This study aims to reveal the trajectory of disparity in receptive vocabulary developments attributable to early-life circumstances. Using the cohort data collected from Ethiopia (N = 840), India (N = 1777), Peru (N = 1624) and Vietnam (N = 1831) as the Young Lives study, this study explores the dynamics of variation in receptive vocabulary acquisitions over a 15-year period. The random forest, which is one of the most popular and versatile machine learning algorithms, is adopted to model the association between receptive vocabulary skills and early-life circumstances. Then the constructed model quantifies how much of the observed variation is due to the early-life circumstances. Such variation has already been shaped by the age of 5 and it persists even when children reach early adolescence. The prediction by the random forest suggests that compensation for disadvantaged circumstances at the early developmental stage would help marginalised children catch up and mitigate the disparity when they reach early adolescence.
ASJC Scopus subject areas