The Evidence on Teaching

Experimental Evaluation of the Tools of the Mind Pre-K Curriculum: Technical Report

Farran DC, Wilson SJ, Meador D, Norvell J, Nesbitt K · 2015

grade Arctindependentfailed
Sample
Cohort 1 - 60 pre-K classrooms (32 Tools, 28 control) in 45 schools across 5 districts in Tennessee and North Carolina, 866 children with a pretest; Cohort 2 - 20 classrooms, 12 schools, 263 children. Followed to the end of first grade.
Population
United States; pre-K children around age 4, followed through kindergarten and first grade.
Design
The independent evaluation that settled Tools of the Mind, and it is the design the archive asks for - randomisation at SCHOOL level blocked by district, low and balanced attrition (816 to 810 to 778 across three years), business-as-usual control, and standardized direct child assessments (Woodcock-Johnson III achievement plus Peg Tapping, Head-Toes-Knees-Shoulders, DCCS, Copy Design and span tasks for executive function). The developers were sub-awarded to deliver training and coaching, so implementation was developer-supported while evaluation was independent. When Cohort 1 came back null the developers were shown the results and intensified coaching for Cohort 2.
Key findings
At the end of pre-K there was no significant effect on any of six self-regulation or executive-function measures or on the EF composite - DCCS 0.01 (p = .94), Copy Design 0.15, Peg Tapping 0.16, HTKS 0.03 - nor on any achievement measure. Every statistically significant effect that did appear FAVOURED THE CONTROL GROUP - end-of-pre-K oral comprehension, end-of-kindergarten letter-word identification, quantitative concepts and the achievement composite, end-of-kindergarten backward digit span, end-of-first-grade Copy Design and the self-regulation composite, and end-of-first-grade spelling. Cohort 2, run after the developers intensified coaching, produced the same result - no difference in achievement, self-regulation or teacher ratings at the end of the pre-K year. Implementation fidelity varied but was NOT associated with greater gains in achievement or self-regulation.
Genetic confound
Low. School-level randomisation blocked by district.
Replication notes
This is the failed replication of Barnett 2008 and Diamond et al. 2007. Reinforced by Solomon et al. 2018 in Canada with an active control and prospective registration, and published in peer-reviewed form as Nesbitt and Farran 2021.

Effects

OutcomeMetricValueMeasureTimingVsHorizonClass
Executive function and self-regulation at end of pre-Keffect sizeno significant effect on any of six measures or the composite (DCCS 0.01, Copy Design 0.15, Peg Tapping 0.16, HTKS 0.03)standardizedend of pre-Kbusiness-as-usualend-of-treatmentnear-transfer
Achievement at end of pre-K (Woodcock-Johnson III)effect sizenull throughout; letter-word -0.11, applied problems 0.03, picture vocabulary -0.05standardizedend of pre-Kbusiness-as-usualend-of-treatmentdomain-skill
Statistically significant effects across all wavesdirectionevery one favoured the CONTROL group, through end of first gradestandardizedend of pre-K, kindergarten and first gradebusiness-as-usualover-2yrdomain-skill
Cohort 2 after intensified developer coachingresultsame as cohort 1 - no difference in achievement, self-regulation or teacher ratingsstandardizedend of pre-Kbusiness-as-usualend-of-treatmentnear-transfer
Implementation fidelity and child outcomesassociationfidelity was not associated with greater gainsstandardizedend of pre-Kbusiness-as-usualend-of-treatmentnear-transfer

Cited by