Subtypes of Mathematical Learning Disability and Their Antecedents: A Cognitive Diagnostic Approach

• In Event: Poster Session 10
  In Poster Session: PS 10 Topic - Developmental Disabilities

Abstract

Background
In view of the complexity of math learning, Mathematical learning disability (MLD) students show deficits in different numerical skills (Geary, 2004), which lead to different subtypes. In the literature, subtypes of MLD are classified mainly through two approaches: top-down and bottom-up. The top-down approach, which refers to identifying subtypes based on prior observations of their errors or cognitive profiles, limits the possibilities of MLD subtypes (Bartelet, Ansari, Vaessen, & Ansari, 2014) and does not consider the fitness with empirical data. The bottom-up approach, referring to identifying subtypes through data-driven method, is generally hard to interpret from a theoretical perspective (Li, Cohen, Bottge et al., 2016).
Aims
The present study utilized a confirmatory latent class model, Cognitive Diagnostic Modelling (CDM), to identify subtypes of MLD based on children’s mathematics performance measured between Grades 1 and 4 and numerical skills measured in preschool. We also evaluated the reliability and validity of using this model for subtyping and examined the cognitive-linguistic antecedents of different subtypes.
Methods
The study is part of a longitudinal project which followed 1,880 Finnish children from kindergarten (T1) to fourth grade (T6) (984 boys; mean age at T1= 74 months, SD = 3.6 months). Among them, 99 was identified with MLD and 420 with low achievement (LA) in a previous study (Zhang et al., 2020). Measures used for identifying subtypes consisted of a counting task, a basic arithmetic concept task, and a number-numerosity mapping task, which were administered at T1. Arithmetic reasoning, measured at T6, was used for validation purposes. Cognitive-linguistic antecedents, including phonological awareness, letter knowledge, spatial visualization, receptive vocabulary, and rapid automatized naming, were measured at T1.
CDM was used to identify subtypes of MLD. For validation purposes, we tested whether each subtype showed lower scores in the corresponding numerical skills than LA children using logistic regressions, and whether the subtypes showed differences in arithmetic reasoning in fourth grade using ANOVA. At last, we examined the cognitive-linguistic antecedents that predicted the classification of each subtype using logistic regression.
Results
We identified eight subtypes, among which five subtypes accounted for more than 90% of the MLD children. These five subtypes were labelled as the arithmetic fluency deficit only subtype, the counting deficit subtype, the pervasive deficit subtype, the symbolic deficit subtype, and the counting and concept deficits subtype. The subtypes showed lower scores in the corresponding numerical skills than LA children, and they also showed significant differences in arithmetic reasoning in fourth grade (F(4, 81) = 4.285, p < .01). Language skills, including phonological awareness, receptive vocabulary, and letter knowledge, predicted the classification of the pervasive deficits subtype, letter knowledge predicted the classification of the symbolic deficits subtype, and spatial visualization and rapid naming predicted the classification of the counting deficit subtype.
Conclusions
The present study was the first study using CDM to identify the subtypes of MLD, which provided the cognitive profile of each student and rendered the interpretation easy.

Authors

• Xiangzi Ouyang, University of Hong Kong (HKU)
  Presenting Author

• Xiao Zhang, The University of Hong Kong
  Non-Presenting Author

• Pekka Räsänen
  Non-Presenting Author

• Tuire Koponen, Jyväskylä University
  Non-Presenting Author

• Marja-Kristiina Lerkkanen, Jyväskylä University
  Non-Presenting Author