Where it is used
Each entry identifies the volume, edition, chapter or appendix in which the work appears.
The evidence behind the series
Follow the research.
Explore every source.
Browse the scientific research and professional literature used across all seven volumes of The Science of Learning for Educators.
7book volumes
8author ranges
1searchable library
Search the collection
Find a paper, author or idea.
Choose an author range, search a title or concept, or narrow the collection by source type. Open any result to see where it appears in the books, read its evidence summary and follow available source links.
44 references
-
- Vol 2 TE: App B
- Vol 4 RE: Chapter 10
historical thinkingsource evaluationcontextualisationevidence corroborationprimary sourcesdisciplinary reasoninghistorical perspective takingpresentismBook Evidence summaryAt a glance
Wineburg argues that historical thinking is “unnatural” because it requires readers to suspend familiar present-day assumptions and interrogate evidence in ways that everyday reading rarely demands. Drawing on studies comparing historians, students, and teachers, the book shows that expertise is not simply possession of more facts.
Key ideas
Skilled readers source a document before accepting its claims, reconstruct the context in which it was produced, corroborate it against other evidence, and notice tensions rather than smoothing them away. Background knowledge remains essential because it makes contextual questions and anomalies visible, but knowledge and disciplinary reasoning work together. Historical understanding also requires disciplined perspective taking: recognising the strangeness of the past while resisting both easy presentism and uncritical empathy. The chapters do not offer a single classroom programme or a simple causal test.
Why it matters
They use cognitive and educational research to challenge textbook recitation and to make historians’ interpretive practices teachable. For educators, the implication is to organise learning around meaningful questions, contrasting primary sources, explicit modelling of reasoning, and discussion in which claims must be warranted by evidence.
View or buy the cited work >>Read a similar article free online >>
-
- Vol 4 RE: Chapter 12
self-regulated learningmetacognitioncognitive regulationwinne–hadwin modelcopes frameworkmetacognitive monitoringstrategy adaptationlearning analyticsBook chapter Evidence summaryAt a glance
Winne describes self-regulated learning as a fusion of cognition, metacognition, motivation, and action rather than a separate study skill. The Winne–Hadwin model represents regulation through four recursively and flexibly sequenced phases: defining the task, setting goals and planning, enacting tactics and strategies, and adapting future work.
Key ideas
COPES identifies the conditions, operations, products, evaluations, and standards involved as learners interpret tasks and compare emerging products with goals. Metacognitive monitoring generates information about fit; control changes operations, effort, goals, or conditions. Because these events are conditional and unfold over time, retrospective questionnaires provide an incomplete picture. Winne argues for evidence from multiple channels and fine-grained traces of what learners do, particularly to study dynamic motivation and links between monitoring and control.
Why it matters
A practical implication is to help learners act as learning scientists: form hypotheses about strategies, collect evidence from their own performance, and adapt methods rather than merely follow decontextualized advice.
View or buy the cited work >>Read a similar article free online >>
-
- Vol 4 RE: Chapters 2, 10
self-regulated studyingtask definitiongoal settingstudy planningcognitive tacticscopes architecturemetacognitive reflectionstrategy adaptationBook chapter Evidence summaryAt a glance
Winne and Hadwin model studying as recursively self-regulated activity rather than a fixed sequence of skills. Four flexible phases organise the model: defining the task, setting goals and planning, enacting tactics and strategies, and adapting studying through metacognitive reflection.
Key ideas
Within each phase, learners interpret conditions, apply cognitive operations, produce outcomes, compare products with standards, and respond to discrepancies—the COPES architecture. Because learners construct task definitions from instructions, knowledge, beliefs, time, and resources, their version of an assignment may differ from the teacher’s. Strategies are conditional combinations of tactics, not universally effective behaviours, and monitoring can alter goals, plans, or understanding at any point. The chapter is a theoretical synthesis, not a trial of one study routine. Support should diagnose where regulation breaks down: task interpretation, standards, planning, tactic knowledge, monitoring accuracy, or adaptation.
Why it matters
Educators can model task definition, make criteria explicit, ask learners to record plans and actions, compare outcomes with standards, and revise methods. Assessment should capture events and changing conditions, not rely solely on retrospective reports of typical study habits.
