Skip to content

Case-Based Learning

Pattern · All patterns
Evidence · 45 claims (19 for, 25 mixed, 1 against) · 33 studies (13 causal, 6 quant-synthesis, 6 review, 3 design, 2 associational, 2 qualitative, 1 theoretical), q2–q4 · 7 of 33 report an effect size · 40 claims rest on one study

Description and scope

A reusable policy for organizing a lesson around realistic cases (scenarios, incidents, dilemmas or worked situations) that learners analyse, and for deciding how many cases to use, how learners work with them, and when the principle they share is made explicit. It applies the case-based learning principle, which has not yet been rewritten as a conditional model, and draws on analogical reasoning and cognitive flexibility. The study configurations below are evidence; the response-dependent policy is an untested design proposal. Use it to gather better evidence about a learner group, not to certify that a learner "can reason from cases".

What the evidence does and does not test. The wiki's strongest evidence concerns one component, comparing cases (case-comparison), against studying the same cases one at a time, single cases, or conventional instruction. Case-based learning as a whole format has one non-randomized classroom comparison and a systematic review that found its effect on learning inconclusive. So this page treats comparison, an explicit principle and a transfer probe as the configuration, and treats "use cases" on its own as an unsupported promise.

Problem-based learning is a neighbouring, converted pattern: it organizes a unit around one problem learners must investigate and decides how much guidance to give before or after it. This pattern is narrower and works at lesson grain: the case is an object of analysis and comparison, not a driving problem, and its decisions concern case selection, comparison and the timing of the principle. Where a case is used as a problem to solve before instruction, use the problem-based pattern's guidance on problem-first phases. The Harvard-method page describes one discussion-centred configuration of the format.

Cases on screen, online and with AI (added 2026-10-17). A case may be read and discussed in an online course, worked alone and asynchronously, or made interactive: a branching case or virtual patient that responds to the learner's decisions, a case written or varied by a generative AI, or a case in which an AI plays the client, patient or colleague. The analysis-and-comparison policy below applies to all of them. Where the point is to rehearse a conversation with the person in the case, use Practice with Simulated People; where an AI responds to a learner's analysis turn by turn, AI Tutoring decides what it may give away; where a learner works through cases alone with a mentor, Mentor Support in Self-Paced Online Learning decides when the mentor steps in.

Inputs: establish the design brief

Field Record or elicit
Learner and starting observation Age band and task-specific starting knowledge (novice, intermediate, advanced, or ?); a brief independent analysis of one case: what the learner thinks matters in it, what principle (if any) they name, and what they would do. Preserve unknowns.
Context and activities Setting (classroom, online-instructor-led, workplace-clinical, ...), time available, group or individual work, who facilitates, and exactly which cases, prompts, explanations and feedback are planned and in what order.
Learner-valued goal Ask what the learner wants to be able to do with cases like these and why (for example, decide in their own practice, argue a position, pass an exam). Record disagreement with the designer objective; negotiate rather than assume agreement.
Designer objective The principle, concept or judgement (principle, concept, complex-skill) the cases are meant to carry, and whether the domain is well structured (one principle applies the same way across cases) or ill structured (the concept plays out differently case by case).
Outcome Separate instruments for factual recall, use of the principle in a new case (near-transfer, far-transfer) and, where it matters, a decision or product; individual as well as group measures; immediate and delayed horizons.

A request for "more realistic, engaging teaching" does not specify these inputs. Ask for them before choosing cases, a number of cases or a predicted gain. Check first that each case actually instantiates the target principle and that a correct analysis of it requires the principle, not only common sense or reading the case.

