Computational Thinking: A Meta-Review of Systematic Reviews and Meta-Analyses

Synthesis: In brief: Astor, Rönnlund, Fawcett and Gredebäck synthesize the 128 systematic reviews and meta-analyses written in the broad field of computational thinking (CT) to clarify the concept's foundations and map research trends. They identify Education (primarily K-12), Programming, Tools, Pedagogical approaches, and Assessment as the dominant themes, show CT review research is concentrated among authors from China, the US, Turkey, Brazil, and Hong Kong, and flag a recent surge of only-loosely-cross-referenced reviews as emerging redundancy. Their central finding is that the CT literature is conceptually fragmented — yet the differing definitions remain fundamentally aligned, allowing a coherent unified definition of CT as reasoning with abstract models that use computational steps and algorithms to solve problems.

The meta-review spans the full CT field rather than any single intervention or technology. It uses content analysis of 128 systematic reviews and meta-analyses to examine definitional coherence, global distribution, potential review inflation, and thematic coverage. The authors find that CT research is increasingly recognized as a critical 21st-century skill comparable to reading, writing, and mathematics, with strong integration into curricula across all levels of education but a heavy emphasis on K-12. Five themes dominate the review literature: Education (primarily K-12), Programming, Tools, Pedagogical approaches, and Assessment.

A distinctive contribution is the paper's treatment of the CT definition problem. While the literature offers many definitions that differ in breadth and character, the authors argue they remain "fundamentally aligned" and converge on a coherent unified definition: computational thinking is reasoning that uses abstract models, computational steps, and algorithms to solve problems. This definition is offered as a common reference frame for the fragmented field. The meta-review also documents a recent surge in CT reviews with limited cross-referencing between them — a sign of emerging redundancy — though it does not yet rise to a critical level.

Key Findings

• 128 systematic reviews and meta-analyses now constitute the broad CT literature; the meta-review synthesizes all of them.
• Five dominant themes: Education (primarily K-12), Programming, Tools, Pedagogical approaches, and Assessment.
• Geographic concentration: CT review research is dominated by authors from China, the United States, Turkey, Brazil, and Hong Kong.
• Conceptual fragmentation is the field's greatest challenge: definitions differ in character, but remain fundamentally aligned, permitting a unified definition of CT as abstract modeling with computational steps and algorithms to solve problems.
• Emerging redundancy, not yet critical: a recent surge in CT reviews with limited cross-referencing suggests the field is beginning to repeat itself.
• Offers a common reference frame for a fragmented literature — valuable for researchers and for curriculum/CS Education integration.

What this means for practice

• Instructors. Teach CT as reasoning with abstract models, computational steps, and algorithms rather than as coding fluency, since the 128 reviews differ in emphasis but converge on that core.
• Instructors. Treat CT as a foundational literacy comparable to reading, writing, and mathematics, and give it curriculum time at every level rather than treating it as an elective computing topic.
• Curriculum designers. Build CT sequences around the two dominant vehicles in the review literature — programming and tools, including educational robotics — while noting that the evidence base is heavily weighted toward K-12, leaving secondary and higher-education pathways comparatively under-evidenced.
• Curriculum designers. Adopt one shared definition, Aho's (2012) formulation of CT as formulating problems so their solutions can be represented as computational steps and algorithms, so that outcomes and assessments stay comparable across courses and institutions.
• Researchers. Justify new CT reviews by demonstrating their net benefit and cross-referencing overlapping work: six thematically overlapping reviews cross-referenced each other only seven times across the corpus, which the authors read as evidence of emerging redundancy.

Limitations

• The synthesis covers only the 128 systematic reviews and meta-analyses identified (130 retrieved, 2 of which could not be obtained), so it maps the synthesizing literature rather than the primary CT evidence base.
• Inclusion required English-language, peer-reviewed reviews published between 2006 and 2023 indexed in EBSCO and Web of Science; 5,214 records were screened down to the final 128, leaving primary studies, non-English reviews, and post-2023 work outside the corpus.
• The definitional analysis rests on how frequently definitions were cited within the 128 included reviews; the authors acknowledge that citation counts are influenced by recency and may not capture the depth of a work's influence.
• Geographic claims come from first-author affiliations retrieved from Scopus for review articles only, and the authors note that other bibliometric analyses aggregate publication types, leaving it unclear whether these patterns hold across research forms. Only 23 of the 128 publications were meta-analyses, so most conclusions rest on review-level synthesis.

Citation

Astor, K., Rönnlund, J., Fawcett, C., & Gredebäck, G. (2026). Computational thinking: A meta-review of systematic reviews and meta-analyses. Educational Research Review, 52, 100794.