Subtypes of mathematical learning disability and their antecedents : a cognitive diagnostic approach

Abstract

Children with mathematical learning disability (MLD) show heterogeneity in their numerical skill deficits due to the complexity of mathematics learning. Previous studies have attempted to identify subtypes of MLD and their antecedents but have rarely considered a comprehensive theoretical framework that includes both numerical skills and cognitive-linguistic competencies. Moreover, the data-driven nature of traditional approaches to classifying MLD subtypes (i.e., cluster analysis) has not only led to difficulties in interpreting the subtypes but has also caused inconsistencies among research findings. Furthermore, little is known about how cognitive-linguistic skills contribute to MLD subtypes. To tackle these issues, the present research aims to identify subtypes of MLD in young children using a new classification approach, namely Cognitive Diagnosis Models (CDMs), and examine the antecedents of the subtypes. CDMs, as a type of confirmatory latent class model, can classify children with different performance profiles based on a series of hypothesized competencies involved in tasks. To achieve this goal, two studies were conducted. Study 1 was an analysis of secondary data collected from 99 MLD children who were identified from 1,839 Finnish children. Based on the children’s performance in numerical skills measured in preschool, five subtypes of MLD were identified. The subtypes were the unknown deficits subtype, the verbal deficits subtype, the pervasive deficits subtype, the symbolic deficits subtype, and the verbal and concept deficits subtype. The results show the CDMs’ good reliability and internal and external validity in identifying MLD subtypes. Furthermore, different subtypes depended on different constellations of language and spatial deficits. To address the limitations of Study 1, such as the small sample size of MLD children and the lack of measures of approximate number system (ANS) and working memory, Study 2 was conducted based on 204 second graders with MLD identified from a sample of 3384 Chinese second graders. Six MLD subtypes were identified: the symbolic and concept deficits subtype, the verbal and concept deficits subtype, the pervasive deficits subtype, the concept deficit subtype, the mapping and concept deficits subtype, and the unknown deficits subtype. Again, the results show the CDM’s good reliability and internal and external validity in identifying MLD subtypes. Additionally, working memory skills predicted the identification of the pervasive deficits subtype as compared with LA children even after including a large pool of control variables. In summary, the present research represents the first endeavour to use CDMs to identify MLD subtypes. The findings underscore that CDMs are a reliable and valid approach in the classification of MLD subtypes. They also highlight the importance of understanding the specific numerical deficits and the cognitive-linguistic antecedents of MLD subtypes. Custom-built interventions could be developed to improve the mathematical ability of MLD children based on each subtype’s unique numerical and cognitive-linguistic profile.