The efficacy of different interventions to foster children's executive function skills: A series of meta-analyses.
Takacs ZK, Kassai R · 2019
grade Bmeta-analysisindependentmixed
Sample
100 independent effect sizes from 90 studies, 8,925 children aged 2-12
Population
Children aged 2-12 across Europe, North America, Asia, Australia and South America.
Design
The most complete map of what kind of executive-function intervention does what, and the only one that separates intervention TYPE properly. Outcomes are restricted to neurodevelopmental tests rather than rating scales, which is the right restriction. 62 effect sizes came from randomised trials and 38 from quasi-experiments with no detectable difference between them; 55 used passive controls, 39 active, 6 both. The funnel plot was asymmetric and trim-and-fill lowered the adjusted overall estimate to g = 0.10 (0.02 to 0.18).
Key findings
Overall g = 0.30 at post-test (working memory 0.35, inhibition 0.22, flexibility 0.18). Explicit EF training g = 0.38, and - the result that matters most - it worked BETTER for typically developing children (0.46) than for atypical ones (0.24), the reverse of what a remediation logic predicts. Computer training 0.42. Aerobic exercise alone 0.05 and not significant. EF-specific CURRICULA managed only g = 0.12, marginal, with Tools of the Mind specifically at g = 0.13 (k = 3, 95% CI -0.11 to 0.37, p = .28). Art 0.07, music -0.04, drama and pretend play 0.10, all non-significant. Self-regulation strategies 0.46 and mindfulness 0.46 (k = 6, fail-safe n = 19, not robust). At FOLLOW-UP (6 weeks to 12 months, k = 15) the raw estimate was 0.18 but trim-and-fill adjusted it to 0.008 (-0.18 to 0.19); inhibition 0.04 and flexibility -0.10 at follow-up, and EF curricula at follow-up g = -0.03.
Genetic confound
Low. Experimental training contrasts.
Replication notes
The headline that EF task performance can be trained replicates. The follow-up null and the curriculum null are consistent with Diamond and Ling's own 2020 review and with the independent Tools of the Mind trials. Scionti et al. 2020 disagree on one moderator, finding larger effects for at-risk preschoolers, which is a live disagreement.
DOI / URL
10.1037/bul0000195
Effects
| Outcome | Metric | Value | Measure | Timing | Vs | Horizon | Class |
|---|---|---|---|---|---|---|---|
| Overall executive-function outcomes at post-test | Hedges g | 0.30 (WM 0.35, inhibition 0.22, flexibility 0.18); trim-and-fill adjusted 0.10 (0.02 to 0.18) | standardized | post-training | unclear | end-of-treatment | near-transfer |
| Explicit EF training, typically developing vs atypical children | Hedges g | 0.46 vs 0.24 — training worked BETTER for children who needed it less | standardized | post-training | unclear | end-of-treatment | near-transfer |
| EF-specific curricula (including Tools of the Mind) | Hedges g | curricula 0.12 marginal; Tools of the Mind 0.13 (k=3, 95% CI -0.11 to 0.37, p = .28) | standardized | post-training | business-as-usual | end-of-treatment | near-transfer |
| Aerobic exercise alone | Hedges g | 0.05, not significant (typically developing -0.15) | standardized | post-training | unclear | end-of-treatment | near-transfer |
| All EF outcomes at follow-up (6 weeks to 12 months) | Hedges g | raw 0.18 (0.02 to 0.35); trim-and-fill adjusted 0.008 (-0.18 to 0.19) | standardized | 6 weeks to 12 months post-training | unclear | under-1yr | near-transfer |
| EF curricula at follow-up | Hedges g | -0.03, not significant | standardized | follow-up | business-as-usual | under-1yr | near-transfer |
Cited by
- Can executive function be trained, and does it transfer to learning?no effectconf: highgc: low