Fadeout in an early mathematics intervention: Constraining content or preexisting differences?

Follow-up of the TRIAD/Building Blocks cluster RCT cohort

US children followed from pre-K into early elementary school.

Tests two competing explanations for the fadeout of the Building Blocks effect: that later instruction constrained the content children could use, versus that stable pre-existing individual differences reasserted themselves. Analysis is longitudinal on an experimental cohort, so the fadeout decomposition is model-based rather than randomized.

The canonical fadeout decomposition: most of the disappearance of a large early-math effect tracks stable differences between children, not a failure of later schooling to build on the content. Early gains look like a head start rather than a trajectory change.

Directly lens-relevant: 'stable pre-existing differences' is the observable signature of partly heritable individual differences reasserting themselves once the intervention stops. The design cannot separate heritable from persistently environmental stability, so it constrains rather than settles the question.

Its mechanism claim is qualified by Watts et al. 2026, which finds fadeout is FASTER where control-group children learn more afterwards — a catch-up channel this paper's decomposition understates.