-
- Vol 4 RE: Chapter 11
self-regulated learning measurementaptitude measuresevent measuresthink-aloud protocolsbehavioural tracesretrospective self-reportmeasurement validitymetacognitive assessmentBook chapter Evidence summaryAt a glance
Winne and Perry examine measurement of self-regulated learning and distinguish two conceptions. Aptitude measures treat regulation as a relatively enduring characteristic and commonly use questionnaires, interviews, or teacher ratings.
Key ideas
Event measures examine regulation as it unfolds in a task through think-aloud protocols, observations, error-detection methods, work products, or behavioural traces. The chapter relates these approaches to metacognition, motivation, monitoring, and strategic action, then analyses what each protocol samples, permitted inferences, and how measurement may affect activity. Retrospective self-report is efficient but depends on memory, interpretation, and aggregation across situations; event data are closer to action but may capture only observable indicators and require demanding coding. No method reveals an entire regulatory process. The chapter is a methodological review rather than validation of one superior instrument.
Why it matters
Researchers and educators should define the process and timescale of interest, align claims with what the measure records, combine complementary sources where feasible, preserve task context, and distinguish reported beliefs about strategy use from evidence that a strategy occurred. Assessment quality depends on defensible inference, not technological detail or questionnaire popularity.
-
- Vol 4 RE: Chapter 12
feedback literacyevaluative judgementfeedback uptakedialogic feedbackpeer feedbackassessment designlearner agencyfeedback processesBook Evidence summaryAt a glance
Winstone and Carless shift feedback from information delivered by teachers to a process in which students interpret, judge, and use information to improve work. Effective feedback depends on student feedback literacy, opportunities for uptake, and assessment designs making action possible.
Key ideas
The book examines how educators can develop evaluative judgement, enable dialogue, connect internal feedback with external comments, use technology purposefully, and organise peer feedback as learning rather than editing. It foregrounds relationships: trust, tone, emotion, credibility, and agency affect whether information is engaged with. Teacher feedback literacy includes designing for uptake while managing disciplinary, institutional, time, and workload constraints. The volume synthesises theory, research, and cases; it is not one comparative trial and does not imply that more comments produce more learning.
Why it matters
Useful designs create cycles in which learners compare work with criteria and exemplars, discuss interpretations, revise, and carry insights into later tasks. Educators should evaluate feedback by the quality of resulting learner action, not only message timeliness or volume, while ensuring that peer, automated, and teacher inputs serve coherent learning purposes.
View or buy the cited work >>Read a similar article free online >>
-
- Vol 3 RE: Chapter 5
computer-supported collaborative learninglearner agencycollaborative interactionlearning analyticsadaptive supportresearch paradigmssocial media learningeducational changeJournal article Evidence summaryAt a glance
Wise and Schwarz consult the computer-supported collaborative learning community to provoke debate about the field’s identity and future. Eight paired dialogues present opposing positions rather than recommendations.
Key ideas
They question whether CSCL needs one unifying framework; whether learner agency should take priority over scripted collaboration; and how rigorously researchers should decide that collaboration or community exists. Further provocations ask whether analytic and interpretive traditions should be reconciled, whether computational approaches deserve greater emphasis, whether learning analytics and adaptive support should become priorities, and whether the field must engage social media and learning at scale. The final exchange challenges the ambition to produce educational-system change. Each “provocateur” is answered by a “conciliator” defending achievements and complexities in existing traditions.
Why it matters
The article arises from interviews and interactive presentations, not an empirical test, and intentionally offers no resolution. Its value is diagnostic: make assumptions about collaboration explicit, examine agency and power, justify units and methods of analysis, connect small-group theory with new scales, and decide how scholarly identity relates to educational impact.
-
- Vol 3 RE: Chapter 1
- Vol 4 RE: Chapters 7, 14
scaffoldingtutoringproblem solvingcontingent supporttask simplificationguided participationfrustration managementgradual release of responsibilityJournal article Evidence summaryAt a glance
Wood, Bruner, and Ross analyse tutoring as preschool children assemble a three-dimensional wooden structure beyond unaided capability. They introduce “scaffolding” to describe how a tutor enables problem solving by controlling aspects of the task while leaving the learner to act on manageable components.
Key ideas
Observing tutor–child interaction, they identify six functions: recruiting interest, reducing the task’s degrees of freedom, maintaining direction, marking critical features, controlling frustration, and demonstrating or modelling solutions. Effective support is contingent; it depends on diagnosing the learner’s current interpretation and performance, intervening enough to sustain productive action, and adjusting as competence changes. Demonstration should preserve recognisable aspects of the learner’s attempt rather than replace it with an unrelated answer. The study is a small, task-specific experiment, not evidence that any prompting sequence works across subjects or ages.
