Course Signals: Purdue University

Synopsis

Course Signals is a predictive learning analytics system originally produced at Purdue University in the USA. The system uses student data to predict those who are at risk of not successfully completing the course. By using predictive modelling of student data and activity in the learning management system (LMS), each student is assigned a ‘risk group,’ the colours of which are those of a traffic signal – red, yellow, or green.

To use the system, the lecturer or tutor must manually run the model to receive students’ ‘signals’, which they can then use to provide targeted feedback or additional resources to those at risk of low performance. Course Signals incorporates the use of intervention emails, which can be written by the teacher and sent to those in each risk group. Notifications can also be given in a student’s LMS course page.

Course Signals allows teachers and tutors to give real-time feedback starting as early as the second week of class, and it can be used by the lecturer at multiple points during the term. In research published at LAK12, it was suggested that there was a 21% retention rate improvement at Purdue between students who took at least one course that used Course Signals, over those who did not. However, this has since been disputed.

Classification

Inventory type:

example at scale

Keywords:

predictive analytics, predictive modelling

Context of Practice

Learning:

post-compulsory

Geographical:

national: United States

Pedagogic:

Purdue Course Signals is not explicit in support of one pedagogic framework over another. This institutional practice relies on predictive modelling of student data to categorise at-risk students.

Practical Matters

Tools used:

Course Signals uses early detection of ‘at-risk’ students to alert teachers of the need for intervention and additional feedback. Data used by Course Signals include student grades, demographic information, academic history, and use of the learning management system.

Design and implementation:

This system was produced at Purdue University in the USA. It uses student data from Blackboard, although no explicit connection between developers of Course Signals and Blackboard is described.

As of 2012, over 2,300 students in more than 100 courses had used the system. At that time, it was suggested a further 20,000 students would gain access within the next 18 months. However, more current data has not been made available.

At present, courses at Purdue are not required to use Course Signals, thus it has not been mobilised yet on an institution-wide scale. Lecturers may choose to adopt Course Signals within their own courses, but the project website suggests it is most effective for classes with over 50 students.

Maturity and Evidence of Utility

Course Signal’s effectiveness was highlighted in a paper presented at LAK ’12, claiming a 21% improvement in the retention rate of students who took at least one course that used the programme. However, criticisms have been made about the methods  underlying these claims. As no follow-up studies have yet been published, it will be necessary to address these issues to demonstrate maturity and utility of the system.

Further Information

Informal accounts: http://bit.ly/22S5InZ, http://bit.ly/1Saf8aK

Comparison of Course Signals and Blackboard Retention Center: http://bit.ly/1Rf95l0

Criticisms of claims http://bit.ly/22S2K2Q, http://bit.ly/1OGwsSd

Academic study:

Arnold, Kimberley E, & Pistilli, Matthew. (2012). Course Signals at Purdue: using learning analytics to increase student success. Paper presented at LAK12, Vancouver, Canada.

See also LAEP Inventory record:

• Progress and Course Engagement (RioPACE) – Rio Salado College