Huntington, Bethany S
(2024)
Predictors of foundational learning for out-of-school children in Tanzania with interactive apps deployed direct to communities.
PhD thesis, University of Nottingham.
Abstract
The world is experiencing a global learning crisis, exacerbated by the COVID-19 pandemic, which has left over 393 children without basic literacy and numeracy skills. Educational technologies (EdTech) that promote autonomous learning may ameliorate this learning poverty. However, little is known about the mechanisms through which out-of-school children may acquire foundational skills using tablet-based technology and app-based learning software.
This thesis aimed to identify potential app, child, and community-level predictors of foundational learning outcomes following a large-scale EdTech intervention - the Global Learning XPRIZE (GLXP) competition - directly deployed to out-of-school communities in Tanzania. An exploratory, mixed-methods approach was adopted to inform a theoretical model of out-of-school app-based learning by identifying predictors at three different levels of explanation.
First, a comparative judgement experiment (Chapter 3) was undertaken to investigate app-level predictors, in which 41 non-expert participants compared five learning apps across 15 key pedagogical features. Results indicated that six pedagogical features were found to be most influential for learning—autonomous learning, motor skills, task structure, engagement, language demand and personalisation. EdTech has been shown to facilitate learning for girls, an at-risk population in low-income countries, so further research was needed to determine what app features may be most influential across genders. A gender and domain app study (Chapter 4) used machine learning methods and inferential statistics to identify which app features most predict girls' and boys’ literacy and numeracy learning. Some app features, such as retrieval-based learning and engagement, were found to be broadly influential for learning. Five app features - engagement, autonomous learning, language demand, personalisation, and curriculum links- showed a differential influence across genders and domains.
A machine learning regression approach was adopted to explore contextual predictors (Chapter 5) of literacy and numeracy learning. Child and community-level features were leveraged using competition survey and assessment data, and contextual covariates derived from open-source geospatial data. Reading habits, small family size, and high social connectedness were shown to be the most predictive factors for improved learning outcomes. Community factors were found to be more predictive of learning improvements than child-level features, highlighting the importance of infrastructure.
Results from the GLXP imply that EdTech deployed to out-of-school children living in remote villages can support autonomous learning. However, no qualitative data regarding implementation was gathered during the competition to draw firm conclusions. Therefore, an expert elicitation (Chapter 6) explored the broader implementational impact and challenges experienced, whereby 14 key informants were interviewed about the autonomous learning process. Four key themes were generated: ‘Technology as a novel concept’, ‘Children don't learn in a vacuum’, ‘Respecting the cultural context’ and ‘Accessibility problems in a mobile world’. Community support was prominent throughout the competition, emphasising its importance for learning with EdTech and raising questions about whether out-of-school children can learn autonomously with this technology.
Key implications and recommendations are outlined for technology developers, educators, and policymakers to consider when designing and implementing app-based learning interventions for out-of-school children. If influential mechanisms are carefully incorporated into the design and implementation of EdTech interventions, they may support even the most marginalised learners.
Actions (Archive Staff Only)
 |
Edit View |
Tools
Tools
