What mediates the effects of wealth on children's vocabulary developmental trajectories in Ethiopia? A longitudinal analysis

Thesis

This D.Phil study is motivated by gaps in educational research in low-income countries. First, it uses a child developmental, rather than economic, model to investigate how the timing of poverty shapes children’s vocabulary trajectories and the mediational pathways through which poverty influences growth at different stages of childhood. Second, it investigates the reliability and validity of the measure used to assess vocabulary, addressing inter-cultural issues that have not been dealt with in the literature and challenging some of the assumptions underlying the use of psychometric approaches in low-income countries. Data from a sample of 1,321 children in Ethiopia, who took part in the Young Lives longitudinal study at ages five, eight, twelve and 15 years, were analysed. The Peabody Picture Vocabulary Test (PPVT) was used to measure vocabulary ability. A composite wealth index was used to measure living standards.

The psychometric properties of the PPVT were analysed within a Rasch measurement model. The test was screened for unidimensionality, item targeting, differential item functioning and scale stability. To compare the scores obtained at different time points, scores from the repeated test were calibrated onto a single metric using a test equating procedure. The Rasch estimates were obtained by anchoring the scores on the most stable items within each of respective language groups: Amharic, Oromifa and Tigrinya.

A multilevel model for change approach was used to estimate individual growth trajectories. On average, children in the highest wealth quintile outperformed children in the lowest quintile. The difference in the mean growth trajectories between top and bottom quintiles followed a constant path, except in Amharic in which the average gap narrowed by 0.47 SD. Initial wealth, measured at one year of age, predicted onset ability at five years for all language groups, and the growth rate between 5-15 years in Amharic. Changes in wealth did not further help to explain the variance in the intercept or the slope.

The analysis was extended to account for the factors that mediate the poverty-growth relationship. The results showed that 0-10% of the wealth effect on onset ability was mediated by the height-for age z-score, a measure of nutrition. Mothers’ education mediated 1-8% of the wealth effect. Preschool participation mediated 6-14%. The findings reinforce existing literature about the importance of these factors in mitigating the effects of poverty on children’s vocabulary development, particularly in early childhood. The study further highlights that expanding this knowledge base requires an in-depth understanding of the contexts in which children develop and learn, drawing on insights from substantive theories and local understandings of development and learning.

Authors: Tesfay, N

• DOI: 10.5287/ora-n6rypmzw5
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: vocabulary assessment; child poverty; cross-cultural assessment; developmental trajectories; Ethiopia; Rasch model; educational measurement; mediated wealth effects; vocabulary development; models of child development; wealth effects
• Subjects: Child development; Education; Poverty