## Dataset (Main folder)

**Reason:** Involve confidential and sensitive data.

# Supporting data for “Subtypes of Mathematical Learning Disability and Their Antecedents: A Cognitive Diagnostic Approach”

A sample was drawn from a pool of 3,384 second graders from 19 primary schools located in 13 different districts in Ningxiang County, Hunan Province, China. Children with MLD and LA were screened in two steps: First, reference was made to the children’s final mathematical exam scores in both Fall and Spring semesters of first grade. Those ranking below the 35th percentile (see also Berch & Mazzocco, 2007) in both semesters within their own school were selected, resulting in 661 children having potential mathematical learning difficulties. Next, an IQ test and a standardized mathematics test with a Chinese national norm (see below for more detail) were used for further screening. For children with normal intelligence (IQ > 85), those with mathematical test scores below the 16th percentile of the national norm were identified as having MLD (see also Bartelet et al., 2016; Devine et al., 2018), and those between the 16th to 35th percentiles were identified as having LA (see also Berch & Mazzocco, 2007). This step led to the final identification of 204 MLD children (6%) and 253 LA children (7.5%).

A composite task battery including numerical and cognitive-linguistic tasks was administered. Mathematics achievement, intelligence, mathematical fact retrieval, word problems, and written arithmetic were measured in a group situation using a paper-and-pencil format. Tasks of counting, number-words comparison, and the domain-general cognitive-linguistic skills were administered one-to-one in a quiet classroom in the schools. Tasks of number-numerosity mapping, dot comparison, and approximate arithmetic were administered in small group situation using a computerized format, with one group consisting of 5-8 children and one tester.

The standardized mathematics achievement test, intelligence test, domain-specific numerical tasks, mathematical outcome measures, and domain-general cognitive-linguistic tasks were administered to the 661 children with potential mathematical learning difficulties.