Why it matters
Its enduring educational implication is responsive assistance: establish a shared goal, simplify without removing the intellectual problem, direct attention to consequential differences, protect persistence, and progressively return control. Scaffolding describes an interactional process between learner, tutor, and task, not merely a worksheet, hint, or permanent support.
-
- Vol 3 RE: Chapters 5, 15, 16, App A
- Vol 7 RE: Chapters 6, 15, App D
collective intelligencesocial sensitivityconversational turn-takinggroup performancegroup coordinationcollective problem solvingteam compositiongroup-level cognitionJournal article Evidence summaryAt a glance
Woolley and colleagues ask whether groups display a collective-intelligence factor analogous to the individual factor found across cognitive tasks. In two studies, 699 participants worked in groups of two to five on tasks involving brainstorming, judgement, planning, problem solving, and coordination.
Key ideas
Performance correlations supported a single factor, labelled c, that predicted performance across tasks and on a criterion task. Collective intelligence was weakly related to the group’s average or highest individual intelligence. It was more strongly associated with average social sensitivity, a more equal distribution of conversational turn-taking, and the proportion of women; the gender association was largely statistically mediated by social sensitivity. These are correlational group-level findings, not proof that adding women or enforcing equal speaking time causes effectiveness, and the laboratory tasks and short-lived groups limit generalisation.
Why it matters
For education and teamwork, the study shifts attention from assembling individually able members to interaction quality: groups may benefit when participants accurately read one another, avoid domination, distribute contribution, and coordinate across different kinds of work. Collective performance should be measured across multiple tasks rather than inferred from one outcome.
-
- Vol 2 RE: Chapter 14
digital technologydeep learninghigher-order thinkingknowledge transferblended learninginstructional guidancecollaborative learningimplementation heterogeneityJournal article Evidence summaryAt a glance
Wu meta-analyses 60 experimental or quasi-experimental studies examining digital technology and educational “deep learning,” meaning understanding, higher-order thinking, and transfer rather than machine learning. The studies span 2007–2023, multiple levels and subjects, and 13,187 participants.
Key ideas
The pooled standardised mean difference favours technology-supported conditions (SMD = 0.68, 95% CI 0.45–0.92), but heterogeneity is extremely high (I² = 99%), so the average does not predict every implementation. Moderator analyses report stronger outcomes when digital work is blended with offline activity, accompanied by guidance, used systematically rather than fragmentarily, and connected with collaborative learning. Tool function, duration, educational level, and several design variables were not significant moderators. Because categories and measures of deep learning vary, subgroup findings are observational across studies rather than causal comparisons among implementation features. The practical message is not that technology itself produces depth.
Why it matters
Educators should align tools with substantive goals, provide guidance, integrate use coherently with teaching and interaction, and evaluate understanding and transfer. Decisions should consider study quality, context, and the wide variability hidden by the pooled effect.
-
- Vol 4 RE: Chapters 12, 25, 27, App C
artificial intelligence chatbotsself-regulated learningadaptive guidanceco-regulationcritical reflectionlearner agencyemancipatory learninglearning analyticsJournal article Evidence summaryAt a glance
Wu and colleagues systematically review 78 studies published from 2018 to 2025 to ask how artificial-intelligence chatbots can be designed to support self-regulated learning. They organize design possibilities through Habermas’s technical, practical, and emancipatory knowledge interests.
Key ideas
Technical features can help learners set goals, obtain adaptive guidance, monitor progress, and receive timely feedback. Practical features emphasize dialogue, social interaction, shared understanding, and co-regulation rather than solitary tool use. Emancipatory features encourage learners to question assumptions, reflect critically, recognize power and bias, and retain agency over decisions. The resulting framework maps chatbot functions onto phases of self-regulation and shows that the literature concentrates more heavily on performance support than on deeper reflection or learner liberation. The review identifies gaps in K–12 settings, multimodal learning analytics, long-term evaluation, and designs that genuinely support emancipatory reflection.
Why it matters
Because the included studies vary in methods and outcomes, the synthesis is a design base rather than proof that any chatbot feature reliably improves achievement. Educators should align tools with learning purposes, provide human guidance, and evaluate both benefits and unintended dependency.
