MOOCs

Instructure Opens Up Canvas Network Data

• By Rhea Kelly
• 03/02/16

Researchers can now access de-identified learning data from more than 230 massive open online courses (MOOCs) hosted on Instructure's Canvas Network. The company has opened up the Canvas Network data — representing thousands of learning experiences from around the globe — to support research on online course design, large-scale informal learning and more.

"We're excited to give researchers this valuable insight into our Canvas Network data as a means of furthering the impact of open education," said Melissa Loble, vice president of partners and programs at Instructure, in a press release. "Canvas Network has always encouraged creativity and experimentation with MOOCs to create more engaged learning. By sharing data we've collected with the educational community, we're providing another tool to help improve learning."

The data comes from Canvas Network open courses running from January 2014 to September 2015 — totaling more than 325,000 aggregate records, with each record representing one individual's activity in one of 238 Canvas Network courses. The dataset's structure is based on the HarvardX-MITx Person-Course data release of 2014 (the data release has no affiliation with Harvard or MIT).

"Our goal is to help educators and designers improve teaching and learning, and providing data from our Canvas Network is a great way to start," commented Jared Stein, vice president of research and education at Instructure. "This information will allow researchers to ask questions about online course design and user behavior from the perspective of large scale, non-formal learning."

According to the company, the data in this release has been anonymized and aggregated to protect user privacy, and does not include Canvas client data or matriculated student data. Researchers who agree to terms of use will be able to access a more robust data set.

The Canvas Network dataset is available on Harvard's Dataverse service.