This project addressed the themes of PBL and digitization and PBL and student retention during first year of study. It aims at enabling relevant stakeholders such as semester coordinator, student counsellors, study boards, and putting retention on better footing by predicting flailing students as early as possible to allocate resources to them.

Dropout and retention in the first semesters of university have been linked to a variety of factors in the literature including but not limited to high school performance, academic performance, demographics, motivation to study, and activities in virtual learning environments. From these factors the University of Eindhoven University has achieved between 75%-80% accuracy for predicting drop-outs in [1] in their electrical engineering degrees that traditionally have high dropout rates (40%) in the first year. They use this information to advise students at risk and to allocate resources. The same can be applied on per course level [2] – allowing for earlier feedback to students.

While much of the required information is being held by AAU, course and semester coordinators have no means of accessing this data nor obtain such an integrated analysis about which students are at high risk of failing and dropping out to direct resources at. Much of the necessary data is currently held by AAU. Medialogy is going to experiment with collecting relevant data regarding demographics and study motivation through the study verification test (studiestartsprøve, SSP). In an ongoing PBL development project (Improving Moodle for flipped classrooms to decrease drop-outs) we introduced new interactive activities to AAU’s Moodle for flipped classroom design to improve both student classroom participation and collect data performance on course participation and performance to facilitate analysis. The next version of Moodle (to be rolled out by ITS in November 2017) will support a course-based prediction plugin (pass/fail) given that courses have sufficient digital interactive activities that generate the required data from student use. These tools do not only serve the purpose of data collection but through their use can help students to reflect on their learning progress, e.g. learning inventories and other self-assessments that are very much in line with PBL practice.  ​

References:

[1] G. Dekker, M. Pechenizkiy, and J. Vleeshouwers, “Predicting students drop out: A case study,” Proc. Educational Data Mining, 2009.

[2] M. Delgado, E. Gibaja, M. C. Pegalajar, and O. Perez, “Predicting students’ marks from. Moodle logs using neural network models,” Proc. Current Developments of Technology-Assisted Educcation, 2006

ANSVARLIG

• Hendrik Knoche, hk@create.aau.dk, Dept. of Architecture Design and Media Technology IT and Design

ØVRIGE DELTAGERE

• Olga Timcenko, ot@create.aau.dk, Dept. of Architecture Design and Media Technology, IT and design
• Jon Bruun-Pedersen, jpe@create.aau.dk, Dept. of Architecture Design and Media Technology, IT and design
• Lise Busk Kofoed, lk@create.aau.dk, Dept. of Architecture Design and Media Technology, IT and design
• Torben Tvedebrink, tvede@math.aau.dk, Dept. of Mathematical Sciences, Engineering and Science
• Poul Svante Eriksen, svante@math.aau.dk, Department of Mathematical Sciences, Engineering and Science
• Brian Møller, brianms@its.aau.dk, AAU IT Services