-
- Vol 4 RE: Chapter 25
generative artificial intelligenceself-regulated learningforethought processesperformance processesself-reflectionpersonalized goalsstrategy generationhuman–machine collaborationJournal article Evidence summaryAt a glance
Xia, Liu, Tlili, and Chiu systematically review how generative AI can support self-regulated learning activities.
Key ideas
They analyze seventy-three articles and map GenAI use across forethought, performance, and self-reflection. Six pedagogical affordances emerge: creating personalized goals; searching, analyzing, and integrating resources; monitoring and evaluating progress; recommending strategies; recording progress and providing feedback; and generating new ideas or examples. Common activities include information searching during forethought, strategy generation for problem solving during performance, and feedback and self-assessment during reflection. The review also separates individual influences, such as prior knowledge and motivation, from environmental influences, including teacher support and task design.
Why it matters
Its central design message is that GenAI should function as a human–machine collaborative scaffold for regulation, not simply an answer generator, and that activities must preserve learner monitoring, judgement, and agency.
-
- Vol 5 RE: Chapters 15, App C
generative artificial intelligencehigher-education assessmentautomated feedbackself-assessmentassessment literacyacademic integritystudent agencyinstitutional policyJournal article Evidence summaryAt a glance
Xia and colleagues scope empirical research on how generative AI is changing higher-education assessment. Following PRISMA-ScR procedures, they search ERIC, Web of Science, and Scopus, screen 969 records, and include 32 studies, mostly from 2023.
Key ideas
Findings are organised across students, teachers, and institutions. At student level, the literature reports opportunities for immediate and diverse feedback, self-assessment, and perceived impartiality, alongside academic-integrity concerns. For teachers, GenAI changes assessment literacy, beliefs about human and automated judgement, and task and feedback design. Institutions face pressure to reconsider policies, staff development, and interdisciplinary provision. The authors argue that assessment should cultivate self-regulated and responsible learning rather than focus only on detecting AI use. This forward-looking scoping review maps an early evidence base; it does not estimate effects or establish that proposed reforms improve learning, and some included studies discuss GenAI generally.
Why it matters
Practical implications include developing assessment, AI, and digital literacy; designing holistic tasks with student agency; clarifying ethical expectations; and aligning institutional policy, teacher capability, and educational purpose. Rapid technological change means these patterns require continuing empirical evaluation.
-
- Vol 4 RE: Chapter 12
self-assessmentperformance criteriafeedback seekingreflective practicecyclical learningstandards interpretationstudent judgementsubsequent actionJournal article Evidence summaryAt a glance
Yan and Brown develop a cyclical model of how learners engage in self-assessment from qualitative interviews with 17 undergraduate students.
Key ideas
The process involves determining performance criteria, seeking feedback in self-directed ways, and reflecting on the evidence gathered. Reflection is therefore not an isolated final activity: it feeds the learner’s next interpretation of criteria and future action. The model helps explain why self-assessment requires more than asking students to assign themselves a mark.
Why it matters
Educators can support the cycle by making standards discussable, teaching learners how to find and interpret feedback, and structuring reflection so that it produces concrete decisions for subsequent work.
View or buy the cited work >>Read a similar article free online >>
-
- Vol 2 RE: Chapter 3
- Vol 7 RE: Chapter 5
retrieval practicetesting effectclassroom learningcorrective feedbackdelayed learningtransfer-appropriate processingretrieval opportunitiesformative assessmentJournal article Evidence summaryAt a glance
Yang and colleagues synthesise classroom evidence for retrieval practice rather than relying only on laboratory studies. The meta-analysis includes 222 independent studies and data from 48,478 learners.
Key ideas
Across comparisons, testing or quizzing produced a medium average improvement in academic achievement over restudy and other control activities (Hedges’ g = 0.499). Effects varied with the comparison strategy, match between practice and final-test formats and materials, corrective feedback, number of retrieval opportunities, timing and location, intervention duration, and research design. The authors find results compatible with additional exposure, transfer-appropriate processing, and motivational accounts; the analysis does not establish that one mechanism explains every effect. The pooled estimate also should not be read as a guaranteed effect for any quiz implementation, because included studies and conditions were heterogeneous.
Why it matters
For educators, retrieval should be low stakes and aligned with important knowledge, repeated over time, followed by corrective information, and used to guide learning rather than merely assign grades. Comparing retrieval with a credible alternative and monitoring delayed learning provide more meaningful evidence than immediate quiz performance alone.
