{
  "schema_version": 1,
  "id": "writing-feedback-level-match-2024",
  "canonical_url": "https://brali-lifeos.github.io/evidence/writing-feedback-level-match-2024/",
  "json_url": "https://brali-lifeos.github.io/evidence/writing-feedback-level-match-2024/index.json",
  "decision": "propose-protocol",
  "decision_label": "Protocol candidate",
  "reviewed_at": "2026-08-21",
  "reviewed_by": "Brali Evidence Reviewer",
  "source": {
    "title": "How effective is feedback for L1, L2, and FL learners’ writing? A meta-analysis",
    "url": "https://doi.org/10.1016/j.learninstruc.2024.101961",
    "type": "meta-analysis",
    "doi": "10.1016/j.learninstruc.2024.101961",
    "citation_text": "Scherer S, Graham S, Busse V. Learning and Instruction. 2024;93:101961.",
    "study_design": "Meta-analysis of 95 papers meeting experimental or quasi-experimental inclusion criteria, comprising 200 treatment/control comparisons and 253 effect sizes for surface- and deep-level writing outcomes.",
    "population": "Secondary-school and university or college students represented in L1, L2, and foreign-language writing studies across the included literature.",
    "intervention_or_exposure": "Writing feedback targeting surface-level features, deep-level features, or both, delivered through sources including instructors, peers, algorithm-based systems, and self-feedback depending on the included comparison.",
    "outcomes": [
      "Surface-level writing outcomes such as grammar and spelling",
      "Deep-level writing outcomes such as writing quality, ideation, organization, genre elements, composition length, and vocabulary"
    ]
  },
  "supported_claim": "In the reviewed secondary-school and university writing literature, feedback effects differed by the level of writing outcome: surface-level feedback improved surface-level outcomes, deep-level feedback improved deep-level outcomes, and combined surface-and-deep feedback could improve both. This supports matching a feedback request to the revision outcome rather than treating all feedback as interchangeable.",
  "unsupported_or_overstated_claims": [
    "More feedback is always better for writing.",
    "Deep feedback should always precede surface feedback.",
    "Peer, instructor, algorithm-based, or self-feedback is universally the best source.",
    "The meta-analytic effects generalize unchanged to professional, technical, marketing, fiction, or other creative writing outside the reviewed educational contexts.",
    "A specific number of feedback rounds, timing interval, or review duration is optimal.",
    "Receiving feedback guarantees improvement for an individual writer."
  ],
  "limitations": [
    "The evidence concerns educational writing by secondary-school and university or college learners rather than all writing contexts.",
    "Effects varied across L1, L2, and foreign-language learners, so a single population-wide estimate can conceal important differences.",
    "The included papers used experimental or quasi-experimental designs but varied in feedback treatment, source, duration, outcome measurement, and instructional context.",
    "Only about one-third of the treatment/control comparisons examined maintenance effects, limiting conclusions about durability.",
    "Many studies did not report enough learner-level variables, and evidence was comparatively scarce for some learner groups and treatment-outcome combinations.",
    "Meta-analytic categories for surface, deep, combined, and feedback source are useful abstractions but do not establish one universally optimal revision workflow."
  ],
  "target_hack_ids": [
    "ask-a-peer-to-proofread"
  ],
  "target_protocol_ids": [
    "brali:ask-a-peer-to-proofread"
  ],
  "risk_flags": [],
  "notes": "Replace the inherited generic peer-proofreading page with a conservative educational-writing workflow: name the current revision level, request feedback at that level, revise against that target, and use a separate pass when another level also matters. Keep the evidence boundary visible and do not convert subgroup findings into a universal ranking of feedback agents."
}