The Evidence on Teaching

Closing the Achievement Gap through Modification of Neurocognitive and Neuroendocrine Function: Results from a Cluster Randomized Controlled Trial of an Innovative Approach to the Education of Children in Kindergarten

Blair C, Raver CC · 2014

grade Brctunclearunreplicated
Sample
29 schools, 79 classrooms, 759 kindergarteners; about half high-poverty (over 75% free/reduced-price lunch) and half low-poverty
Population
Massachusetts, United States; kindergarten children.
Design
The strongest positive Tools of the Mind trial and the one the programme's defenders rely on. Block randomisation at school level with two districts randomised at classroom level, about 5% attrition unrelated to condition, no baseline differences, objective child assessments and salivary biomarkers. Two weaknesses matter. The control was BUSINESS AS USUAL while Tools teachers received two years of professional development and coaching, so dose and attention are confounded with content. And independence is unclear - the authors are not the programme's originators but are the champions of the kindergarten version, designing and running a trial of a programme they advocate.
Key findings
Whole-sample effects were small - backward digit span 0.14 (p < .05), EF composite reaction time 0.12 (p < .05), maths 0.13 (p < .05) - with NO overall effect on EF composite accuracy. The large effects are all in the high-poverty subgroup - EF composite reaction time 0.48 (p = .10), accuracy 0.31 (p < .10), attention-bias 0.82 (p < .05), vocabulary 0.43, Raven's reasoning 0.46. At first-grade follow-up reading was 0.14 (p < .05), having grown from a non-significant 0.07, vocabulary 0.10 and maths 0.07 (p < .10). Salivary cortisol was higher and alpha-amylase lower in Tools high-poverty schools, which the authors read favourably.
Genetic confound
Low for the randomised contrast. The subgroup analyses are where the reported effects live and they are not protected by randomisation in the same way.
Replication notes
Unreplicated. It is the only positive large Tools trial using objective EF measures, there is no independent kindergarten replication, and no follow-up beyond first grade has been published. The pre-K version of the programme failed independent replication comprehensively.
DOI / URL
10.1371/journal.pone.0112393

Effects

OutcomeMetricValueMeasureTimingVsHorizonClass
Backward digit span (working memory), whole sampleeffect size0.14, p < .05standardizedend of kindergartenbusiness-as-usualend-of-treatmentnear-transfer
Executive-function composite accuracy, whole sampleeffect sizeno overall effect (high-poverty subgroup 0.31, p < .10)standardizedend of kindergartenbusiness-as-usualend-of-treatmentnear-transfer
Mathematics, whole sampleeffect size0.13, p < .05standardizedend of kindergartenbusiness-as-usualend-of-treatmentdomain-skill
Reading at first-grade follow-upeffect size0.14, p < .05 (grew from a non-significant 0.07)standardizedend of first gradebusiness-as-usual1-2yrdomain-skill
High-poverty subgroup, reasoning and vocabularyeffect sizeRaven's 0.46, vocabulary 0.43standardizedend of kindergartenbusiness-as-usualend-of-treatmentfar-transfer

Cited by