-
- Vol 6 PE: App E
teacher leadershipprofessional learning communitiesdistributed leadershipteacher collaborationschool organisational capacityinstructional improvementleadership conditionsJournal article Evidence summaryAt a glance
York-Barr and Duke review two decades of scholarship to clarify what teacher leadership means, how it operates, and what supports it. They describe teacher leadership as a process in which teachers, individually or collectively, influence colleagues, principals, and school communities to improve teaching and learning.
Key ideas
Leadership can be formal or informal and may involve mentoring, professional development, curriculum work, decision making, inquiry, and collaboration beyond one classroom. Their synthesis links teacher leaders’ knowledge, skills, roles, and relationships with conditions such as supportive principals, time, trust, access to information, recognition, and shared responsibility. Reported benefits include teacher learning, stronger professional community, organisational capacity, and possible improvements in practice and student learning. However, the evidence was conceptually inconsistent and often descriptive, small-scale, and weak on direct outcome attribution.
Why it matters
The review provides a framework and research agenda, not proof that assigning leadership titles causes achievement gains. Schools should define purposes, select credible leaders, protect time, build trust, align authority with responsibility, and evaluate influence on practice while monitoring workload, role conflict, and resistance.
View or buy the cited work >>Read a similar article free online >>
-
- Vol 7 RE: Chapter 4
artificial intelligence applicationshigher educationprofiling and predictionadaptive learning systemsintelligent tutoring systemseducational data ethicsinterdisciplinary researchJournal article Evidence summaryAt a glance
Zawacki-Richter and colleagues systematically review artificial-intelligence research in higher education published from 2007 through 2018. From 2,656 records, they include 146 studies and classify applications into four areas: profiling and prediction, assessment and evaluation, adaptive systems and personalisation, and intelligent tutoring systems.
Key ideas
Much of the literature comes from computer science and STEM contexts, with limited participation by educators and little connection to pedagogical theory. Studies often emphasise technical performance while giving less attention to teaching practice, learner perspectives, ethics, privacy, or broader consequences. The review maps an emerging field; it does not establish that AI applications generally improve learning, and its search period predates generative AI. The authors call for interdisciplinary collaboration, theoretically informed educational questions, and critical attention to risks and benefits.
Why it matters
Institutions and researchers should involve educators and students in problem definition, evaluate educational value rather than model accuracy alone, make data and decisions accountable, and study inclusion, agency, privacy, and possible harms alongside effectiveness. Later AI developments require new evidence rather than automatic extrapolation from this corpus.
-
- Vol 5 RE: Chapter 15
generative artificial intelligencefeedback engagementfeedback literacystudent agencyfeedback processingfeedback enactmentcritical ai literacyJournal article Evidence summaryAt a glance
Zhan, Boud, Dawson, and Yan develop a conceptual framework for understanding student engagement with feedback in generative-AI environments. Drawing on feedback research and an ecological perspective, they examine three stages: eliciting feedback, processing it, and enacting it in subsequent work.
Key ideas
Generative AI can reduce barriers of time, access, and social anxiety by providing rapid, repeatable, personalized responses. Yet useful engagement depends on students’ feedback literacy: they must formulate productive requests, judge credibility and relevance, compare advice with standards, and decide what to implement. Risks include fabricated or biased information, weak prompts, passive acceptance, reduced human dialogue, and ethical concerns. The authors propose a cyclical model of feedback forethought, control, and retrospect, emphasizing interaction between the opportunities a tool affords and a learner’s capacity to recognize and use them.
Why it matters
This is a theory-building synthesis, not an experimental test of learning gains. Its practical message is that institutions should design feedback environments and teaching activities that develop judgment, agency, and critical AI literacy while preserving meaningful teacher and peer relationships.
-
- Vol 4 RE: Chapter 12
- Vol 4 RE: Chapter 6
- Vol 4 RE: Chapters 18, 25, App A, App C
artificial intelligenceself-regulated learningadaptive learning platformspersonalized guidancelearning analyticsai literacyteacher orchestrationJournal article Evidence summaryAt a glance
Zhu and Sari systematically review 25 empirical studies of artificial-intelligence applications intended to support self-regulated learning. The included tools span chatbots, intelligent tutoring systems, adaptive platforms, electronic books, and educational games, and address planning, performance, monitoring, reflection, and iteration in different ways.
