Skip to content

Immediate Feedback

Principle · All principles
Evidence · 25 claims (10 for, 14 mixed, 1 against) · 33 studies (12 review, 7 causal, 5 quant-synthesis, 5 qualitative, 3 theoretical, 1 design), q1–q4 · 3 of 33 report an effect size · 19 claims rest on one study

Conditional relationship

This page covers when corrective feedback arrives relative to a learner's attempt: during or straight after it, while the learner can still use it on the next attempt, or later, after a fuller performance or a delay. The relationship it proposes is conditional. For a learner practising a task where an error can be seen and corrected (a recall item, a procedural step, a short constructed response), corrective information delivered before the next attempt is expected to keep the error from being rehearsed and to change that next attempt. Whether delivering it at once, rather than at the end of the item set, the session or the task, raises delayed-retention or near-transfer is a separate question, and the wiki's evidence does not settle it: in the one direct comparison recorded here, immediate and delayed feedback were equally effective, and one second-hand report favours delayed feedback for vocabulary learned incidentally from reading.

So the page's working rule is that feedback presence and content are the better-supported levers, and timing is a design choice to be made for the task and tested locally, not a default to maximise. What the designer can rely on is narrower: correct errors, especially confident ones, before the learner repeats them; make sure an attempt comes first; and keep a chance to use the correction. The Formative Assessment principle holds the broader model of using a response to choose the next action, and timing is one parameter of that action. The Retrieval Practice principle holds the condition that retrieval benefits depend on initial success or corrective feedback; this page does not repeat it. The feedback element's Design Decisions carry the same reading of the timing evidence ("timing is not a strong lever in the evidence the wiki holds"), and this page is written to agree with it. No pattern page instantiates this principle; the strategies timely feedback and timely_feedback are recipes, not tested models.

Default design, while the relationship is untested

The page's earlier guidance, kept as a concrete default a designer can act on and revise. Each step is a design proposal; the evidence status is stated beside it.

  1. Make the learner's action visible. Feedback needs a step, response, draft move or decision to respond to. Untested; from the earlier page.
  2. Let the attempt come first. Give feedback after the learner has tried, not instead of trying: a correction after a failed retrieval is where feedback has shown its largest benefit, and feedback after correct answers added little. Evidence: Feedback Enhances Retrieval Practice [+M], laboratory word pairs and classroom quizzing; the timing of that feedback is not reported on the claim page.
  3. Correct errors before they are practised again, confident errors first. If the learner will answer the same or a similar item again before any feedback arrives, deliver the correction before that attempt. On multiple-choice practice, always give feedback, because chosen lures are reproduced later; immediate or end-of-set delivery both worked. Evidence: lure intrusions halved by feedback, immediate and delayed equally [~M]; the confident-errors-first ordering rests on the hypercorrection claim under Further evidence, not re-read here.
  4. Say what to change, not only whether it was right. A correctness signal without the correct answer or a next step can leave the misconception in place. Untested here; from the earlier page. The wiki has no claim comparing elaborated with answer-only feedback.
  5. Give an immediate chance to use it. Feedback matters most when another attempt follows. Untested; from the earlier page.
  6. For whole performances (writing, projects, discussion, complex skills), consider synthesised feedback after the performance rather than interrupting each step, and keep step-level correction for steps that would otherwise be repeated wrongly. Untested; from the earlier page. No claim here compares the two for complex-skill goals.
  7. Avoid correcting in front of peers by default where it may embarrass; instructors in one interview study held back oral correction to the end of class for this reason. Evidence: teachers' stated reasons [~W] (beliefs, not learner outcomes).
  8. Consider letting the learner ask for feedback after an attempt. In one motor-skill laboratory study, learners who decided after a trial whether to see their result retained the skill better than those who decided before it. Evidence: self-controlled knowledge of results decided after the trial [~M]; carry no further than simple motor tasks until tested elsewhere.

How far each neighbouring claim may be carried: the multiple-choice and retrieval claims are about verbal-association and fact learning, mostly in laboratories and classrooms of university students; the vocabulary-timing claim is one study reported in a review; the knowledge-of-results claim is one laboratory motor task. None tests timing for procedural skills in school mathematics, clinical simulation or writing.

