The essence of creative thinking in PISA 2022: An item-level machine learning approach to construct validity at the facet level
Thinking Skills and Creativity · 2026-09 · 61 · 102205
DOI: 10.1016/j.tsc.2026.102205 ↗Kurz gefasst
Fachartikel in der Zeitschrift Thinking Skills and Creativity, 2026-09. Verfasst von Elisabeth Barbara Kraus und 2 weiteren Autor:innen. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 68 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
The definition and measurement of creative thinking remain subjects of debate. In 2022, the Programme for International Student Assessment (PISA) introduced creative thinking as a new domain, conceptualized across three facets: generating diverse ideas, generating creative ideas, and evaluating and improving ideas. While this framework provides a valuable foundation, questions regarding the construct validity of PISA’s creative thinking assessment persist. This study adopts a data-driven approach to examine the construct validity of the three creative thinking facets by exploring their relationships with psychological constructs, demographic and process data, and item characteristics. Using item-level response data from PISA 2022, we modeled how individual item scores relate to these variables. To account for the hierarchical data structure and potential confounding between item content and score, we employed a multi-level predictive modeling approach based on machine learning algorithms. Explainable AI (XAI) techniques were applied post-hoc to identify the most influential predictors and their effects. Results showed that item characteristics such as content domain, process-related indicators, and academic competences in mathematics, reading, and science were the strongest predictors for item-level scores across all facets. In contrast, self-reported creativity and engagement in creative activities showed limited relevance. Accumulated local effects analyses revealed systematic but modest differences in predicted scores across item domains and facets, indicating subtle interactions between content, competences, and performance. However, these differences were small, suggesting limited empirical differentiation between the three facets as implemented in PISA 2022. From a validity perspective, these findings imply that while PISA’s creative thinking assessment captures meaningful variance related to engagement and aptitude, it may not fully distinguish between theoretically distinct aspects of creative thinking. The study highlights the utility of machine learning and XAI in construct validation research.
Zitieren
APA
Kraus, E. B., Goecke, B., & Jaggy, A. (2026). The essence of creative thinking in PISA 2022: An item-level machine learning approach to construct validity at the facet level. Thinking Skills and Creativity, 61, 102205. https://doi.org/10.1016/j.tsc.2026.102205
Vancouver
Kraus EB, Goecke B, Jaggy A. The essence of creative thinking in PISA 2022: An item-level machine learning approach to construct validity at the facet level. Thinking Skills and Creativity. 2026;61:102205. doi:10.1016/j.tsc.2026.102205
BibTeX
@article{Kraus2026,
author = {Kraus, Elisabeth Barbara and Goecke, Benjamin and Jaggy, Ann-Kathrin},
title = {The essence of creative thinking in PISA 2022: An item-level machine learning approach to construct validity at the facet level},
journal = {Thinking Skills and Creativity},
year = {2026},
volume = {61},
pages = {102205},
doi = {10.1016/j.tsc.2026.102205}
}
EUVIMED