Block-based versus text-based programming environments on novice student learning outcomes: a meta-analysis study

13 publications, 52 effect-size comparisons

Novice programmers across K-12 and undergraduate education.

Small k. Both pooled estimates are non-significant under random effects, and the authors detected publication bias. Moderator analyses by education level, learning environment and duration are underpowered at k = 13, and hybrid block/text environments could not be handled by the coding scheme. Read at abstract depth. This is the only independent meta-analytic test of the single design principle the entire K-12 computing movement is built on.

Blocks beat text on cognitive outcomes by g = 0.245, 95% CI -0.078 to 0.567, p = .137 — a confidence interval straddling zero. Affective outcomes g = 0.195, 95% CI -0.289 to 0.678, p = .429. Publication bias was detected. The field's most confidently asserted design principle, that blocks are a gentler on-ramp, is not supported by the pooled evidence.

Low. The comparison is between two environments delivered to comparable novices; the threat is design quality, not heredity.

No competing meta-analysis of the same contrast was located. Descriptive and behavioural comparisons of block, text and hybrid modalities exist but neither confirm nor refute the null, and a 2025 systematic review of the block-to-text transition could not be obtained.