Developmental differences in classifiers of low and high mathematics performance: A classification and regression tree analysis [Author Accepted Manuscript]

Background: Given the contribution of domain-general and domain-specific skills to children’s mathematics performance and the importance of children’s early mathematical skills for future success, it is imperative to determine and compare the predictive skills that classify children’s mathematical performance. Previous efforts examining classifiers of mathematics performance have focused on development within a year (e.g., preschool; Purpura et al., 2017) or general predictive relations (e.g., Jordan et al., 2002). The question remains whether the classifiers of low and high mathematics performance differ across multiple years. Method: The current study used data from a longitudinal project that assessed children on a number of academic and behavioral measures across four time points spanning preschool and kindergarten. The analytic sample for this study consisted of 322 children (T1 Mage = 4.81 years; SD = 0.30; 50.6% female) from three cohorts. Results: Prior mathematics performance was the most consistent classifier for low and high mathematics performance at the end of kindergarten. Broad school readiness at the beginning of preschool was also a consistent classifier of low and high mathematics at the end of kindergarten. A consistent classifier of only low mathematics performance at the end of kindergarten was behavioral self-regulation skills in the spring of preschool. Further, reading skills at the beginning of kindergarten were a consistent classifier of only high mathematics performance. Conclusion: Findings can guide researchers and educators in identifying the important skills at different developmental periods in early schooling to support children at risk of low mathematics performance or those with exceptionally high mathematics performance. reviewed acceptedVersion