We view adaptivity as just one application of a broader discipline – data-driven learning. Critical aspects on our views of data-driven learning fall into three areas:
- What data do we have (and how do we design learning to collect that data)?
- How do we model that data?
- What do with the information coming from our models to optimize learning?
What data do we have? We focus primarily on collecting and modeling learning data, which we distinguish from other forms of data that one might work with in an educational setting
- Learning Analytics: what are students learning? What level of mastery against the learning outcomes of the course have they achieved at any moment in time?
- Engagement Analytics: what are students doing? Are they logging in? When? What are they clicking? How engaged do they appear based on their behavior?
- Institutional Analytics: based on high-level results data, demographic data, and models, what do we know about the effectiveness of our courses, curricula, student support services, etc.?
More information on these types of data analytics: Read more: Analytics in Online Higher Education: Three Categories Watch the webinar: Q&A Panel Webinar: Essentials of Analytics in Online Higher Education
How do we model that data?
Every student interaction with Acrobatiq activities is recorded, whether the activities are summative checkpoints or practice questions. This data informs our model about changes in learning tied to specific skills or knowledge components related to those activities.
By using sophisticated statistical modeling in conjunction with sound principles of cognitive science, we can model their learning on learning objectives and even course-wide competencies. This allows us both to identify weak points for students and to give us the ability to target specific misconceptions. We think it is essential to have a precise and nuanced view of a student’s learning state relative to the desired outcomes if we are going to drive actions based on that data.
What do we DO with that data? (adapt the learning environment amongst other things)
Once we have a clear view of a student’s learning state relative to the desired course outcomes, we can leverage that data (often in real time) to improve learning outcomes and to create efficiencies in the teaching process. Examples of the kinds of things we do with that data include:
- Learning dashboards for faculty and mentors
Allowing teachers to see at a glance where to focus their efforts in the classroom or intervene with individuals or with small groups that need similar instruction
- Dashboards for learners
Giving insight to students on their progress in the course allowing, them to optimize their time spent
- Recommendations for learners
Based on those insights, providing guidance to learners on optimizing their experience by letting them know when it’s time to move on and where to focus their attention
- Adapt the environment to the learner
Using carefully designed questions that synthesize a number of objectives, we present a small scenario to students. A student who we believe has a good understanding of all of the component objectives may be presented with simple a couple of questions to validate their learning, before being encouraged to move on. If, on the other hand, we are presenting to a student that has struggled in some component skills, we scaffold that student up front within the context of the scenario by presenting additional questions targeted to their learning needs. This is different from scaffolding a problem simply based on the student giving a wrong answer. We optimize the experience to the student need.