• COVID-19 has had a big and lasting effect on education.
  • The pandemic has powered the adoption of educational technology, but the impact of EdTech on learning outcomes has been mixed.
  • If policy makers wish to establish an adaptive learning system, 5 requirements must be met first.

Educational technology, or EdTech, has been a promising avenue to address some of the most challenging policy questions within educational systems in low- and middle-income countries. However, after decades of promises to disrupt and revolutionize education, the impact of EdTech on learning outcomes has been mixed. The drive to provide tech-based support to improve teachers’ instruction and facilitate student learning has become urgent, given school closings and re-openings due to the COVID-19 pandemic. One of the few silver linings of the pandemic, is that it has transformed the field, shifting the focus from disruption to mitigation and inclusion, with the potential to reduce inequity.

The evidence-base for what EdTech interventions work for improving student learning outcomes in low- and middle-income settings is limited. A recent systematic review (by Rodriguez-Segura, 2020) of existing credible studies found that technology as a standalone intervention will not work to improve teaching practices or student learning. Of all the EdTech interventions included in the review, those focused on self-led learning were of the most salient ways to improve student learning outcomes.

Facilitating self-led learning through an adaptive learning system

Self-led learning, or the ability for students to take charge of their learning, at their own pace or level, can be a game-changer during (and after) the pandemic. Self-led learning can be developed without technology but with intensive teacher support. However, Rodriguez-Segura explains that because EdTech interventions can enable students to learn at a fitting pace with minimal external support, they seem particularly enticing. One example are adaptive learning systems that respond to interactions in real-time by automatically providing students with individualized support. One common challenge teachers face is tailoring their instruction to each student’s learning needs. Teachers can address this challenge by differentiating instruction and dividing the classroom into smaller groups based on students’ learning levels.

COVID-19 Education and Skills Technological Transformation
A model of adaptive learning.
Image: Dreambox Learning

Adaptive systems can automatize this process by modifying the presentation of material in response to student performance. The systems (can) “learn” from student progress and adjust the path of learning. In some cases, they have been shown to be cost-effective and can sharply improve productivity in delivering education, with the potential to complement classroom instruction, reinforce lessons, and fill in content gaps. This methodology is closely related to rigorous evaluative research that has shown how targeting teaching instruction by learning level, not by grade, can lead to significant learning outcomes.

Five enabling conditions for establishing an adaptive learning system

Adaptive systems’ primary function is to provide remedial education and help students improve curricular mastery. They can complement instruction when schools are open and can be used by students independently when schools are closed. For instance, in Ecuador, adaptive systems have helped prevent student dropout during school closures. However, adaptive systems are not a “silver bullet”. In order to work, there are several enabling conditions that must be established.

1. Most adaptive systems are proprietary and require high upfront costs to develop or adapt. Adaptive systems have only been developed for a few subject areas (i.e. math and early reading) and are currently limited to these subjects. Thus, when considering an adaptive system, policymakers have two main options: (i) adapt a pre-existing, proprietary system for their context or (ii) develop a system from scratch. Both have their drawbacks, the former involves adapting and translating a software that is likely only available in English/Spanish and mapped to a foreign curriculum; policymakers must also pay licensing fees for users to access the platform, which can incur ongoing costs. Both options have a host of associated costs, including translations, server maintenance, training, software updates, a support desk, among others.

2. Regardless of whether a government adapts or develops a system from scratch, adaptive systems must be mapped to the curriculum. Adaptive systems require a detailed curriculum mapping and content development to support the learning objectives outlined in the curriculum. If a government decides to adapt a pre-existing, proprietary system for their context, they may run into issues in changing the content to match their curriculum. If they develop a system from scratch, they can either outsource the curriculum mapping or embed this step in the planning process. Irrespective, it is essential for policymakers to involve teachers in this process, which has been shown to improve teacher uptake and buy-in during implementation.

3. Adaptive systems require a robust digital infrastructure to ensure widespread adoption. Although evidence suggests the impact of adaptive learning can be more acute for students of lower socioeconomic status, and thus minimize learning losses, it’s often the case that the students who are most at risk of learning losses often do not have access to online solutions. Thus, it’s imperative for teachers and students to access the system, both by having a device with the appropriate content on it and wireless capability to connect to that content. Although the price of tablets and smartphones has dropped significantly, and low- and middle-income countries have invested in access to affordable, high-quality data, adaptive systems require wireless devices and strong bandwidth, which could incur high up-front costs (and other associated cost like technical support, maintenance, monitoring, among others).

4. Adaptive systems must include training for teachers to be deployed effectively. Teachers’ willingness to adopt new practices and buy-into the system play a major role in determining the efficacy of the program. Adaptive systems are not meant to replace teachers, but rather enhance their role (see “Reimagining Human Connections”). It’s important that teachers are trained on what is expected of them, as without an understanding of this role shift, policymakers risk disenfranchising them. Specifically, teachers should be trained on how to: (i) use the platform to shift the role from an instructor to a tutor, (ii) utilize and access the platform and associated technology, (iii) use the technology for students’ knowledge development, (iv) utilize the findings from students’ participation on the platform to automatically provide individualized support to students, (v) utilize the data from the platform to plan future lessons and differentiate instruction based on students’ needs. This training should be paired with an engagement plan, in which teachers can share good practices and promote the benefits of the platform through different channels and contexts.

5. In addition to teacher training, adaptive systems must effectively engage students. Like teachers, it’s important to engage students in the rollout process. Engagement initiatives targeted for students aim to create intrinsic and extrinsic incentives for them to use the platform consistently. One such program nominated a number of students who received training on how to navigate the platform, handle connectivity errors, troubleshoot technical issues, and contact the helpdesk. An additional approach can be to gamify the experience, students can earn ‘points’ (or other form of social recognition) based on the usage or progression on the system per week (top performers can be recognized among peers). Similar approaches can be used with teachers.

When these enabling conditions are met, adaptive systems have the potential to improve how teachers instruct, enhance student learning, and help policymakers to better understand what is happening in classrooms (or outside of them). At a minimum, these systems require (i) adequate curriculum calibration, (i) initial assessment, (ii) customization of the instructional process, and (iii) ongoing monitoring. However, for them to work, these systems require an extensive rethinking of the role of and dynamics between students and teachers. Even in the best-case scenario, these changes take substantial time to roll out.

The COVID-19 pandemic has shown us that we can no longer rely on traditional forms of schooling. It is increasingly likely that in the future of learning, technology could diversify the means to support students. As shown, adaptive systems could be an opportunity to support self-led learning as well as other forms of learning (making it more accessible, impactful, and engaging). However, given the complexity of adopting and deploying adaptive learning systems, education systems must address the five basic enabling conditions. Without these conditions in place, adaptive systems will not only underdeliver, but can also be a costly lesson for low- and middle-income countries.

Special thanks to Diego Angel-Urdinola, Iñaki Sanchez, and Mike Trucano for their insightful contributions to this blog.