Validation of the English version of TOY8 Developmental Screening Tool: Examining Measurement Invariance Across Languages, Gender and Income Groups

Abstract

Background: National Health and Morbidity Survey in Malaysia (2022) revealed that there is a significant increase in developmental delay among young children. Early detection is required by using a developmental screening tool that is easy to access and cross-culturally effective. Thus, the TOY EIGHT developmental screening tool (TOY8) was developed, utilizing artificial intelligence (AI) technology and a standardized parent proxy questionnaire. The objective of this study is to examine the validity and reliability of the English TOY8 (based on existing validated Malay Toy8) and to examine measurement invariance across language versions, gender, and income groups. Methods: TOY8 was designed and developed to screen the developmental aspects in children aged 3 to 5 years old in two languages: Malay and English, by an interdisciplinary expert panel drawing upon both national and international guidelines. Two samples of parents and their children were recruited: 1767 dyad to complete the English Toy8 and another 1724 dyad to complete the Malay Toy8. Results: The CFA results indicate that the model structure of the English version of TOY8 matches that of the Malay version of TOY8. The split-half reliability coefficient indicated adequate to high reliability which is also consistent with the previous Malay version of TOY8. Our results showed that all configural and metric invariance model across groups showed good fit to the data demonstrating that a multiple-group CFA was appropriate. Lastly, scalar invariance was only achieved in certain domains across gender, not in the language version of income groups. Conclusion: The English TOY8 is found to be a valid and reliable screening tool to identify developmental milestones of children aged 3-5 years old in Malaysia. In addition, configural and metric invariances across groups in all domains were established indicating the cross-cultural equivalent of the items; scalar invariance was established across genders in most of the domains of 3-5 years old. This implies that our study aligns with previous literature regarding child development, which indicates that there is a general similarity in gender and cross-cultural development domains in the first years of life but not for older children in term of language and socioemotional skills.