The development of the web-based cognitive assessment battery for the identification of twice-exceptionality

This study aims to develop a Web-Based Cognitive Assessment Battery (WECAB) tailored to the cognitive characteristics of twice-exceptional students and evaluate the effectiveness of machine learning algorithms in identifying these students. A meta-analysis was conducted to determine the areas to be assessed by the WECAB. The cognitive characteristics of gifted students and those with learning disabilities are taken into account, and the battery was designed to evaluate four domains: Non-verbal Ability, Memory, Naming Speed, and Pseudoword Reading. The study included a sample of 425 elementary school students, encompassing typically developing, gifted, twice-exceptional, and learning-disabled students. Cluster sampling was used for typically developing students, and criterion sampling was used for other students. Construct validity was assessed through hierarchical confirmatory factor analysis, which demonstrated an excellent fit with the data set. Criterion validity was established by showing significant correlations between the WECAB tasks and four different measurement tools: TONI-3 (.92), the Rapid Naming Test (.91), Working Memory Test (.65), and reading speed (.54). The internal consistency reliability of the WECAB tasks was high (α = .84-.93, ω = .85-.93) and test-retest reliability for two administrations conducted 11-12 weeks interval ranged from .78 to .92. Moreover, the developed supervised machine learning algorithm achieved an accuracy of 96.88% in distinguishing between twice-exceptional, gifted, typically developing, and learning-disabled students. Thus, the WECAB emerges as a valid and reliable tool for assessing primary school students, offering the capability to discriminate twice-exceptional, gifted, and learning disabled students from their typically developing peers using machine learning algorithms.