Bringing child development research online at scale

Childhood occupies a significant chunk of the human lifespan. This extended period of development is where many uniquely human traits emerge: language, social cognition, and literacy all develop gradually over the course of childhood and all vary across individuals and groups. Since development entails change over time, understanding these traits requires a view that spans multiple time points. A full understanding of human development involves measuring a diverse set of children longitudinally. To accomplish this, we need a tool that is standardized, widely accessible to researchers and participants online, and easy to use.

LEVANTE is designed to help researchers better understand childhood and development all over the world. LEVANTE provides a ready-made research tool for advancing our understanding of variation in child development. We’ve built a full-stack open source tool that researchers can use for free, without any web-development background. While the tool is currently in limited release to a handful of sites, we aim to make this available to all interested researchers sometime in 2026. The tool supports supervised data collection with children as young as 2 years old on any internet-connected tablet or computer. The platform also supports tracking children over time ethically and anonymously – allowing for researchers to test their task performance across many years. Our tasks span cognitive capacities from literacy and numeracy to social cognition and executive function.

Many large scale longitudinal studies struggle with data loss, participant attrition, and accommodating different methods of data collection and recruitment. To address these issues, we’ve built a researcher dashboard to work for researchers in human development across a variety of recruitment contexts. We already support sites collecting data in schools, traditional labs, and homes.

To accomplish this, we were lucky enough to have a working model: the Rapid Assessment of Online Reading (ROAR), led by Jason Yeatman at Stanford. ROAR had already developed a researcher dashboard for internet-based reading assessment in schools. In collaboration with ROAR, LEVANTE has created a “fork” of their researcher dashboard and adapted it for the kind of longitudinal research our international sites carry out. We are very grateful for ROAR’s support and continuing partnership.

How does the dashboard work?

Researchers at individual sites manage recruitment in their own way – they may be visiting schools, inviting families to a university lab, traveling to family homes with portable devices, or recruiting children and families online. The researcher dashboard supports this diverse range of approaches by having sites manage their recruitment data locally. The dashboard provides a ‘Link Users’ function which links child data to information about their home and educational environment provided by their caregivers and teachers.

Beyond tracking individual users over time, the dashboard includes tools for grouping users: each location is tied to a site, which can include many schools and classes, or cohorts for sites taking a different approach to recruitment. Groups can then be given assignments which have specific tasks, allowing a site to test and re-test children flexibly. Assignments can use customised bundles of some or all of our 13 tasks, including language tasks from the Rapid Online Assessment of Reading (ROAR), computer adaptive math tasks, and a suite of social cognition and executive function tasks.

Privacy and ethics

Combined with linking users together, the kind of longitudinal tracking the platform provides presents some ethical and privacy-related concerns. How can we keep track of participants and release open data without compromising privacy and anonymity? LEVANTE only stores anonymous ids and basic information about participants. We estimate a child participant’s age using their birth month and year and child participants are linked to caregivers and teachers through their anonymous ids. Any identifying information required – for example, contact information needed for longitudinal data collection – is stored separately by researchers, and never connected to the LEVANTE dataset.

A tool with this level of flexibility allows for research teams to address a wide variety of questions related to child development and cognitive science. For example, researchers at the Donders Institute in Nijmegen are using the LEVANTE platform to investigate the relationship between mother-infant skin-to-skin contact and later cognitive development. A team based at Universidad Torcuato di Tella in Buenos Aires, Argentina will focus on peer relationships: exploring how children learn by teaching themselves and others at school. We also have sites using the tool in Canada, Germany, Colombia, and the US – with more sites to come.

The biggest challenge ahead of us is continuing to adapt the researcher dashboard for a geographically and scientifically diverse user base. We want the tool to be usable in many places all over the world, across a range of diverse recruitment contexts. This comes with significant challenges in terms of translating and adapting our tasks, but also optimising tasks for a range of devices, and potentially intermittent internet connectivity in remote areas. For example, in Colombia, we’ve worked with researchers to optimise our task audio and image assets for delivery via Starlink.

We continue to anticipate challenges as new researchers adopt our tools. For example, we are thinking carefully about how to accommodate areas with low rates of adult literacy. As we continue to support a broader range of sites and researchers, we look forward to continuing to build a tool that finally makes large-scale online research in child development accessible.