In the realm of online education, one of the challenges faced by educators is the issue of false positives in identifying at-risk students. These are individuals who are flagged by the system as being at risk of dropping out, even though they may not actually be in danger of doing so. This problem often arises with students who are not consistently active in the virtual learning environment. To address this issue, a new model has been developed that takes into account a temporal window that is automatically calculated based on the course, type, and difficulty of the activity. In order to confirm that a student is truly at risk of dropping out and activate the necessary intervention mechanisms, they must remain in the dropout risk category for a consecutive number of days determined for each activity. If a student is deemed to be at a high risk of dropping out, an automated intervention message is generated for the student.
The goal of the intervention system is to increase student motivation by providing personalized automated messages. These messages may include time management recommendations, short-term goal setting, information on the negative consequences of not completing an activity, as well as supplementary learning materials and exercises to help students achieve their objectives. Additionally, instructors have the ability to design and customize the content of these messages, tailoring them to the level of risk identified. The system also offers various dashboards that allow both instructors and students to individually track their current status and potential risks.
To evaluate the effectiveness of this new predictive model, a comparison was made between student dropout rates in courses where the system was used and those where it was not. The results, published in the International Journal of Educational Technology in Higher Education, co-edited by the UOC, showed a significant decrease in student dropout rates at the end of the course across all activities. There was a 12% difference in dropout rates between participants in the pilot study and non-participants, and a 5% difference compared to the previous semester when only the LIS system was used without the new predictive model.
This new system empowers instructors to proactively address student issues. It allows for early detection of problems and enables round-the-clock monitoring of students. Moreover, it is a scalable tool that enables instructors to effectively manage large classes without overwhelming themselves or their collaborators. For example, a pilot study was conducted in a course with 1,500 students, and the system allowed instructors to monitor at-risk students without overburdening themselves.
An important advantage of the LIS system is its adaptability to any online learning environment. It is not dependent on a specific learning management technology, making it compatible with various platforms as long as there are historical academic data available. This flexibility allows for the system to be configured for each course, adapting to the activities involved and training the necessary prediction models using data from previous students who have taken the course.
In conclusion, the research conducted by the UOC supports the goal of sustainable development (SDG) 4, which focuses on quality education. This innovative approach to identifying and intervening with at-risk students in online education demonstrates the potential for improving student outcomes and retention rates. By leveraging technology and data-driven insights, educators can better support their students and enhance the overall learning experience.