Objective: While computational thinking (CT) is a key competency in elementary education, its valid assessment faces significant conceptual and psychometric challenges. This research provides a diagnostic analysis that moves beyond mere classification, employing a dual approach: first, mapping the scientific landscape through scientometric analysis, and second, critically evaluating existing assessment tools.

Method: In this systematic critical review (PRISMA 2020), 45 eligible articles were identified from Scopus and Google Scholar (2006–2025). The analysis was conducted in two phases: (1) a scientometric analysis using R and VOSviewer to map the field's conceptual structure, and (2) a qualitative critical analysis of eight key instruments, focusing on their theoretical foundations and psychometric properties.

Findings: The scientometric analysis revealed a field characterized by rapid yet imbalanced growth. Keyword co-occurrence maps highlighted a strong focus on "educational applications" and "programming tools," alongside a significant absence of core assessment concepts like "validity" and "fairness." This structural gap is mirrored in the instruments, revealing three primary limitations: 1) psychometric weakness, due to over-reliance on traditional metrics (e.g., Cronbach's alpha) while neglecting modern standards (e.g., McDonald's omega, DIF analysis); 2) conceptual narrowness, focusing on algorithmic knowledge at the expense of process skills (e.g., debugging) and dispositions (e.g., perseverance); and 3) an instrumental gap for upper elementary students.

Conclusion: No single instrument is sufficient for a comprehensive assessment of CT. The maturation of this field requires adopting modern psychometric standards, designing contextually-relevant tests, and developing hybrid assessment frameworks that integrate standardized tests with performance-based measures.