John Hattie

Record year2007

summaryOrganizes into a testable model why feedback may be positive or negative, what questions information should answer, and at which cognitive level it operates.

M rationaleThe 2007 original abstract explicitly states that feedback may be positive or negative, differs by type and delivery, and proposes a named model; a 2020 three-author review also shows substantial heterogeneity, supporting the model's conditional value rather than a universal prescription. It is rated M5.4 as a reusable educational explanatory model. The Root adjudication retains M: this is the end-state of conditional knowledge, a model, or an index, not a claim that instructional interventions are universally effective.

attribution rationaleHattie 0.25, Timperley 0.25, feedback-research prehistory 0.18, original research/classroom-data teams 0.13, institution 0.05, unresolved 0.14. This final version does not change a: this revision has no new auditable attribution facts.

overlap boundaryThe 2007 joint paper reviews concepts and evidence on feedback and proposes the three questions of goals, progress, and next steps, plus a four-level model of task, process, self-regulation, and self; it does not treat “feedback” as a single uniform intervention or claim teachers' implementation, student outcomes, or all value of the 2020 re-meta-analysis.

Evidence gapsThe two authors' division of modeling work and literature selection are not confirmed by research archives. Independent replication of different feedback levels across ages, subjects, and outcome measures still requires item-by-item checking.

Record year2008

summaryOrganizes many dispersed educational meta-analyses into a navigable map of influences and exposes citations, individual effects, and evidence-scale information in MetaX, reducing cross-topic retrieval costs.

M rationaleThe book and MetaX do provide large-scale synthesis, continuing updates, and citation drill-down as a knowledge service, with clear cross-topic navigation value; however, ranking averaged effects across heterogeneous designs and constructs, and interpreting 0.4 as a year's growth, have direct statistical and causal boundaries. It is rated M5.5 as a “useful but not a uniform prescription” synthesis–index family; controversy enters the achievement's net boundary rather than appearing only as a gap. The Root adjudication retains M: this is the end-state of conditional knowledge, a model, or an index, not a claim that instructional interventions are universally effective.

attribution rationaleHattie 0.19, Visible Learning/MetaX team 0.16, meta-analysis authors 0.21, original-research/school-data teams 0.14, methodological prehistory 0.14, institution 0.05, unresolved 0.11. This final version does not change a: this revision has no new auditable attribution facts.

overlap boundaryPublished in 2008, Visible Learning organized more than 800 meta-analyses by influence factors and effect sizes; subsequent MetaX updated the same synthesis chain into an index showing meta-analysis citations, individual effect sizes, and sample/study counts. MetaX receives credit only for the additional work on traceability, maintenance, and updating, and does not create a second complete M. 0.4 is only a project average-effect marker, not a universal causal threshold, guarantee of annual-learning conversion, or uniform classroom prescription. This revision ends at 2025-12-31: historical IDs are retained for traceability, but 2026 new material, database updates, or effects are not counted.

Evidence gapsMetaX inclusion and exclusion rules, version differences, overlapping meta-analyses, and treatment of dependency effects have not undergone an independent whole-corpus audit.; The cross-scale conversion from 0.4 to ‘one year’s growth’ and comparability among factors require methodological review.; Commercial training, school adoption, and learning outcomes are not accounted for and cannot be inferred from the existence of an index.; The task cutoff is 2025-12-31; additions in 2026 are explicitly excluded.

Record year1982

summaryPlaced highly dispersed and inconsistently defined self-concept research within a multidimensional hierarchical model and quantitative synthesis, clarifying that measurement, correlation, and intervention evidence must be assessed separately.

M rationaleThe 1982 joint synthesis, the 1985 joint model paper, and the 1992 monograph form a continuous research terminal state independent of Visible Learning; M5.0 is assigned for their bounded contribution to educational psychometrics and synthesis methods, without inflating correlations or later applications. Root adjudication retains M: this is a conditional knowledge, model, or indexing terminal state, not a claim that teaching interventions are generally effective.

attribution rationaleHattie 0.24, Hansford 0.16, Song 0.07, original-study and data teams 0.14, theoretical/methodological prehistory 0.20, institutions 0.05, unresolved 0.14. This final version does not change a: this revision provides no new auditable attribution facts.

overlap boundaryThe 1982 synthesis with Hansford of self-concept and academic/performance relationships, the 1985 test with Song of a hierarchical multidimensional model, and the 1992 monograph’s synthesis of dimensions, measurement, correlational relationships, and interventions; it does not state correlation as self-concept causing achievement, and does not claim later self-efficacy or social-emotional-learning outcomes.

Evidence gapsThe studies included in 1982, effect-size conversions, and publication bias have not been recomputed item by item.; The specific conceptual consumption relationship between the 1992 monograph and the joint papers still requires chapter-by-chapter review.