Adaptive Learning¶

Strategy · All strategies
Evidence · 1 claim (1 for) · 2 studies (2 quant-synthesis), q4 · 2 of 2 report an effect size

Description¶

Adaptive learning adjusts instructional content, task difficulty, pacing, or feedback in response to ongoing evidence of each learner's performance, rather than delivering a fixed sequence to all learners. Adaptation may be rule-based (e.g., mastery thresholds) or algorithmic (e.g., Bayesian knowledge tracing, item-response-theory-driven selection), implemented by a system or by a teacher using formative data.

Design Implications¶

Adaptive systems work by keeping each learner operating at the edge of their current competence — challenging enough to induce learning, easy enough to avoid overload or discouragement Adaptive learning improves learning outcomes relative to non-adaptive instruction. [+M]. The mechanism is largely indirect: adaptation is valuable insofar as it enforces mastery, manages Cognitive Load, and delivers timely feedback — not because personalization is inherently powerful. Poorly calibrated adaptation (e.g., advancing on weak evidence of mastery) can be worse than fixed sequencing.

Context¶

Requirements¶

• A valid, sensitive measure of learner state (accuracy alone is weak; latency and confidence add signal)
• An explicit adaptation rule: what triggers remediation, advancement, or Fading of support
• A sufficiently large pool of tasks at graded difficulty levels, aligned to the target Learning Goals
• Mastery thresholds set high enough that "passed" means durable learning, not momentary luck

Constraints¶

• Adaptation based on noisy or sparse performance data misroutes learners; early errors in the learner model compound over a sequence [-M]
• Over-adaptation can narrow exposure — learners never see content the algorithm predicts they will fail, blocking productive struggle [~M]
• Effects shrink or reverse for advanced learners, who are often better served by less guidance and more challenge (the expertise-reversal pattern) [~M]
• Adaptive difficulty tuned to maintain a fixed high success rate (~90%+) can reduce effortful retrieval compared with desirable difficulties [~S]

Implementation Variability¶

• Macro-adaptation: routing between whole units or tracks based on pre-assessment
• Micro-adaptation: selecting the next item or hint level within a task (Adaptive Difficulty)
• Mastery-based adaptation: requiring criterion performance before advancement (Adaptive Mastery Learning)
• Teacher-mediated adaptation: instructors adjust instruction using dashboard data rather than automating decisions

Target Learners¶

• Novices and lower-prior-knowledge learners, who benefit most from calibrated task difficulty and immediate remediation Adaptive learning improves learning outcomes relative to non-adaptive instruction. [+M]
• Learners in large heterogeneous groups where a single fixed pace fits few students
• Less beneficial for experts or fast learners, for whom adaptation often adds redundancy and slows progress [~M]

Target Learning Goals¶

• Procedural fluency and automaticity in well-structured domains (mathematics, language learning, coding)
• Mastery of hierarchical prerequisite knowledge
• Less suited to ill-structured goals — argumentation, design, collaboration — where "correct performance" is hard to model

Instructions¶

1. Define the knowledge/skill model: decompose the domain into prerequisites and map task difficulty levels.
2. Establish a baseline through diagnostic assessment or embedded early items.
3. Set adaptation rules: mastery thresholds for advancement, remediation triggers, and support-fading conditions (Adaptive Mastery Learning).
4. Deliver tasks with immediate, actionable Feedback; adapt difficulty to keep learners in a productive challenge band (Adaptive Difficulty).
5. Review system decisions periodically — audit for learners stuck in remediation loops or advancing without mastery — and override the algorithm where the model is wrong.

Related Strategies¶

• Mastery Learning — the instructional logic most adaptive systems automate; adaptation without a mastery criterion is mere pacing variation
• Spaced Retrieval — adaptive scheduling of review items (e.g., expanding intervals) is a common and well-supported adaptation dimension
• Formative Assessment — supplies the performance evidence on which any adaptation depends
• Use ongoing assessments to tailor early-childhood learning experiences
• Profile-driven adaptive strategy rules composed dynamically and injected into the system prompt each turn
• Use strategy-based evidence models to trigger adaptive scaffolding toward more productive problem-solving strategies

Examples¶

• ASSISTments — free math platform that adapts problem selection and scaffolding based on student responses; evaluated in randomized controlled trials in Maine middle schools.
• Khan Academy — mastery-based progression where exercises unlock as learners demonstrate proficiency, with hints and remedial tasks triggered by errors.
• Duolingo — adaptive item scheduling and difficulty adjustment using learner error and response-latency data.
• ALEKS — knowledge-space-theory system that maps what a learner is ready to learn and restricts topic choice accordingly.

Key Sources¶

• Kulik, C.-L. C., Kulik, J. A., & Bangert-Drowns, R. L. (1990). Effectiveness of mastery learning programs: A meta-analysis. Review of Educational Research, 60(2), 265–299. doi:10.3102/00346543060002265
• VanLehn, K. (2011). The relative effectiveness of human tutoring, intelligent tutoring systems, and other tutoring systems. Educational Psychologist, 46(4), 197–221. doi:10.1080/00461520.2011.611369
• Pane, J. F., Steiner, E. D., Baird, M. D., Hamilton, L. S., & Pane, J. D. (2017). Informing Progress: Insights on Personalized Learning Implementation and Effects. RAND Corporation. doi:10.7249/RR2042
• Corbett, A. T., & Anderson, J. R. (1994). Knowledge tracing: Modeling the acquisition of procedural knowledge. User Modeling and User-Adapted Interaction, 4(4), 253–278. doi:10.1007/BF01099821