Sequence and conditional policy

  1. Elicit briefly. Give one case and ask each learner (not only the group) what matters in it and why. Keep this short and record it as a learning exposure.
  2. Select at least two cases that share the target principle. Choose cases whose surface details differ while the principle stays the same; for an ill-structured target, add cases in which the concept plays out differently. Record how the cases differ and on which dimension they are meant to contrast.
  3. Have learners compare, not just read. Present the cases together and prompt for what is the same, what differs and why. Studying the same cases one after another is a different configuration with weaker recorded results; record which was delivered.
  4. Make the principle explicit after the comparison. State the principle, connected to what learners found in the cases; correct a principle that learners have abstracted wrongly.
  5. Reobserve individually on a new case. Use a case with new surface details, scored separately for recall and for use of the principle, at the agreed horizon. If the response disagrees with the expectation, revise the interpretation or the configuration.
Observation Discriminating probe Proposed action, with its uncertainty
Learner retells the case but names no principle Ask what would change the outcome of the case; give a second case and ask what the two have in common. If the comparison releases a principle, continue with comparison. If not, the learner may lack the concept the cases presuppose: teach it explicitly or use a worked example before further cases. Neither branch is a validated diagnosis.
Learner names a principle, but a surface feature of the case Give a case with the same surface feature and a different principle, and one with different surface and the same principle. Choose cases that vary the misleading surface feature and hold the principle constant, and prompt for the contrast. The comparison evidence does not establish how many cases are enough.
Lively discussion, but individual answers on a new case do not use the principle Individual analysis of a changed case, matched in demand to the discussion. Add an individual comparison task and an explicit statement of the principle; record who contributed what. A review of the format notes that any benefit may come from group work rather than from the cases, so discussion alone does not show learning.
Principle used on the new case, factual recall weak Score recall separately under the same conditions. Add targeted recall practice if recall is part of the objective. In one small experiment, multiple-case study improved transfer essays while a drill group recalled more facts, so this pattern is expected rather than a failure.
Correct on an immediate new case, not at a delay A matched new case after the agreed interval. Plan a later comparison or retrieval of the principle. The comparison meta-analysis found larger benefits on immediate than on delayed tests, so an immediate success does not forecast a delayed one.

Choosing configurations from evidence

  • Whether to expect anything from the format itself. Case-based learning improves exam performance [~M] rests on one non-randomized within-cohort comparison in an introductory community-college biology course (56 students; case-taught topics scored higher on regular exams than matched topics taught by short lecture, reading and discussion, an 18-percentage-point average difference, no standardized effect size) and on a systematic review of 104 papers in health professional education that found the evidence inconclusive on whether case-based learning improves learning more than other activities, noting that gains may come from the group work. Plan and evaluate a specific configuration; do not promise a gain from the label. Neither entry reports a delayed outcome.
  • Compare cases rather than study them one at a time. Comparing contrasting cases improves learning [+M]: across 57 experiments (336 tests) case comparison produced greater learning than single, sequential or non-analogous case study, traditional instruction and controls (d = .50, 95% CI [.44, .56]); in a randomized experiment with undergraduate novices learning negotiation, 48% of those who compared two cases on one page used the principle in a new negotiation against 19% of those who studied the same cases separately. The pooled estimate mixes several comparison conditions, so it is not solely a "compare vs. same cases separately" effect. The same evidence underlies multiple contrasting cases support abstraction [+M].
  • When the principle is stated, and on which outcome to judge. The same meta-analysis found larger benefits when learners were asked to find similarities, when the principle was given after the comparison, with perceptual content, and on immediate tests. That supports step 4's order and the delayed probe in the table; it does not set a lag or a dose. In seventh-grade algebra, comparing solution methods side by side raised procedural knowledge and flexibility but not conceptual knowledge (abstract only), so a comparison may change how learners use and choose methods without changing their grasp of the concept.
  • Multiple perspectives in ill-structured domains. Presenting multiple cases from different perspectives supports transfer in ill-structured domains [~M]: in a small randomized experiment (34 paid university volunteers, the social impact of technology), revisiting short cases under several themes produced better transfer essays than computer drill, while drill produced better factual recall; the treatment bundled several features, so which one mattered is not known. The comparison meta-analysis does not isolate ill-structured domains or multiple perspectives, and its smaller delayed benefit cuts against the theory's prediction that the advantage grows at delay. Choose this configuration for an ill-structured target judged on transfer, and measure recall separately.
  • Novices. No claim recorded here tests case-based learning by prior knowledge. The claim pages' discussions propose, without evidence entries, that novices may need the concept supplied or comparison scaffolded; the comparison experiments themselves used novices. Treat expertise as an open moderator and elicit the starting response rather than assume. The nearest evidence since this was written: in simulation-based learning in higher education, examples were associated with larger effects for learners at lower education levels and reflection phases for those at higher levels (Prior knowledge moderates scaffolding effectiveness [~M]), and in one experiment novices who learned the causal mechanism of artificial diseases diagnosed difficult cases with irrelevant findings better than those who learned feature lists (Biomedical causal knowledge improves novice diagnostic accuracy on difficult cases with irrelevant findings [+M]), which supports supplying the concept before novices work difficult cases. Neither tests case comparison by prior knowledge.

