Abstract
Background: The Senior South African Individual Scale – Revised (SSAIS-R) has served as a cornerstone of cognitive assessment for children in South Africa for three decades, playing a critical role in educational and clinical contexts. However, its psychometric properties have not been updated since its publication in 1991, raising doubts about its continued reliability and validity. The original standardisation project was limited to English- and Afrikaans-speaking children and did not include all population groups, raising concerns about applicability in South Africa’s diverse multilingual and multicultural population. Since 1991, the South African education system has undergone substantial transformation, and established psychometric practice recommends periodic revision and renorming to account for score inflation associated with the Flynn effect. Comprehensive psychometric re-evaluation using modern methods has been lacking, creating uncertainty about the instrument’s fairness and diagnostic accuracy across diverse linguistic and cultural groups. Establishing the psychometric integrity of this locally developed tool is essential for psychologists to make accurate diagnoses of neurodevelopmental disorders, develop educational intervention programmes, and make informed decisions regarding school placement.
Aim: This study investigated the robustness of the psychometric properties of the SSAIS-R when assessing South African children aged seven to 16 years, 30 years after its initial publication.
Objectives: The study aimed to (1) determine the reliability of the SSAIS-R, (2) evaluate the construct validity of test scores, and (3) examine how test items could be re-sequenced to better reflect the cognitive abilities of contemporary South African children.
Methodology: Data were analysed from a full sample of 897 children aged 7 years 0 months to 16 years 11 months, representing all four major population groups and nine official languages. A comprehensive dual-framework approach integrated classical test theory methods (descriptive statistics, Cronbach’s alpha, Spearman-Brown split-half reliability, confirmatory and exploratory factor analyses) with item response theory techniques (Rasch and partial credit modelling, Wright maps). Differential item functioning analysis compared English-speaking and isiZulu-speaking children to examine potential linguistic bias.
Results: Reliability analyses revealed variable internal consistency across subtests, with most subtests demonstrating adequate reliability while others showed weaker results and larger measurement error. IRT analyses indicated acceptable measurement precision for average-ability participants but reduced sensitivity at higher ability levels. Construct validity analyses did not fully support the original two-factor theoretical structure; instead, a four-factor model distinguishing verbal reasoning, visual-perceptual processing, quantitative reasoning, and cognitive proficiency presented improved statistical fit and greater theoretical coherence. Item-level analyses identified several misfitting items and significant differential item functioning favouring English speakers, particularly in Vocabulary and Story Memory, indicating construct-irrelevant variance. Logit-based difficulty estimates supported item resequencing in multiple subtests.
Conclusion: The SSAIS-R retains diagnostic potential but requires systematic revision to improve reliability, validity, fairness, and theoretical coherence...