Identifying Early Mathematics Learning Difficulty Risk: A Cross-Sectional Evaluation of Rating- and Performance-Based Tiered Screening

Technical Summary: Identifying Early Mathematics Learning Difficulty Risk

Problem Statement
Early identification of mathematics learning difficulties (MLD) during the preschool years is critical for preventing later academic failure, yet it remains methodologically complex. Preschool children exhibit high developmental variability, and indicators of difficulty may manifest differently across behavioral, cognitive, and academic domains. A single assessment approach is often insufficient; performance-based measures provide direct evidence of competence but are resource-intensive, while rating-based measures offer practical efficiency but may reflect informant perceptions rather than objective deficits. Furthermore, formal diagnosis is often inappropriate in early childhood due to the lack of stabilized performance patterns, necessitating a "risk-oriented" perspective. The literature lacks sufficient comparative data on how rating-based and performance-based tools relate to one another and how they can be optimally organized within a sequential screening framework to identify MLD and broader learning difficulties (LD).

Methodology
The study employed a cross-sectional correlational design with a sample of 111 preschool children (mean age = 65.9 months) from three kindergarten settings in northeastern Türkiye. The research utilized a multi-method assessment approach integrating three distinct instruments:

1. Rating-Based Behavioral Screening: The Learning Difficulty Early Signs Screening Scale (LDESSS), a parent-report tool assessing language, cognitive, psychomotor, and social-emotional domains.
2. Performance-Based Cognitive Assessment: The Anadolu-Sak Intelligence Scale (ASİS), specifically focusing on the Memory Capacity Index (MCI) and a VSM–WoM composite (sum of Visual Sequential Memory and Words–Meanings subtests).
3. Performance-Based Academic Assessment: The Test of Early Mathematics Ability–Third Edition (TEMA-3), serving as the primary criterion for early mathematics competence.

Data analysis included Pearson correlations, hierarchical and multiple linear regressions, and screening validity analyses (sensitivity, specificity, predictive values, and Receiver Operating Characteristic [ROC] curve analyses). Risk classifications were dichotomized into "at risk" and "not at risk" based on inclusive cutoffs (e.g., TEMA-3 scores at or below the 40th percentile) to maximize sensitivity for initial screening purposes.

Key Results

• Correlational Findings: LDESSS total scores showed a significant negative correlation with TEMA-3 performance ($r = -.380$) and ASİS cognitive indicators. LDESSS scores explained an additional 14.2% of the variance in TEMA-3 scores beyond preschool attendance duration. Notably, the LDESSS Language and Communication subscale uniquely predicted MCI scores.
• Screening Validity of Rating-Based Measures (LDESSS): The LDESSS demonstrated high sensitivity for detecting MLD risk (0.90) and LD risk (0.91), effectively minimizing false negatives. However, it exhibited low specificity (0.35 for MLD; 0.33 for LD) and modest discriminative accuracy (AUC = 0.658 for MLD; AUC = 0.685 for LD). The Language and Communication subscale showed slightly better discriminative accuracy (AUC = 0.716) for LD risk.
• Screening Validity of Performance-Based Measures (ASİS): Cognitive indicators demonstrated significantly stronger discriminative accuracy. The MCI showed an AUC of 0.855 with high specificity (0.87) and a positive predictive value (PPV) of 0.81. The VSM–WoM composite showed an AUC of 0.838 with extremely high specificity (0.98) and PPV (0.96), though lower sensitivity (0.48).
• Comparative Utility: While rating-based measures captured broad observable difficulties, performance-based cognitive measures provided more precise evidence regarding the cognitive vulnerabilities underlying mathematics competence.

Key Contributions
The study contributes to the field by empirically validating the complementary roles of rating-based and performance-based assessments rather than treating them as competing alternatives. It demonstrates that:

1. Parent-reported behavioral indicators are highly sensitive tools for broad initial detection but lack the specificity required for definitive classification.
2. Performance-based cognitive indicators (specifically MCI and VSM–WoM) offer superior discriminative precision for identifying children at genuine risk for MLD.
3. Language-related behavioral indicators reported by parents are particularly informative for detecting broader cognitive vulnerabilities, not just mathematics-specific deficits.

Significance and Proposed Framework
The authors propose an evidence-informed, tiered screening framework that integrates these distinct assessment types into a sequential process:

• Tier 1 (Universal Screening): Utilize high-sensitivity, rating-based behavioral tools (e.g., LDESSS) as an efficient first-stage filter to broadly detect potential risk and minimize false negatives, acknowledging that this will result in a higher rate of false positives.
• Tier 2 (Cognitive Refinement): Apply performance-based cognitive indicators (e.g., ASİS MCI, VSM–WoM) to children flagged in Tier 1. These measures provide high specificity to refine risk classification and confirm underlying cognitive vulnerabilities.
• Tier 3 (Academic Confirmation): Conduct domain-specific academic assessments (e.g., TEMA-3) to confirm observable weaknesses in mathematics competence.

The paper argues that this tiered approach is both psychometrically sound and pragmatically viable for large-scale preschool screening. It addresses logistical constraints by using low-cost parent reports for initial screening while reserving resource-intensive direct testing for targeted refinement. The authors conclude that early identification should not rely on a single measure but rather on a multi-source system that leverages the strengths of both behavioral observation and cognitive performance to support targeted preventive interventions.