Evidence added 2026-10-17: cases online, interactive, and with AI.

Do not rank these results against one another: their comparators (separate cases, drill, lecture and reading, other activities) and outcomes (exam items, transfer essays, a negotiation, procedural knowledge) differ. A precise abstention names the missing comparison, outcome or observation and the next way to resolve it.

Fitting the design to a situation

Ask for four facts when the brief omits them: is the domain well structured or ill structured?, is the outcome recall, use of a principle on a new case, or a procedure?, is the test immediate or delayed?, and will learners work in groups or alone?

If the brief says… Then change… Basis
The goal is applying a principle to new cases Present two or more cases that share the principle side by side and prompt for what is the same and why, rather than one case at a time Comparing cases side by side improves learning and transfer [+M]; Analogical reasoning improves transfer [+M]
The plan states the principle first and then shows cases as illustrations Reverse the order: have learners compare first and state the principle after, connected to what they found Analogical reasoning improves transfer [+M] (moderator in the meta-analysis)
The domain is ill structured (the concept plays out differently case by case) Revisit several short cases under different themes; expect better transfer essays, and plan separate work for factual recall, which drill did better Multiple cases from different perspectives support transfer in ill-structured domains [~M]
The goal is understanding why a mathematical method works, not only choosing a method After comparing solution methods, add an explicit explanation of why each works; comparison raised procedural knowledge and flexibility but not conceptual knowledge Comparing cases: procedural gain, no conceptual gain in one algebra study [~M]
The test that matters is weeks later Add a later comparison or retrieval of the principle; benefits were larger on immediate than delayed tests Analogical reasoning improves transfer [~M]
The course is in the health professions and relies on small-group case discussion Do not expect the case format alone to raise learning; a review judged the evidence inconclusive and said any benefit may come from the group work. Add individual analysis of a new case Case-based learning improves exam performance [~M]
Learners work alone (online-self-paced) Put both cases on one screen with a written prompt to describe their similarities Comparing cases side by side improves learning and transfer [+M] (undergraduates comparing two cases on one page)
Cases are complex and supported by AI tools or team workflows (medical education) Keep a faculty member guiding how the tool is used in each case; complex cases and team workflows showed null or unfavourable effects Case-based learning benefits depended on faculty-guided embedding [~W]
Learners lack the concept the cases presuppose Teach the concept or give a worked example before the cases, then compare no claim here tests case-based learning by prior knowledge; nearest: Biomedical causal knowledge improves novice diagnostic accuracy on difficult cases [+M] and Prior knowledge moderates scaffolding effectiveness [~M] (examples for lower levels, reflection for higher)
Many learners discuss one case in a large group Add an individual comparison task on a changed case before the group discussion ends untested proposal
Learners work through cases self-paced online (online-self-paced), with a mentor available on demand Put the cases side by side with a written comparison prompt; require a short individual written analysis of each new case; have the mentor read a sample and return the principle statement with feedback; add an asynchronous peer exchange on each case; recap the principle in writing; use cases from the learners' own workplaces where possible Comparing cases side by side improves learning and transfer [+M]; Co-regulation shows stronger relationships with the three CoI presences than self-regulation in an online case-based course [~W]; Some Chinese students felt uncomfortable with case-based learning due to lack of structured instruction content [~W]; mentor role from Mentor Support in Self-Paced Online Learning, untested for case work
An AI is the case partner: it writes or varies the cases, plays the client or patient, or questions the learner's analysis Have the AI generate surface-varied cases holding the principle constant, and check each one before use; let it play the person in an interactive case; have it ask for the principle rather than supply it; judge the learner's individual analysis of a new case without the AI; keep a person reviewing complex cases Case-based learning benefits depended on faculty-guided embedding [~W]; Controlled GenAI virtual-patient and communication studies reported short-term skill gains, but positive findings were not universal [~M]; see AI Tutoring and Practice with Simulated People
An interactive case simulation (virtual patient, branching case) is available Use it alongside the case discussion, not in place of it; expect gains on reasoning and skills more than on recall; give feedback inside the case Virtual patients blended with traditional education improve knowledge more than traditional education alone [+M]; Virtual patients show larger advantages over passive traditional learning than over active learning [~M]; Virtual patients are as effective as traditional education for knowledge outcomes [~M] (health professions)
The cases are static documents only (a written case, a record, a work sample) Build in decision points with consequences revealed in stages, or pair the document with a worked analysis for novices Live simulations with real patients showed the greatest effects among simulation types [~M] (document-based least effective, g = 0.31, higher education)
Teacher education: cases of classroom practice for observation Start with a simplified simulated classroom for reflection, then move to authentic classroom video for analysis and proposing support In the initial skill acquisition phase, authentic video yielded higher overall observation assignment scores than virtual simulation [~M]; Virtual simulation stimulated higher self-reported purposefulness than authentic video [~M] (one design study, 34 candidates)
Learners come from varied countries or workplaces (international, working professionals) Mix global and local cases, and ask learners to bring or adapt a case from their own work International students wanted more global cases and closer connections between case analyses and their local work contexts [~W]
Business or technology cases (a strategy decision, an incident review, a troubleshooting case) Keep the comparison policy: two cases sharing one principle, an individual written recommendation before discussion, the principle stated after untested in these domains in this wiki; the comparison evidence includes a negotiation task (Comparing cases side by side improves learning and transfer [+M])

