Peer-reviewed article
Early Identification of Children at Risk for Long-Term Difficulties with Mathematics Learning and Innumeracy in Adolescence.
Ferrer E, Geary DC, Shaywitz BA, Holahan JM, Shaywitz SE ·
Summary
Written by AI, not yet AICS's view · claude-haiku-4-5-20251001This study followed 445 children from 1st to 10th grade to identify which early indicators best predict long-term mathematics learning difficulties and innumeracy in adolescence. Researchers tested different criteria for identifying at-risk students, including achievement scores and teacher observations of behavior. Children who scored below the 25th percentile in mathematics in 1st grade were found to be the best indicator of future difficulties, and these at-risk students fell further behind their peers as they progressed through school.
Key points
- An achievement cutoff below the 25th percentile in 1st grade mathematics provided the best balance of sensitivity (0.589) and specificity (0.878) for identifying future innumeracy risk
- At-risk students identified in 1st grade remained behind their peers throughout schooling, with the achievement gap widening from 1st to 10th grade
- Low achievement in 1st grade was predictive of difficulties independent of cognitive ability, attentive behavior, and parental socioeconomic status
- Not all at-risk students (those predicted to struggle) actually experienced long-term difficulties, indicating additional unmeasured factors influence outcomes
- Achievement-only criteria were more effective for early identification than ability-achievement combinations
Limitations
The study does not identify what additional factors cause some at-risk children to succeed despite low 1st grade achievement. The sample characteristics and country context are not specified in the available metadata, limiting generalizability.
Abstract
Children who start school behind in mathematics are at-risk of staying behind throughout schooling and into adulthood. Unfortunately, there is no consensus on the best criteria for the early identification of these at-risk students. Using a unique 1st to 10th grade study of academic development (n = 445), we assessed the sensitivity and specificity of different ability-achievement and achievement-only criteria for identifying risk of mathematics learning difficulties (MLD) and adolescent innumeracy. The outcome criterion was based on the risk of innumeracy in 10th grade (< 25th mathematics percentile). The achievement-only cutoffs were mathematics scores below the 10th to 35th percentile (at 5 percentile increments, inclusive) in 1st grade and across 1st and 2nd grade, and teacher-reported in-class attentive behavior. The best combination of sensitivity (.589) and specificity (.878) was found for an achievement cutoff below the 25th percentile in 1st grade. We then examined growth in mathematics from 1st to 10th grade for different at-risk and typically achievement groups using latent score change (LCS) models. The models revealed that at-risk students were behind their peers in 1st grade, with a widening gap across the school years. However, not all at-risk students were identified indicating that additional factors need to be considered in follow-up studies. Until that time, low-achieving 1st graders, independent of cognitive ability, attentive behavior, and parental socioeconomic status, are at risk for long-term difficulties in mathematics and innumeracy in adolescence.