Enhancing computational thinking assessment: A multimodal cognitive diagnostic approach

Abstract

Computational thinking (CT) is an essential skill students need if they are to adapt to the rise of digital society, and the evaluation of CT is crucial for enhancing students’ CT. However, the current evaluation of CT is based on a summative paradigm and ignores real CT performance during the assessment process. This study thus used a multimodal cognitive diagnostic model (H-MCDM) of students’ CT based on multimodal data in situational tasks, to achieve precise evaluation and explore the current mastery status of students’ CT and their cognitive and behavioral characteristics. The research results indicate that: 1) students’ CT was at a moderately high level and indicated a good grasp of the CT components of problem decomposition and algorithm design, but students had not yet mastered problem abstraction, 2) a total of six different CT latent patterns were discerned in the overall student results, 3) individual students who shared the same CT latent patterns could differ in their cognitive and behavioral characteristics, while those with different CT latent patterns could share similar characteristics. This study showed that it is possible to capture the degree of students’ mastery of CT and their cognitive and behavioral differences with the same or different CT latent patterns at a granular level, providing a basis for developing individualized strategies to enhance students’ CT.