Programming and 21st century skill development in K-12 schools: A multidimensional meta-analysis

74 studies (2000-2023), 136 effect sizes

K-12 students, international.

Pools eight heterogeneous "21st century sub-skills" — critical thinking, creativity, communication, collaboration, problem solving, learning to learn, ICT literacy, and personal and social responsibility — into a single estimate. Almost none of these are programming competence, and the composite is exactly the undifferentiated "improves learning" bucket the archive's outcome taxonomy exists to prevent. No control-type or publication-bias moderator is visible at abstract depth. Read at abstract depth. Included mainly because it is the most-cited recent aggregate on the policy claim.

Overall Hedges g = 0.621, p < .01. The recommendations include starting programming in kindergarten, keeping interventions under six months, and using visual tools — and the second of those is another effects-shrink-with-dose result, which is a small-study and measure-alignment signature rather than a dosing recommendation.

Medium. Mixed-design pool with no control-type moderation.

Overlaps heavily with Scherer et al. (2019) and Li & Oon (2024) in its primary studies, so the apparent agreement across these three aggregates is partly the same studies counted three times.