Chun Wang

My scientific career starts in the field of educational and psychological measurement, and in recent years gradually pivots to the emerging intersection between measurement and data science. The goal of this gradual transition is to leverage artificial intelligence (AI) and machine learning (ML)-based psychometric tools to produce assessments that are reliable and secure, fair, inclusive, and provide interpretable diagnostic feedback. I served as a member on the Design and Assessment Committee (DAC) of National Assessment of Educational Progress (NAEP) since 2017, and I was the editor-in-chief for the Journal of Educational Measurement between 2022-2025. My work is funded by federal grants from Institute of Education Sciences, National Science Foundation, National Institute of Health, and private companies such as McGraw-Hill, Pearson, and Duolingo. 

My main line of work spans three interrelated strands, each of which has roots in item response theory (IRT). They are: (1) developing computerized adaptive testing (CAT) for measuring multifaceted constructs; (2) expanding cognitive diagnostic models to move the needle on formative use of diagnostic assessment in the classroom and at scale, and to explore complex dynamics of student learning that are essential to scalable, individualized, and adaptive instruction; and (3) proposing new models and methods for analyzing both micro level (e.g., item level response time, intra-individual level data) and macro level (e.g., population level longitudinal data) assessment data to accommodate idiosyncratic data structure and analysis needs. 

My recent work is inspired by and will fuel cross pollution between psychometrics and AI. On one hand, I use modern AI techniques to discover complex patterns from massive, heterogeneous, and irregular data such as Electronic Health Record (EHR) data. Meanwhile, I also develop novel psychometric models to better evaluate GenAI models’ performance across diverse contexts (e.g., multilingual, multiple prompt).