of Hong Kong). Introduced the concept of student feedback literacy and developed the theory of feedback as a dialogic, future-oriented process that learners must act on, rather than a comment to be transmitted and received. ORCID

John Hattie

(Laureate Professor, University of Melbourne). With Helen Timperley, author of The Power of Feedback, whose model organizes feedback around three questions across four levels; author of the Visible Learning syntheses that rank feedback among the strongest influences on achievement. ORCID · Wikipedia

Avraham N. Kluger

(Professor, Hebrew University of Jerusalem). With Angelo DeNisi, author of the Feedback Intervention Theory meta-analysis, which found that over a third of feedback interventions reduce performance and that feedback directing attention to the self rather than the task is the cause. ORCID

D. Royce Sadler

(Professor Emeritus, Griffith University). Author of the 1989 theoretical account that defines formative assessment as requiring the learner to hold a concept of the standard, compare it with current performance, and act to close the gap — the conceptual spine of the field. ORCID · Google Scholar

Valerie J. Shute

(Professor Emerita, Florida State University). Author of the 2008 Focus on Formative Feedback review — the canonical definition of formative feedback and the derivation of its guidelines for timing, specificity, and complexity — and a leader in stealth assessment embedded in digital learning. ORCID · Faculty

Dylan Wiliam

(Emeritus Professor, UCL Institute of Education). With Paul Black, co-author of Assessment and Classroom Learning and Inside the Black Box; the field's most influential advocate for embedding formative assessment in everyday classroom practice. Wikipedia · Faculty

Glossary

Assessment for learning.

The use of assessment to inform and improve ongoing teaching and learning, as opposed to assessment of learning, which certifies attainment after the fact.

Elaborated feedback.

Feedback that goes beyond right-or-wrong to supply additional information — why an answer is wrong, the relevant principle, or what to do next.

Feed forward.

Feedback oriented to the next task — the activities needed to make better progress — answering the question “where to next?” in the Hattie and Timperley model.

Feedback gap.

The difference between a learner's current performance and the standard being aimed for; the object that feedback exists to help the learner close.

Feedback Intervention Theory (FIT).

Kluger and DeNisi's account explaining the variable effect of feedback by the locus of attention it induces, from the task up to the self.

Feedback literacy.

The understandings, capacities, and dispositions a learner needs to make productive use of feedback — appreciating it, judging quality, managing affect, and taking action.

Formative assessment.

Assessment whose purpose is to generate information that improves the learning still to come, of which formative feedback is the operative act.

Formative feedback.

Information given to a learner to modify their thinking or behaviour for the purpose of improving learning, rather than to grade or rank completed work.

Knowledge of correct response.

Feedback that supplies the correct answer without explaining it; more informative than bare verification but weaker than elaborated feedback.

Locus of attention.

The level of the task-to-self hierarchy toward which a feedback message directs the learner's attention; the central explanatory variable in Feedback Intervention Theory.

Self-regulated learning.

Learning in which the student sets goals, monitors progress against them, and adjusts strategy; the regulatory cycle through which external feedback has its effect.

Standard.

The goal or reference level of quality against which current performance is compared; the first of Sadler's three conditions for feedback to work.

Summative assessment.

Assessment that measures and reports attainment after a period of learning, contrasted with the forward-looking purpose of formative feedback.

Verification.

Feedback that signals only whether a response is right or wrong, carrying no information about why or what to do next.

Frequently Asked Questions

What is the difference between formative and summative feedback?

The difference is purpose, not form. Formative feedback is intended to change the learning still to come, so the learner must be able to act on it; summative feedback measures and reports attainment after the fact. The same comment can be either, depending on whether there is an occasion to use it.

Does more feedback always produce more learning?

No. Feedback that only verifies a right-or-wrong verdict can add nothing when the learner already knows the outcome, and feedback that is excessively long or complex can overwhelm the learner and be ignored. What matters is the information the feedback carries and whether the learner can act on it, not its quantity.