Fitting the design to a situation

Ask for three facts when the brief omits them: what kind of material or skill is practised (word pairs, facts, prose, a second language, a motor or clinical skill), whether the learner will attempt the same or a similar item again before feedback would arrive, and whether correction would happen in front of peers or mid-performance.

If the brief says… Then change… Basis
Multiple-choice practice or quizzes Always give the correct answer; immediately after each item or at the end of the set both work Feedback after multiple-choice tests halves lure intrusions [~M]
Pretests on paired associates, such as vocabulary pairs Show the correct answer right after each pretest item; delaying it removed the benefit Feedback timing moderates pretesting [~M]
Pretests on trivia facts or other material richer in meaning than word pairs Feedback can wait until the end of the set or a later session; with trivia facts the benefit held with delays up to 24 hours Feedback timing moderates pretesting [~M]
Second-language vocabulary learned from reading Consider delayed rather than immediate feedback; the one reported comparison favoured delay Incidental vocabulary retention and feedback timing [~W]
Oral correction of second-language errors in class Correct errors (it helps, and the benefit lasts to delayed tests); hold correction to the end of the activity or give it privately where public correction would embarrass Corrective feedback effect on L2 development is durable [+M]; Teachers favour delayed oral correction [~W]
The outcome is a delayed second-language test, not the next lesson Do not choose the correction type on the immediate post-test: explicit correction leads there, implicit correction (recasts) holds up as well or better later Explicit feedback stronger immediately, implicit better maintained [~W]; Recast gains held while metalinguistic gains declined [~M]
Homework or practice sets learners complete alone Return scored feedback before the next set, so that errors are not practised; automated scoring makes this feasible Timely feedback improves homework learning [+W]
A tutoring system or programming exercises Give immediate, elaborated feedback that says what to change at each step Tutoring systems improve outcomes with immediate actionable feedback [+W]; Immediate elaborated feedback improved programming learning [+W]
A simple motor skill practised in trials Let the learner decide after each trial whether to see the result, rather than fixing the schedule Self-controlled feedback decided after the trial improves retention [~M]
A whole performance, such as a simulated clinical consultation Debrief after the performance instead of interrupting it; interrupt only for errors that would otherwise be repeated or cause harm Participants preferred post-session debriefing [~W]
Computer-administered tests used to measure achievement Showing results item by item does not change the scores; decide on it for learning reasons, not measurement ones Immediate knowledge of results has no effect on achievement estimates [~W]
Writing or projects, with feedback on drafts Choose the timing locally and record it; no claim in the wiki compares immediate with delayed feedback on drafts untested proposal

Observation, state and explanation

What can be observed is a response on an attempt, the feedback given (its content, its timing relative to the attempt, who gave it, in front of whom), and the next response. "The feedback came too late" or "the learner was interrupted from productive struggle" are inferred states. Record the delay, what happened in between, and whether the learner had another attempt.

Response Plausible explanations Observation that could change the interpretation
Repeats the same error on the next attempt despite immediate feedback Feedback not read or not understood; feedback said only "wrong"; the error reflects a missing concept, not a slip; no time to process before the next item Ask the learner to restate the correction before the next attempt; vary answer-only against explained feedback on matched items; note the gap between feedback and the next item.
Corrects at once with feedback, fails the same item a week later Correction held in short-term memory only; the learner copied the shown answer; later learning interfered Compare a delayed probe after immediate feedback with one after end-of-session feedback on matched items; record whether any retrieval came between.
Stops trying and waits for the hint Feedback arrives before the learner has attempted; dependence on step-level support; low confidence or fear of error Delay feedback until the learner commits to an answer or confidence rating, and see whether attempts return.
Performance on a whole task drops when step-level feedback is removed Capability was supported by the feedback, not held; integration demand of the whole task; the removed feedback carried task information Probe the steps alone and the whole task without feedback; compare with a version that gives synthesised feedback after the whole task.
Withdraws or reacts negatively to correction in class Public correction threatens face; the feedback was perceived as judgement; disagreement with the correction Offer the same correction privately or in writing, and ask the learner which they prefer; see whether participation changes.

