[Anonymized project name]: Digital Learning Platforms as Research Infrastructure (Poster 4)

• In Event: Advancing Methods, Theories, and Conceptualizations of Infrastructuring in the Learning Sciences

Abstract

Objectives: As researchers turn to infrastructure, so have schools. More learning is now online, and more of online learning is centralized on platforms which collect data. The central objective of the [project] network is to enable researchers (together with practitioners) to define and conduct improvement-oriented research using the large-scale platforms. To this end, [funder] funded five platforms to open their capabilities to research teams, and research teams have applied for funding to use those capabilities. More generally, similar efforts to use digital learning platforms (DLPs) as research infrastructure are emerging via funding from the Gates Foundation (the AIMS Collaboratory) and NSF (the SafeInsights midscale infrastructure). We broadly reflect on the promise and the challenges of this movement.

Theoretical Framework: [Project] can be seen as rooted in four long-term evolutions in educational research (Schellinger & Roschelle, 2024). First, a body of knowledge has developed on how to organize data repositories for secondary analysis, and more recently, how to support open science. Second, researchers began to collect and model fine-grained learning process data that arises in DLPs. Third, teams have created experimental interfaces to standardize how researchers interact with platforms to conduct comparative research. Fourth, conferences and societies have emerged, such as Learning Analytics and Knowledge, Educational Data Mining and Learning@Scale, and these convene to advance and publish the work. [Project] and similar efforts draw on all four evolutions; they bring them together in a new way to enable research about innovations that could scale up through the medium of widely-used DLPs.

Methods and Data Sources: We broadly reflect on our observations as network organizers.

Results: The funder initially imagined research that would be quicker, less expensive, and conducted on a larger scale because DLPs were used as research infrastructure. Those have not turned out to be the most salient aspects of the work. The opportunities that have arisen bring specialized expertise into a platform setting, creating an ability to work on an improvement that the platform-owner would not have the time or expertise to conduct. Overall, this can look like extending and replicating a theoretically-driven insight to a practical setting. Another opportunity has been more quickly testing new technological capabilities (e.g. LLMs) in a realistic context that already pre-exists. Further, when things work, there can be an easier path to scaling them when the research is done in the context of a platform operating at scale. However, there are challenges, which could be summarized as a transfer problem: it is not straightforward to transition from going from old ways to new ways of research. At each step, such as problem definition, research design, data access, etc., teams need to learn new things to effectively utilize DLPs. Another difficulty has been ensuring incorporating a broader view of the educational system into the research.

Scholarly significance of the study or work: Using DLPs as research infrastructure could result in a learning sciences that has clearer path to broad impacts. However, the field will need to evolve and transform to realize these opportunities.

Authors

• Jeremy Roschelle, Digital Promise Global

• Xin Wei, Digital Promise

• Adam Schellinger, Empirical Education Inc.

• Jenna L. Zacamy, Empirical Education Inc.