As academic technologies are increasingly incorporated into online and onground classrooms, the data collected about student interactions with these systems can create valuable insights for administrators, faculty/teachers, and students themselves. In this presentation, we will discuss highlights of recent Blackboard research in learning analytics, and demonstrate solutions that integrate these findings. By using learning analytics, faculty can better understand how students are interacting with course materials and use this understanding to increase student engagement, ultimately leading to better student outcomes and retention.
Presented by John Whitmer, Rachel Scherer, Marlen Rattiner, and Jesse Lyman
In 22% of courses, there was a significant relationship between Learn use and student achievement, but the amount of predictability varies quite a bit between courses. Most courses had R-squared values (effect size) between 0 and .5.
Student access to grades is an indicator of achievement – at every level, probability of higher grade increases with increased use of the Grades tool. This is likely not causal, but it is a good indicator.
Retention Center sees low usage, so the goal with analytics is integrating it into the core workflow of teaching and learning. For example, notifications to students within the Course Stream.
Similarly, using analytics and measures of quality (critical thinking, grade level, etc) and quantity to provide a recommended grade. Coming in September 2017!
Surfacing Patterns with Analytics for Learn
Analytics for Learn, part of the Intelligence suite of products
Includes reporting, analysis, data warehouse, and integration services. The value is in the structure provided by A4L.
Descriptive Analytics vs. Predictive Analytics
What happened? Why did it happen?
What WILL happen?
The power of predictive analytics is using what is known to predict that there will be a problem and intervene earlier for greater effect. Bb Predict uses years of data from the institution so that every model is uniquely adapted to the local context. Model changes over time, too, based on what is known about the student.
For example, an early semester prediction might be based on what is known about similar students and past performance. A mid-semester prediction would be more heavily weighted on current grade and LMS activity.
For Advisors, this is about providing accesss to data they have never had before, so they can provide more direct support and intervention. For Instructors, it’s about providing data and analysis embedded within their workflows, so that they can know their students better. For students, it’s about showing them how they are doing individually and compared to their peers. Instead of a predicted score, though, Bb shows a grade projection chart that shows the best they could do, the worst they could do, and how they would score if they kept the same pace and behaviors.
Tailoring the Analytics Experience
Bb Consulting is able to work on customizing data prediction and modeling when an existing product doesn’t meet the need. For example, they helped to build a custom student dashboard to focus on retention. They have also built “micro-analytics” within Learn to nudge student achieve more through comparison with others in the class, such as number of posts on a discussion forum, within the forum or content item itself.