These are proposals, untested with learners. A probe can shift confidence between explanations; it does not identify a cause, and feedback during probing is itself an intervention.

Evidence and qualifications

No claim in the wiki tests the page's original statement, that immediate feedback beats delayed feedback for learning, with its own evidence. The claims below bear on the timing question directly or on the conditions around it.

Taken together: feedback after an error, especially in multiple-choice and recall practice, has the strongest evidence here; the timing comparisons that exist find no advantage for immediacy, or an advantage for delay, each in one reported study; and learner choice and public setting may matter alongside timing. Learners, tasks and outcomes differ across these claims, so do not rank them by effect label. A nonsignificant difference between immediate and delayed feedback is not equivalence.

The evidence by question

Each question below links a general claim that gathers the studies behind it into one table, one row per distinct study, with each study's design and effect as recorded and no averaged verdict (eval/synthesis/feedback-timing.json, scripts/synthesize_claim.py). The count is of distinct studies across the claim and the narrower claims that bear on it.

Further evidence, not yet read against this model

Claims this page cited before it was rewritten as a conditional model. They are evidence about the relationship, but none has yet been re-read against the model above; each says how far its own sources have been checked.

Objective and learner-valued goal

The designer's objective fixes what the feedback is for and the horizon it is judged at: getting today's items right, remembering them next month, or performing a whole task unaided. Immediate correction can raise accuracy during practice without showing anything about the later horizon, so name the horizon before choosing a timing. Elicit separately what the learner wants from feedback: to know at once whether they are right, to work a problem through before being told, to avoid being corrected in front of others. Agreement and divergence are both observations.

For example, an adult learning Italian on a phone app may want each answer marked at once, while the designer's objective is recall of the vocabulary a week later. Immediate marking serves the learner's wish and keeps lures from being rehearsed; whether it serves the week-later objective better than a short review of errors at the end of the session is untested here, and the one vocabulary report favours the delay. Recording both lets the designer try end-of-session review on a subset of items and compare delayed recall, rather than assuming either.

What would revise this model?

An advantage for immediate over delayed feedback should be stated if controlled comparisons with comparable learners, tasks and aligned delayed outcomes reproduce one, for example in procedural skills or school mathematics, where none is recorded here. If more studies like the vocabulary report find delayed feedback better for retention, say so as a default for that task type. If immediate feedback is shown to reduce learners' attempts or later unaided performance, add that as a constraint with its evidence. Do not protect the model by relabelling every null as feedback that was too slow or too vague after the fact.

Accuracy during practice, delayed retention, transfer and the learner's wish for prompt answers are separate claims. The present evidence does not establish an optimal delay, a timing rule across domains, or that faster feedback is better.

Claims

  • Formative Assessment — immediate feedback is one of the most actionable forms of formative assessment.
  • Guided Practice — guided practice depends on feedback that arrives while practice is still underway.
  • Error Analysis — timely correction helps learners understand and revise errors before they harden.
  • Check-ins — quick check-ins can trigger immediate feedback and support adjustments.

Examples

  • Step-level tutoring feedback: A digital system flags an incorrect move and gives a hint before the learner proceeds.
  • Live coaching during practice: An instructor responds to a reading, writing, or math step while the learner is still working.
  • Immediate corrective feedback on retrieval: Learners answer a prompt, see the correct answer, and try again quickly.
  • Inline draft comments during conferences: Feedback is given during revision rather than after the writing task is fully over.

Key Sources

Page metadata
Field Value
type principle
id immediate-feedback
title Immediate Feedback
description For a learner practising a task where errors can be seen, correction before the next attempt is expected to stop errors being rehearsed, but whether the timing matters, immediate rather than delayed feedback improving delayed retention, is not settled: the comparisons recorded here find no advantage or favour delay.
status review
canonical True
generated by claude/unspecified on 2026-10-02
trust unverified — no one has confirmed this page's content
sources 3 — hattie-2007, metcalfe-2017, hooley-2017

Frontmatter is the page's machine-readable record — see the Schema & Guide.