The evidence by question

Further evidence, not yet read against this model

Claims this page cited before it was rewritten as a conditional model. Both claims it cited before the 2026-10-02 rewrite are now read in the model above, so none remains here. - A training-design argument, not tested by any study recorded here, holds that practising whole complex tasks improves transfer of complex skills [+W] aligns with case-based learning when the case requires integrating multiple concepts into a coherent response - Transfer is stronger when learners articulate why a response fits the case rather than merely selecting an option Prompting learners to self-explain improves understanding and problem solving on immediate tests by a moderate average amount, with little evidence yet on delayed or classroom gains. [+S] - Facilitation matters: questioning and hints often outperform simply telling learners what the right answer was Contingent scaffolding improves learning more than fixed or absent support. [+M] - Part-task practice reduces cognitive load for absolute novices during initial skill acquisition. [~M] — some novices need component instruction or worked examples before cases become productive - Studying theoretical content with text-based cases yields a small advantage in productive scientific knowledge application over studying the same content with theoretical models [+M] — attached 2026-10-18 from Baier et al. (2021), which proposed "Use text-based cases to connect scientific theory to real-world classroom situations in teacher education, but treat them as one option rather than a panacea against inert knowledge"; tests this page's relationship. - Both case-based and model-based interventions increase pre-service teachers' self-reported use of theory-based knowledge when analysing classroom situations [+M] — attached 2026-10-18 from Baier et al. (2021), which proposed "Use text-based cases to connect scientific theory to real-world classroom situations in teacher education, but treat them as one option rather than a panacea against inert knowledge"; tests this page's relationship.

Illustrative design instance and observation record

Illustration only, invented, not a tested lesson. A workplace session for new team leads on handling conflict between staff. The designer's objective is that each lead can recognise when a dispute is about a task rather than a relationship and choose a response that fits; one lead says she wants to stop avoiding difficult conversations. Agree that both a correct reading of a new case and a planned first conversation matter. Ask each lead to analyse one short case; most describe who was at fault and none distinguishes task from relationship. Give two further cases with different settings and the same underlying distinction, ask leads to compare all three in pairs, then state the distinction explicitly against what they found. Assess individually with a new case at the end of the session and again three weeks later, scored separately for reading the case and for the planned response.

Record: cases and how they differ → each learner's first analysis → comparison prompts, explanation and facilitation actually given, with timing → principles learners named → candidate interpretations and confidence basis → distinguishing probe → principle statement delivered → individual analysis of a new case at the stated horizon. Preserve the learner's stated purpose and any change in it. An agent may summarize this record but must not fabricate missing inputs or assign numerical state probabilities without a calibrated model.

