Enhancing testing and assessment in the digital age with computerized adaptive testing

Computerized Adaptive Testing (CAT) is a highly efficient and flexible measurement approach that is based on item response theory models and originates from psychometrics and educational measurement. Parallel to CAT-related research and development, the computer science-driven field of Artificial Intelligence (AI) has also developed strongly in recent years. An important area of AI research and development is adaptive measurement and testing. However, CAT and AI have so far only partially taken note of each other’s approaches and results. This may lead to one area struggling in certain points and having to start from scratch even though the other area has already resolved those same points. As a starting point to overcome this, the IACAT 2021 conference in Frankfurt will link the two areas in an interdisciplinary fashion. Researchers from both areas, CAT and AI, are warmly invited to present and discuss their current research and developments regarding adaptive measurement and testing. We believe that both areas can thereby learn from each other and thus take full advantage of the technology that is available in the digital age.

Strands

  • Automatic item generation with machine learning algorithms
  • Automatic scoring in CAT with AI models
  • Bayesian approaches to CAT
  • CAT applications
  • CAT in large-scale assessments
  • Cognitive diagnosis CAT
  • Constraint Management in CAT (e.g., content balancing, exposure control)
  • Detection of disengagement and cheating with AI
  • Formative adaptive assessment
  • Item banking
  • Item pool design
  • Item selection methods
  • (M)IRT models in CAT
  • Multi-stage testing
  • Natural Language Processing in CAT
  • Online-calibration
  • Usage of AI to score oral responses
  • Software for CAT development & application
  • Statistical foundations of CAT
  • Systems and models connecting learning and adaptive testing