Combining expert judgment and psychometric information to create the Global Scale for Early Development (GSED)

McCray, Gareth; Gladstone, Melissa; Cavallera, Vanessa; McCoy, Dana; Waldman, Marcus; Eekhout, Iris; van Buuren, Stef; Janus, Magdalena; Kariger, Patricia; Black, Maureen; Weber, Ann

Abstract

Background: The Global Scale for Early Development (GSED) a global and culturally neutral measure of child development under 3 years of age is designed to allow comparisons within and across countries. The project has been undertaken by an experienced team of international researchers led by the World Health Organisation. Ultimately, the GSED will comprise two tools anchored on the same scale: firstly, a caregiver reported measure aimed at population level monitoring, and secondly, a direct assessment tool to be used for programmatic evaluation. Aims: The initial GSED tools were designed using a data-driven paradigm. Data from over 70,000 children, assessed with 22 different instruments and using 2275 items were collated. A modified Rasch model was used to jointly model the data and select a subset of 807 items which showed promise of stability across countries. The psychometric performance of items is not sufficient for principled item selection and subject matter expert (SME) judgement data must be collected and synergised with the item parameters for robust item selection. The aim of this study was to create a novel, robust and reproducible methodology to collect and combine SME judgement about items with the psychometric properties of those items to streamline item selection for the creation of new global measurement tools. Methods: Eight internationally renowned SMEs in child development measurement provided inputs relating to three main areas. Firstly, data were collected on the extent to which groups of items measured the same behaviours. This was important so we could be cognizant of overlap between items in the final tool. Secondly, the domain of child development that each of the items measured was probed. We did this to ensure content validity in the final tool. Thirdly, global feasibility, relevance and appropriateness data was gathered for each item. Some items were clearly unsuitable for our purpose and an opportunity was given to the SMEs to flag those they felt did not meet the requirements to be included in the prototype for field testing. The Judgement data were combined with the psychometric data using an R-ShinyApp (online-dashboard) which facilitated the selection of the items for the final tool. Results & Conclusions: Two prototype tools were constructed to be tested in the field—the GSED Short Form, the caregiver reported tool, and the GSED Long Form the programmatic evaluation tool. The GSED is currently being validated gathering data from more than 8700 participants across seven countries. Disclosures: The authors declared no competing interests.

Citation

McCray, G., Gladstone, M., Cavallera, V., McCoy, D., Waldman, M., Eekhout, I., …Weber, A. (2021, June). Combining expert judgment and psychometric information to create the Global Scale for Early Development (GSED). Paper presented at Proceedings of the Patient Reported Outcome Measures (PROMs) Research Conference, Sheffield, UK

• Presentation Conference Type: Conference Paper (unpublished)
• Conference Name: Proceedings of the Patient Reported Outcome Measures (PROMs) Research Conference
• Conference Location: Sheffield, UK
• Start Date: Jun 16, 2021
• End Date: Jun 17, 2021
• Deposit Date: Jun 21, 2023
• Publisher URL: https://www.iriseekhout.com/publication/mccray-2022-combining/