Characteristics and psychometric properties of computational thinking assessments in children and adolescents: A systematic review

Background: Computational thinking is a problem-solving approach that utilizes computer science principles, enabling a computer to carry out the solution. Over the past decade, computational thinking has become essential in 21st-century education, particularly in K-12 curricula. Assessment tools have been growing in numbers for computational thinking in K-12 education, often equating it with programming or digital skills. However, these assessments frequently lack consensus regarding what should be measured and often fail to meet standard psychometric criteria. To address this, the present study conducted a systematic review of the assessment instruments available to date, analyzing their characteristics, psychometric qualities, and underlying theoretical constructs with the aim of identifying instruments that demonstrate robustness and reliability, as well as highlighting new lines of research in this area. Method: This systematic review followed the parameters of the PRISMA statement guidelines. The reviewed studies were identified and retrieved from the following electronic databases and publication platforms: Science Direct, Springer Link, Taylor & Francis, Sage, IEEE Xplore, Web of Science, and Psychinfo. The automatic filters applied included research articles and full-text publications. The review protocol was prospectively registered in INPLASY202340069. Results: Sixty-four studies met the eligibility criteria and reported on 65 assessment instruments. Most instruments were questionnaires and task-based rubrics, with elementary school children being the most frequently targeted population. Of the included studies, 48 (75%) reported at least one measure of reliability or validity; 36 (56%) reported reliability evidence and 36 (56%) reported validity evidence. At the study level, internal consistency was the most frequently reported reliability measure (42.2%), followed by inter-rater reliability (14.1%), test–retest reliability (6.3%), alternate-form reliability (4.7%), split-half reliability (1.6%), and plausible values reliability (3.1%). Criterion validity (26.6%) and content validity (23.4%) were the most frequently reported forms of validity evidence, followed by construct validity (21.9%) and face validity (9.4%). The instruments with the strongest psychometric evidence were further examined, while those requiring additional validation were identified. Conclusion: The review identifies numerous tools for assessing computational thinking in K-12 education, showing a wide variety in the computational thinking components they aim to evaluate. While some studies report on the reliability of these tools, evidence of their validity is less frequently documented.