Can feedback actually harm performance?

Yes. A meta-analysis of over 600 effect sizes found that more than a third of feedback interventions reduced performance relative to no feedback. The harm is associated with feedback that directs attention to the self (praise, grades, social comparison) rather than to the task and how to improve it.

Why is praise a weak form of feedback?

Praise directed at the self carries almost no information about the task, and it draws the learner's attention away from the work and onto self-evaluation. Both the Hattie and Timperley model and Feedback Intervention Theory place self-level feedback as the least effective and sometimes counterproductive.

What are Sadler's three conditions for effective feedback?

The learner must hold a concept of the standard being aimed for, be able to compare their current performance with that standard, and take appropriate action to close the gap between the two. Feedback is the whole loop, not just the teacher's comment.

What is elaborated feedback?

Feedback that goes beyond signalling right or wrong to explain why an answer is incorrect, state the relevant principle, or indicate what to try next. It generally produces more learning than verification or knowledge of the correct response alone, especially for higher-order outcomes.

What is feedback literacy?

The understandings, capacities, and dispositions a learner needs to make use of feedback: appreciating what feedback is and their role in it, judging the quality of work, managing the emotions critical feedback provokes, and taking action to improve. It reframes the learner as the agent who completes the feedback loop.

How does feedback relate to self-regulated learning?

Feedback acts through the learner's own regulatory cycle rather than on the learner directly. A self-regulating learner uses external feedback to calibrate their monitoring of progress toward a goal and to adjust strategy, which is why the same message can help one student and leave another unchanged.

References

Bangert-Drowns, R. L., Kulik, C.-L. C., Kulik, J. A., & Morgan, M. (1991). The instructional effect of feedback in test-like events. Review of Educational Research, 61(2), 213–238. https://doi.org/10.3102/00346543061002213

Black, P., & Wiliam, D. (1998). Assessment and classroom learning. Assessment in Education: Principles, Policy & Practice, 5(1), 7–74. https://doi.org/10.1080/0969595980050102

Butler, D. L., & Winne, P. H. (1995). Feedback and self-regulated learning: A theoretical synthesis. Review of Educational Research, 65(3), 245–281. https://doi.org/10.3102/00346543065003245

Carless, D., & Boud, D. (2018). The development of student feedback literacy: Enabling uptake of feedback. Assessment & Evaluation in Higher Education, 43(8), 1315–1325. https://doi.org/10.1080/02602938.2018.1463354

Hattie, J., & Timperley, H. (2007). The power of feedback. Review of Educational Research, 77(1), 81–112. https://doi.org/10.3102/003465430298487

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. https://doi.org/10.1037/0033-2909.119.2.254

Lipnevich, A. A., & Panadero, E. (2021). A review of feedback models and theories: Descriptions, definitions, and conclusions. Frontiers in Education, 6, 720195. https://doi.org/10.3389/feduc.2021.720195

Nicol, D. J., & Macfarlane-Dick, D. (2006). Formative assessment and self-regulated learning: A model and seven principles of good feedback practice. Studies in Higher Education, 31(2), 199–218. https://doi.org/10.1080/03075070600572090

Sadler, D. R. (1989). Formative assessment and the design of instructional systems. Instructional Science, 18(2), 119–144. https://doi.org/10.1007/BF00117714

Shute, V. J. (2008). Focus on formative feedback. Review of Educational Research, 78(1), 153–189. https://doi.org/10.3102/0034654307313795

Van der Kleij, F. M., Feskens, R. C. W., & Eggen, T. J. H. M. (2015). Effects of feedback in a computer-based learning environment on students' learning outcomes: A meta-analysis. Review of Educational Research, 85(4), 475–511. https://doi.org/10.3102/0034654314564881

Wisniewski, B., Zierer, K., & Hattie, J. (2020). The power of feedback revisited: A meta-analysis of educational feedback research. Frontiers in Psychology, 10, 3087. https://doi.org/10.3389/fpsyg.2019.03087