Modern test design

Methodological advances in modern test design unfortunately move only very slowly into actual testing practice. The major consequence is that many cognitive ability tests as well as educational and psychological assessments currently in use or newly developed, are still constructed ad hoc and validated at most by relating scale-level scores to external variables or by using mere face validity or trusting expert opinion. Doing research on and research with modern test design methodology in a strong item-level validation framework will be one of the recurring topics in this research area that will be specifically defined through a) theoretical item models defining radical and incidental item features, b) explanatory item response models for item-level validation experiments, c) automatic item generation and pre-calibration assisted by machine learning and generative artificial intelligence to systematically build effective item banks and d) adaptive testing for efficient test administration.