Berkeley edX

Adapting an existing on-campus course for BerkeleyX (i.e. into an edX MOOC or SPOC) is a condensed HOWTO with pointers to more details in each area of course creation and delivery.

This information is for developers who want to create custom functionality for a particular course or research project (see below) based on the edX platform.

• The edx-dev group is for people across campus hacking edX directly. We meet weekly and sometimes there's free food. The requirement is you sign up to present something useful at the meeting.
• Developer-facing wiki on GitHub has info on installing the platform and documentation on its evolving APIs.
• ReadTheDocs documentation on both how to create a course and how to get/work with learner data; you can browse online, download as PDF, or download as an ebook.
• Sef Kloninger's Five Ways to Extend edX gives a good overview of the different ways to think about customizing the platform.
• All About Autograding has some specific info on building an external “custom grader” (one of the Five Ways described in Sef's overview, sometimes referred to as an “external grader” or “external checker” in the official edX developer documentation).
• RuQL is a Ruby-embedded domain-specific language for creating quiz questions that can be exported to edX, to a printable quiz, or to AutoQCM, an open-source tool that lets you create scantron sheets and grade scanned filled-in scantron sheets.

Generally speaking, you must:

• Be a Berkeley Principal Investigator, or a Berkeley student working with a Berkeley PI, or a Visiting Scholar or Visiting Industrial Fellow/Researcher working with a Berkeley PI
• Have completed Human Subjects training. Berkeley-affiliated researchers can do this online: follow the instructions for "Online Training" on this page.

In general, an edX MOOC produces three different data sources:

• MySQL databases store info about students, their grades, and earned certificates.
• The clickstream log or event log, which has an event for every host-side or client-side (typically Javascript) action. Each such event is a JSON object. There's a folder for each server, and under that a file for each calendar day, so getting all events for a calendar day requires examining that day in each folder. The number of events is very large so most applications will pre-filter it with a script or with a simple tool such as grep.
• The MongoDB databases that store the edX forum data (discussion boards)–the full text of posts, replies, and so on. Importing them into Mongo is straightforward using mongoimport.

To learn more about the data:

• The above are described in the Internal Data Formats section of the edX developer documentation.
• Graduate student Kristin Stephens-Martinez has created "Where in the world is that piece of data?", which discusses in more detail the terminology used to describe the structure of an edX course and how those terms are related to the names of database tables and fields used to store course data.
• These scripts (not maintained) import “raw” edX course data into local databases; as they go, they anonymize the data by removing personally identifying information and re-encoding user IDs.
• edx2bigquery lets you import edX data dumps to Google BigQuery, and also to Tableau, which has a BigQuery connector module.

There's also a variety of user-contributed tools (both for data management and other stuff) on the edX Tools wiki.

Journals

• RPA (Research and Practice in Assessment)
• Educational Researcher
• Transactions on Computing Education
• Computers & Education

Sites

• MITx working papers
• HarvardX working papers
• Lytics Lab Publications (Stanford)
• MOOC Research Hub Reports (Bill & Melinda Gates Foundation)

Books

• How Learning Works (pdf)
• Handbook of Research on Learning and Instruction (google search, first hit)