At least twelve meta-analyses, studies that pool the results of many others, have examined feedback in the classroom. Together they cover 196 studies and give feedback an average effect on achievement of 0.79: twice the typical effect of schooling, which is 0.40 and serves as the yardstick for judging any influence. But that power cuts both ways: “Feedback is one of the most powerful influences on learning and achievement, but this impact can be either positive or negative” (p. 81).
With that warning, John Hattie and Helen Timperley of the University of Auckland open “The Power of Feedback,” published in 2007 in the Review of Educational Research. In it, they review the evidence and propose a model to explain what separates effective feedback from feedback that does not help, or even does harm.
What Feedback Is
The authors define feedback as information about how someone is performing or understanding something, given by an agent: a teacher, a peer, a book, a parent, experience or oneself. A peer can suggest another strategy, and a student can look up the answer to check a response. Feedback, in short, is a consequence of performance.
It follows that feedback comes second, after a student has responded to some initial teaching; it does not work in a vacuum. It is most powerful when it targets mistaken interpretations rather than a complete lack of understanding, and in that case more instruction works better than feedback.
A Large Effect, but an Uneven One
The 0.79 average places feedback among the top 5 to 10 influences in a 1999 synthesis by Hattie of more than 500 meta-analyses, alongside direct instruction (0.93) and reciprocal teaching (0.86), and above homework (0.41) or reducing class size (0.12). These figures are effect sizes, a measure that lets researchers compare how much each factor matters.
The results vary widely, though. The largest effects come when students receive information about a task and how to do it better; the smallest, with praise, rewards and punishment. In one of the paper’s tables, cues, which steer the search for information and strategies, reach 1.10, while praise reaches 0.14. In Kluger and DeNisi’s 1996 meta-analysis of 131 studies, the average was 0.38, and 32% of the effects were negative.
The authors even doubt that rewards count as feedback, since they say almost nothing about the task. In Deci, Koestner and Ryan’s (1999) meta-analysis, tangible rewards such as stickers or awards undermined intrinsic motivation, the drive that comes from interest in the task itself. The study of Amabile’s componential model explains how rewards and the expectation of being evaluated also weigh on creativity.
The Three Questions of Effective Feedback
To make sense of this evidence, the authors propose a model. Feedback exists to close the gap between what someone understands or does now and a goal, and to do so it has to answer three questions, which the teacher or the student can ask:
- Where am I going? What the goals are. The authors call this feed up.
- How am I going? How far the student has come toward the goal: the feed back.
- Where to next? Which activities will move the student further: the feed forward.
The gap can be closed well or badly. Students can work harder or look for better strategies, but they can also abandon the goal, blur it, or lower it until they accept work far below what they can do. Teachers help by setting challenging, specific goals, which focus attention, and clear success criteria: without them, a goal is often too vague to improve learning.
One of the paper’s examples: comments on presentation, spelling and length do not help when the criterion for success is creating mood in a story. And when it comes to what should happen next, teaching tends to answer with more: more information, more tasks, more expectations. Used well, the authors argue, this question can have some of the strongest effects on learning.
Four Levels: Task, Process, Self-Regulation and Self
Each question works at four levels, and the level that feedback targets shapes how effective it is:
- Task: whether the work is right or wrong, or what information is missing; for example, that a piece of work needs more on the Treaty of Versailles.
- Process: the strategies for doing the task; for example, suggesting that a page would make more sense with strategies discussed earlier.
- Self-regulation: how students monitor, direct and regulate their actions toward the goal; for example, reminding a student that they already know the key features of a strong opening to an argument and asking them to check their first paragraph for them.
- Self: judgments about the student as a person, such as telling them they are a great student.
According to the authors, the self level is the least effective. Process and self-regulation feedback are powerful for deep understanding and mastery of tasks, and task feedback is powerful when the information then helps improve strategies or self-regulation, which happens far too rarely. About 90% of the questions teachers ask in class target the task level, and that feedback loses force when it is mixed with praise of the person. The most effective path runs from task to process, and from process to self-regulation.
Form matters too. Page (1958) found that short written comments, instead of grades alone, improved test results for students in 74 classrooms, and R. Butler (1988) showed that only comments on their own led to learning gains; grades, grades combined with comments, and praise did not.
The Problem With Praise
The fourth level is in the model not because it works, the authors explain, but because it is common in classrooms and often takes the place of the other three. Praise aimed at the person carries little information about the task and rarely turns into more engagement, better understanding or greater self-efficacy, a person’s confidence in their ability to succeed at something. Wilkinson (1981) found an effect of 0.12 for teacher praise; in Kluger and DeNisi’s meta-analysis, feedback with praise reached 0.09, and feedback without praise, 0.34.
There is an important distinction, though. Praise that pulls attention away from the task and toward the person is not the same as praise for effort, self-regulation or the process: the latter can strengthen self-efficacy and lead to better work.
Feedback about ability carries risks of its own. In six studies with elementary school students, Mueller and Dweck (1998) found that those who received it became more focused on performance than on learning and, after a failure, performed worse and enjoyed the task less.
Timing, Mistakes and Assessment
Timing depends on the level. In a study by Clariana, Wagner and Roher Murphy (2000), delayed feedback, compared with immediate feedback, had an effect of −0.06 on the easy items of a test, 0.35 on midrange items and 1.17 on difficult ones. According to those authors, difficult items call for more processing, and the delay gives time for it.
A student’s confidence matters as well: according to Kulhavy and Stock (1989), feedback has its greatest effect when someone was sure of an answer that turns out to be wrong. As for whether feedback is positive or negative, the level it targets matters more.
In practice, feedback is scarce in classrooms. Bond and colleagues (2000) closely documented the daily work of 65 teachers, and the most common form of feedback was praise.
For the authors, assessment should provide information that answers the three questions, not just measure each student’s level. Black and Wiliam (1998) reviewed 578 publications and concluded that classroom assessment tends to encourage superficial, rote learning. They conclude: “It is the feedback information and interpretations from assessments, not the numbers or grades, that matter” (p. 104). The article on how to assess creativity in university students applies that idea to creative work, with rubrics that describe each level.
What the Paper Leaves Us
The authors do not present feedback as a cure-all but as one powerful tool among others, and one that can only build on something: it achieves little when there has been no initial learning. Effective feedback is clear and purposeful, connects with what students already know, relates to specific goals and carries little threat to the student’s sense of self. What matters most is that it targets the task, the process or self-regulation, not the self.
For researchers, the paper calls for qualitative and quantitative studies of how feedback works in the classroom. For educators, it recommends classroom climates that encourage self-assessment and peer assessment and allow students to learn from mistakes, and it reminds teachers that they, too, must learn from feedback, such as students’ answers on a test. The article Can Creativity Be Taught? counts feedback among the learning conditions that foster creativity. For professionals, the paper warns that asking for it has costs, such as the effort or how one looks to others, and that feedback that feels good does not necessarily improve performance.
In the creativity course, each week includes an hour of self-assessment, and this paper is part of its bibliography.
Original source: Hattie, J., & Timperley, H. (2007). The power of feedback. Review of Educational Research, 77(1), 81–112. https://doi.org/10.3102/003465430298487
Learn creativity with a method
Mario Miqueles’ creativity course: 20 weeks, 4 units and a final project, online or as a book.