Key ideas
The synthesis finds promising support for personalized guidance, feedback, engagement, and strategy use, but also recurring problems involving inaccurate outputs, limited adaptability, behavioral tracking, weak AI literacy, and possible negative effects or dependency. The authors propose an AI–SRL framework with two connected layers. The learner layer follows forethought, performance, self-reflection, and iteration; the instructor layer includes teaching AI literacy, integrating tools into purposeful activities, providing continuing support, monitoring progress, and improving designs over time. This makes teacher orchestration part of the model rather than assuming AI independently produces self-regulation.
Why it matters
The review is limited to English-language studies found in three databases and a still-small, heterogeneous evidence base. Its conclusions should guide careful design and further testing, not be read as proof that AI tools uniformly improve self-regulated learning.
View or buy the cited work >>Read a similar article free online >>
-
- Vol 4 RE: Chapters 1, 2
self-regulated learningsocial cognitive theoryself-efficacyself-observationself-judgementself-reactionreciprocal causationJournal article Evidence summaryAt a glance
Zimmerman presents academic self-regulation as a dynamic interaction among personal beliefs, behaviour, and environment rather than a fixed learner trait. Within this social-cognitive account, students regulate learning through self-observation, self-judgement, and self-reaction.
Key ideas
They monitor aspects of performance, compare them with goals or standards, and respond in ways that can sustain or alter subsequent effort and strategy use. Self-efficacy is central because beliefs about capability influence goal choice, persistence, monitoring, and reactions to results. Regulation is reciprocally caused: students shape study environments and behaviours, while feedback from those environments changes beliefs and later action. The article integrates theory and existing evidence; it is not a single classroom experiment establishing one intervention’s effect. It also avoids equating self-regulation with solitary learning, because models, feedback, rewards, and social support contribute to its development.
Why it matters
Educationally, teachers can make goals and criteria explicit, model strategic processes, support accurate monitoring, provide informative feedback, and help learners interpret errors as information for adaptation. Assessment should attend to processes and changing contexts, not infer a stable capacity solely from an outcome score or retrospective questionnaire.
-
- Vol 4 RE: Chapters 1, 2, 4, 8, 13, 25, App A, App C
- Vol 4 TE: App C
- Vol 5 RE: Chapters 2, 3, 5, 6, 13, 14, 15, App B
- Vol 5 TE: App C
- Vol 7 PE: App D
- Vol 7 RE: Chapters 2, 7, App D
self-regulated learningforethought phaseperformance phaseself-reflectiongoal settingstrategic planningcausal attributionBook chapter Evidence summaryAt a glance
Zimmerman defines self-regulation as context-sensitive processes through which people pursue goals, not as an inherited stage, general trait, or purely metacognitive skill. His social-cognitive model is cyclical.
Key ideas
Forethought includes task analysis, goal setting, strategic planning, self-efficacy, outcome expectations, interest, and goal orientation. During performance, learners deploy strategies, manage attention, use self-instruction or imagery, structure environments, and observe their actions and conditions. Self-reflection then combines self-evaluation and causal attribution with satisfaction and adaptive or defensive reactions. These reactions feed the next cycle, so interpretations of success and failure influence later goals, effort, and strategy choice. The chapter also distinguishes proactive regulation, in which learners set goals and create favourable conditions, from reactive attempts made only after difficulty occurs. Social models, instruction, feedback, and guided practice contribute to development. This is a theoretical synthesis of research, not one intervention trial.
Why it matters
For teaching, it supports modelling strategies, helping students set specific proximal goals, making performance information visible, prompting defensible attributions, and revising approaches after reflection. Difficulties should be diagnosed across the cycle rather than attributed to motivation or ability alone.
View or buy the cited work >>Read a similar article free online >>
A transparent research trail
More than a bibliography.
What it contributes
Evidence summaries explain the central idea and why the source matters to educators.
How to access it
Entries link to lawful full text where available, or to a publisher, library or purchase page.
Edition abbreviations
- TE
- Teacher Edition
- PE
- Practice Edition
- RE
- Research Edition
Citation checks
References are checked against publisher records, scholarly indexes, DOI or ISBN metadata, repositories and author records where available. An entry marked [Citation not verified] preserves the wording found in the relevant volume without attributing findings to an unconfirmed work.
From sources to a coherent pathway