Development and Validation of a Construct Map and Assessment Instrument for Abstraction in STEM

Abstract Abstraction is a foundational cognitive skill in science, technology, engineering, and mathematics (STEM) disciplines, enabling learners to generalize concepts, recognize patterns, and navigate complexity across domains. Despite its recognized importance, abstraction remains poorly defined and difficult to assess in educational settings, limiting efforts to understand and support its development. This study presents the development and pre-validation of a construct map and corresponding assessment instrument designed to measure abstract thought in STEM education. The construct map defines abstraction as a learning progression across five levels: from basic recognition of concrete patterns to sophisticated evaluation of abstract principles. We designed 58 scenario-based items aligned with this framework and administered the instrument to more than 600 undergraduate students across multiple STEM fields at two major German universities. By leveraging the Berkeley Evaluation and Assessment Research (BEAR) Assessment System and Item Response Theory, we present evidence of the validity, reliability, and applicability of the instrument in disciplinary contexts. The findings demonstrate the sensitivity of the assessment to developmental differences and its potential to reveal how students engage in abstraction across STEM domains. This work offers a theoretically grounded practical framework for assessing abstraction and contributes to ongoing efforts to better understand, measure, and ultimately foster the development of abstraction skills in STEM learners.