Psychometric Properties of the Early Childhood Development Index (ECDI) Used Among 3- to 4-Year-Old Children in Nigeria

Background/Purpose: Early childhood is associated with the rapid development of physical, motor, cognitive, language, and social skills. Landmark studies suggest that millions of children in low and middle-income countries (LMICs) are at risk of poor cognitive development. Nigeria accounts for over 31 million children under five of which over 6 million are estimated to be at risk of not reaching their developmental potential. Past studies note a dearth of culturally and context specific early childhood development (ECD) assessment instruments in LMICs. Nigeria’s diverse language, education, socioeconomic and cultural nuances may present a challenge for direct application of instruments used in western cultures. To our knowledge, there are no studies reporting the performance of the 10-item Early Childhood Development Index (ECDI) in the Nigerian context. This study assessed the performance of the ECDI used in the fifth round and 2016-2017 Multiple Indicator Cluster Survey (MICS) to track the developmental milestones among 3- to 4-year-old children in Nigeria.

Methods: Representative data was collected from mothers who responded to ten ECD questions (yes/no) that examined four domains namely the literacy/numeracy (3 items), physical (2 items), the cognitive development/approaches to learning (2 items), and the socio-emotional (3 items). The child’s age (years), gender (male or female), place of dwelling (rural or urban), maternal (level of education) and household characteristics (wealth index, religion) were also examined. This study used the Item Response Theory (IRT) modeling to examine the differential functioning of each item and the degree to which the items measured the desired ECD construct. Exploratory Factor Analysis (EFA) was used to highlight the factor structure, while a Confirmatory Factor Analysis (CFA) examined the dimensionality of the ECDI domains.

Results: The sample of 11,207 children (3-year-old, 51% and 49% female) resided primarily in rural areas (74%). Most responding mothers were between 30 to 39 years (44%) and had either non-formal or primary education (37%). The Cronbach’s alpha of the 10-item ECDI was 0.68 when children are three years and increased to 0.71 for 4-year-olds. The principal axis factoring (paf) EFA and parallel analysis supported a 4-factor model, with the first factor accounting for 75% of the variance. The CFA confirmed a 4-factor model as opposed to a 1-factor solution and yielded adequate fit and minor modifications (χ2(29) 503.25, p<.001); RMSEA = .067; CFI = .94; and SRMR = .05). Results from the one parameter logistic IRT model yielded an item discrimination parameter of 0.96, suggesting that the items are not particularly discriminating, and a poor indication from the test of item fit. The overall Test Characteristic Curve (TCC) fulfilled the IRT monotonicity condition.

Implications: The confirmation of a 4-factor model of the ECDI suggests its better utility for assessing specific domains rather than ECD overall. Further, the weak correlations and low discriminatory power of some of the items suggest a review of the criterion validity. Understanding the performance of the ECDI will be instructive as countries await the launch of the new ECDI2030 and will provide a benchmark for assessing changes related to the update.