A meta-analysis of the impact of the implementation of thinking skills approaches on pupils

29 studies

School-age pupils; named programmes include Feuerstein's Instrumental Enrichment, Cognitive Acceleration through Science Education (CASE) and Philosophy for Children, plus general metacognitive approaches.

The EPPI-Centre synthesis that supplied the evidence-base sentence for a generation of UK thinking-skills policy, and the ancestor of the EEF Toolkit's "metacognition and self-regulation" strand that the LTSS trial was funded on. Twenty-nine studies, heavily weighted to small, developer-run evaluations with programme-aligned outcome measures; CASE's own non-randomised results are inside the pool. The review itself concedes it "is not clear to what extent the benefits are due to specific aspects" of the programmes or their implementation. The subject pattern is the giveaway: the largest effects appear in maths (0.89) and science (0.78) and the smallest in reading (0.4), which is what you would expect if the outcome measures sit closest to the puzzle-like content of the programmes rather than if a general reasoning capacity had been raised.

Pooled effect of thinking-skills approaches ~0.62 on cognitive measures, ~0.62 on curricular outcomes, and ~0.74 overall across cognitive, curricular and affective outcomes — numbers so large that on METHODOLOGY's own benchmark they are a red flag rather than a result. Impact varied by subject: maths 0.89, science 0.78, reading 0.40. Every one of the three flagship programmes in the pool has since been put to a well-powered independent randomised test in England and failed: CASE/Let's Think Secondary Science (EEF 2016, g = -0.01 science, -0.15 English, -0.11 maths) and Philosophy for Children (EEF 2021, g = 0.02 reading, 0.04 maths, very high security). This source's value to the archive is now as the before-picture in an efficacy-to-effectiveness collapse, not as an estimate.

Medium-to-high: pools pre-post and non-randomised designs from volunteer schools; the aggregate cannot separate programme effects from the effects of being a school that adopts programmes.

Failed. The two most heavily represented named programmes were later randomised at scale by an independent What Works Centre and produced null-to-negative results on independent standardised measures. The gap between 0.62-0.74 here and ~0.00 there is roughly the whole efficacy-to-effectiveness decay curve in one comparison.