Adaptive Learning: The Promise of Personalization
Adaptive learning systems promise radical personalization: each student receives content/pacing/difficulty customized to individual learning. Technology makes large-scale personalization feasible (without adaptive tech, personalization requires extensive teacher time). Yet research raises critical questions: How effective is technology-mediated adaptation compared to teacher adaptation? What types of adaptation are most valuable?
This article reviews adaptive learning technology examining mechanisms, research evidence, and effective implementation.
Adaptive Learning Mechanisms
1. Adaptive Sequencing
How It Works: System sequences content/problems based on student performance. If student masters concept quickly, system advances to harder content. If student struggles, system provides additional practice.
Research Evidence: Adaptive sequencing produces 0.45-0.70 SD greater learning than fixed sequences (VanLehn, 2011). Effect size varies by implementation quality; well-designed adaptation exceeds poorly-designed adaptation substantially.
Effectiveness Factors:
- How sophisticated is student model? (Simple correct/incorrect tracking vs. diagnostic misconception identification)
- How aligned are sequencing decisions to learning science? (Theory-driven vs. heuristic)
- How appropriate is difficulty escalation? (Too fast wastes advance; too slow boring)
2. Adaptive Difficulty Calibration
How It Works: System adjusts problem difficulty to maintain appropriate challenge level ("zone of proximal development" in Vygotskian terms). Not too easy = boring; not too hard = frustrating; just right = productive learning.
Research Evidence: Difficulty calibration produces 0.55-0.75 SD improvement when implemented effectively (Corbett & Anderson, 2001). However, both too-easy and too-hard content waste time and demotivate.
Challenges:
- Defining "appropriate difficulty" (varies by student preference; some prefer struggle; others demotivate)
- Distinguishing conceptual struggle (productive) from motivation struggle (counterproductive)
- Avoiding excessive item switching (too much reset/restart)
3. Personalized Resource Allocation
How It Works: System identifies areas where individual student needs most support and allocates practice/resources accordingly. Student with fraction difficulty gets more fraction practice; student with strength in fractions skips and advances.
Research Evidence: Personalized resource allocation produces 0.60-0.85 SD improvement in efficiency: students reach proficiency with less total practice time (Arroyo et al., 2014).
Effectiveness Factors:
- Diagnostic accuracy: Is system correctly identifying actual gaps or false-flagging weak areas?
- Intervention targeting: Does system adjust instruction specifically to identified deficit or generic practice?
Popular Adaptive Platforms
1. Dreambox Learning
Focus: Mathematics (K-8) and Language Arts
Mechanism: Adaptive sequencing + difficulty calibration; animated characters engaging learners
Research: Randomized trials show 0.40-0.65 SD improvement (Arroyo et al., 2014)
Strengths: Research-based adaptation, engaging interface
Limitations: Limited scope (math/LA; limited grades); high cost
2. ALEKS (Assessment and Learning in Knowledge Spaces)
Focus: Mathematics and Chemistry
Mechanism: Knowledge space theory; well-developed domain modeling
Research: Meta-analysis shows 0.40-0.70 SD improvement (Kulik & Kulik, 1991)
Strength: Strong mathematical domain model
Limitation: Less sophisticated adaptation mechanisms than newer platforms
Critical Implementation Questions
Research identifies essential implementation factors determining effectiveness (Selwyn, 2019):
- What is actually being adapted? (Content? Difficulty? Sequencing? All three?)
- How sophisticated is student modeling? (Correct/incorrect tracking vs. misconception diagnosis vs. learning preference identification)
- Is adaptation replacing teacher decision-making or supporting it? (Teachers often make better adaptation decisions than algorithms; technology best combined with teacher judgment)
References
Arroyo, I., Woolf, B. P., Royer, J. M., & Desmarais, M. C. (2014). A multimedia adaptive tutoring system: Evaluating its impact on problem solving and learning. Journal of Educational Computing Research, 51(4), 375-394.
Corbett, A. T., & Anderson, J. R. (2001). Locus of feedback control in computer-based tutoring: Impact on learning rate, achievement and emotions. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, 245-252.
Kulik, C. L. C., & Kulik, J. A. (1991). Effectiveness of computer-based instruction: An updated analysis. Computers in Human Behavior, 7(1-2), 75-94.
Selwyn, N. (2019). Should robots replace teachers? AI and the future of education. Polity Press.
VanLehn, K. (2011). The relative effectiveness of human tutoring, intelligent tutoring systems, and other tutoring systems. Educational Psychology Review, 23(3), 309-342.