Self-Directed Learning in an Engineering Program - Student Planning and Bingeing

Background

In February 2016, Charles Sturt University (CSU) in Australia, received its first intake of students in a new Civil Systems Engineering degree. As the only Australian engineering program based in a Faculty of Business, we set ourselves the goal of doing something different to standard engineering programs. We would educate a very different type of engineering graduate and do so in a very different learning environment.

When developing the new engineering course CSU was mandated to produce a program that was orthogonal to existing Australian engineering degree offerings. After careful consideration of existing engineering programs, we developed a combined Bachelor of Technology and Master of Engineering degree with several key points of distinction (Morgan & Lindsay, 2015). The program would be based in the College of Business and focus on producing entrepreneurial graduates, with students completing 18 months on campus, followed by 4 x 1-year work placements in combination with studying the underlying theory online.

Given the constraints on students, particularly during their work placements, learning would have to take place in an asynchronous manner using a self-directed model. This model is the primary differentiator between this program and those of other institutions. John Carroll (1963) recognized that if educational time is held constant, student learning will be the variable. He suggested altering the paradigm so that learning becomes the constant, making time the variable. In this program the students would be in control, deciding what they wanted to learn, when to learn it, and the pace at which they would progress. Importantly, the structure of the program would give them the time they needed to succeed.

This model of learning has the advantage of ensuring the student covers the relevant topics at the point in time which they are most relevant. However, it also introduces risk in terms of how the students manage their progression through the learning. Students have considerable amount of freedom in deciding how and when they engage with the topics in the online environment which could impact positively or negatively on learning.

The Topic Tree

The delivery of the online asynchronous element of the course is through the Realizeit adaptive learning platform (Howlin & Lynch, 2014). Within this system, the Topic Tree was created. This is a structured curriculum which breaks concepts into granular 'topics' that each require approximately three hours to complete. These are then arranged using the fundamental prerequisites relationships that exist between them, and a tree-type structure emerges which makes the recommended learning order visible and explicit. The Topic Tree is the mechanism which enables students to measure, plan, and organize their learning.

The Topic Tree moves the students to a Mastery-based learning paradigm. Each topic is assessed on a ‘mastered’ or ‘not yet mastered’ basis. Students can progress when they have acquired the knowledge to a required standard. If this occurs quickly, they can advance rapidly; if it takes longer, then the student can take the time needed, rather than missing out. Where possible (and appropriate), automated assessment and feedback are used to support the student learning.

The research presented in this session shares initial findings and insights on how students in a self-directed learning environment are making use of the topic tree and its impact on their learning. We examine how students plan their learning, how they move between topics available to them, what kind of time gaps are they leaving, and if there is a relationship between these behaviors and student performance.

Student’s Planning

The platform allows students to select subsections or branches of the topic tree to add to a plan. They specify the topics they want to learn; then the system selects the required topics from the topic tree they need to master. A plan essentially describes the gap between their current knowledge of topics and their goal topics as chosen by themselves. Students have the flexibility to create large plans (macro-planning) by picking advanced topics to the right of the tree, or create multiple small plans (micro-planning) and incrementally build their way to the more advanced topics.

We examined how students build plans and use them to manage their learning by looking at the number of plans each student created, along with the median size of their plans. We found that there is broad variability in how students use plans, from the students who create many relatively small plans to the students who form a single large plan. There are no clusters; students fall along the range from micro- to macro-planners. It is also worth noting that some students engage in both micro-planning for day-to-day activities and macro-planning to track long-term goals.

The Netflix Effect

Introduction of a self-directed learning environment in conjunction with the topic tree enables new learning behaviors that were previously not possible. In another collaborative study, Dziuban, Howlin, Johnson, and Moskal (2017) examined the pace of learning for students moving through similarly structured curricula within the bounds of traditional 15-week courses at the University of Central Florida, and 5-week full online courses at the Colorado Technical University. They found that students displayed a range of behaviors. Some students rapidly complete the course, some progress at a steady pace, and some leave everything to the last minute. However, these courses were not self-directed.

Given the open nature of the learning, we were interested in examining how students moved about in the topics tree. We concentrated on two aspects:

• the distance measured as the number of prerequisite links between topics in the topic tree for consecutive learning activities, and

• the time between consecutive learning activities.

We examined each pair of consecutive learning activities for each student and measured the time between learning activities, as well as the distance between consecutive attempted topics. We found that students tend to stay in a set neighborhood of the topic tree and complete all related topics before moving on. Examination of the time metric revealed that students typically attempt multiple activities on any day that they are active. Further investigations found that this behavior is generally replicated across several days, proceeded and followed by days with no activity. Students are engaging with the learning in a manner which is similar to how many people watch series on Netflix: they tend to stay on the same series and watch all episodes in a short period of time.

We discovered several other striking behaviors as part of this analysis. For example, some students engaged in bingeing to catch up when they have fallen behind. This behavior resulted in a short-term closing of the gap to the other student, but it ultimately proves not to be a very successful strategy as the gap opens again after a few weeks. Another example is the identification of students who never immediately repeat a topic; they always seek out new concepts to learn.

Conclusion

This program puts students in control of their learning, with the tools they need to manage and organize it, allows them to learn the topics when they are most relevant to them, and gives them the time and flexibility they need to do all this. It is a substantial departure from traditional courses. The results presented here focus on just one aspect of this model: how students are behaving. We need to be able to provide feedback to students, faculty and the institution on the success of the model - what is effective, and what is not. This session shares the results from our first step in that direction. While some of the uncovered behaviors are striking, how it affects outcomes and performance remains a little murky and will form a key piece of some of our follow up investigations.

The research presented in this session is the result of a collaboration between the faculty engaged in the development and implementation of the engineering program at CSU, researchers working at Realizeit, and a senior researcher working at the Technology Centre for Engineering Education in Delft University of Technology, Netherlands.

References

1. Carroll, J. B. (1963). A model of school learning. The Teachers College Record, 64, 723–733. Retrieved from http://www.tcrecord.org/Content.asp?ContentId=2839
2. Dziuban, C., Howlin, C., Johnson, C., & Moskal, P. (2017, December 18). An Adaptive Learning Partnership. EDUCAUSE Review.
3. Howlin, C. P., & Lynch, D. (2014). A framework for the delivery of personalized adaptive content. 2014 International Conference on Web and Open Access to Learning (ICWOAL), (pp. 1- 5). doi:10.1109/ICWOAL.2014.7009203
4. Morgan, J.R., Lindsay, E. D. (2015), The CSU Engineering Model, Proc of the 2015 Australasian Association for Engineering Education Conference, Geelong, Australia