Elements and limits

Case studies, case-based learning, comparing cases, discussion-based learning, self-explanation, scaffolding and worked examples.

This pattern is scoped to lessons in which cases are analysed and compared and the principle they share is made explicit. Whole case-based curricula, cases used as unguided problems before any instruction, discussion formats judged on participation, and assessment of professional competence need distinct configurations and evidence. Apart from the virtual-patient trials pooled by Kononowicz et al. (interactive cases against traditional teaching, in the health professions), the wiki records no randomized comparison of case-based learning as a whole format with another approach, and none of online or AI-run case work against face-to-face case work; the diagnostic branches above remain to be tested with actual learners.

Claims

Key Sources

  • Herreid, C. F. (2011). Case study teaching. NSTA Press.
  • Kaddoura, M. A. (2011). Critical thinking skills of nursing students in lecture-based teaching and case-based learning. International Journal for the Scholarship of Teaching and Learning, 5(2). doi:10.20429/ijsotl.2011.050220
  • Hoffer, E. R. (2020). Case-based teaching: Using stories for engagement and inclusion. International Journal on Social and Educational Sciences, 2(2), 75-80.
  • Vinney, L., Friberg, J. C., & Smyers, M. (2019). Case-based perspective-taking as a mechanism to improve metacognition and higher-level thinking. Journal of the Scholarship of Teaching and Learning, 19(3), 91-104. doi:10.14434/josotl.v19i2.24006
  • Ellet, W. (2007). The case study handbook: How to read, discuss, and write persuasively about cases. Harvard Business School Press.
  • Christensen, C. R., Garvin, D. A., & Sweet, A. (1991). Education for judgment: The artistry of discussion leadership. Harvard Business School Press.
  • Kim, S., Phillips, W. R., Pinsky, L., Brock, D., Phillips, K., & Keary, J. (2006). A conceptual framework for developing teaching cases: A review and synthesis of the literature across disciplines. Medical Education, 40(9), 867-876. https://doi.org/10.1111/j.1365-2929.2006.02544.x
  • Kononowicz, A. A., Woodham, L. A., Edelbring, S., Stathakarou, N., Davies, D., Saxena, N., Tudor Car, L., Carlstedt-Duke, J., Car, J., & Zary, N. (2019). Virtual patient simulations in health professions education: Systematic review and meta-analysis by the Digital Health Education Collaboration. Journal of Medical Internet Research. https://doi.org/10.2196/14676
  • Chernikova, O., Heitzmann, N., Stadler, M., Holzberger, D., Seidel, T., & Fischer, F. (2020). Simulation-based learning in higher education: A meta-analysis. Review of Educational Research. https://doi.org/10.3102/0034654320933544
  • Sadaf, A., Kim, S. Y., & Olesova, L. (2022). Relationship between metacognition and online Community of Inquiry in an online case-based course. Online Learning, 26(4), 79-93. https://doi.org/10.24059/olj.v26i4.3474
  • Liu, X., Liu, S., Lee, S.-h., & Magjuka, R. J. (2010). Cultural differences in online learning: International student perceptions. Educational Technology & Society, 13(3), 177-188. https://www.jstor.org/stable/jeductechsoci.13.3
  • Zhang, B., et al. (2026). A sequential integrating virtual simulation and authentic video for preschool teacher candidates' observational skills training: A design-based intervention study. PLoS One. https://doi.org/10.1371/journal.pone.0358951
Page metadata
Field Value
type pattern
id case-based-learning
aliases case-based-learning-harvard-method
title Case-Based Learning
description A reusable policy for learning from realistic cases: establish what learners can already do with the target concept, have them compare several cases that share a principle rather than study one at a time, make the principle explicit, and judge the unit on transfer to a new case rather than on enjoyment or recall.
status review
grain_size lesson
moved_from principle/case-studiescase-based-learning
canonical True
generated by claude/unspecified on 2026-10-17
trust unverified — no one has confirmed this page's content
sources 5 — kim-2006, kononowicz-2019, chernikova-2020, sadaf-2022, zhang-2026

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