Learning Analytics For Course Adaptivity4 min read

Deep Learning vs Surface Learning

In her blog on FacultyFocus.com, Maryellen Weimer argues that the difference between deep learning and surface learning is not always sufficiently clear for either teachers or students. Deep learning is often defined through what it isn’t. For example, deep learning is not just memorizing text or reciting passages.

The essence of deep learning, however, is true knowing. According to Weimer, deep learning begins when a concept will stay in the long term memory, and other new concepts will be linked to this concept in the brain. In other words, deep learning means students understand a concept and are able to work with it.

Surface learning on the other hand is very temporary; it helps a student pass an exam, but that knowledge won’t be readily available to them in the second or third year. Education is supposed to be a collaboration, or interplay, of all the courses, so it becomes problematic when new knowledge does not build upon previously gathered knowledge. Through deep learning, students will be able to continue building on a foundation, using previous knowledge to link new knowledge with.

Learning Analytics

So deep learning is important. Now what? Well, consider this: when a student takes a test and gets a good grade, does that mean the student used deep learning? Unfortunately, this is hard to determine. Instead, the best place to find out what type of learning the student engages in, is during the learning process.

A recent article by Tempelaar et al. researched how data gathered through a digital learning application can be used to influence the learning process. They implemented external e-learning tutorials and used the feedback they gathered as trace data. In previous research they proved that an online learning environment is an important asset in improving the learning process and learning experience. The data available through the e-learning platform can be used to help detect at-risk students, and to help adapt the course to meet the individual needs of each student. In this follow-up research, they aimed to prove that this data can be used to identify whether a student engages in deep learning or surface learning.

What did this research team look at? They used our mathematics learning platform, SOWISO, to analyze, amongst others, the following trace data; mastery, attempts, solutions and hints.

Mastery shows the proportion of exercises which are correctly solved. The other data shows the number of attempts the student has made to solve exercises, the number of hints called upon and the number of solutions checked. Looking at all these elements can help show not just how a student is doing, but also gives insight in the learning process behind the results. With this data, teachers can intervene exactly when and where they need to, with precise personal tips to help the student get back on track. This is especially useful in higher education, where classes are big and often impersonal.

Tempelaar et al. argue that this data can also be used to recognize which students have a disposition for surface learning rather than deep learning. Students who show high activity levels will not be flagged as at-risk, but might only engage in surface learning. They argue that the comprehensive Learning Analytics should also be used to identify students who engage in surface learning, and should help to design “learning interventions directed at changing surface learning approaches into more deep learning approaches has more potential than just telling students they are using more worked out examples than the best students in their class are doing.”

reporting overview.png


SOWISO stands out from competitors in the Learning Analytics that are available. Currently, it’s possible to see:

  • Mastery
  • Theory pages accessed
  • Exercises finished
  • Number of mistakes made
  • How many solutions and hints were called upon
  • Activity per date

Screenshot Report 2.jpg

To support a more in-depth understanding of which students are engaged in surface learning vs. deep learning, we are currently working on making even more comprehensive statistics available:

  • Time spent on theory pages
  • Time spent on exercises
  • Number of attempts per exercise
  • Number of attempts in total
  • Number of hints used while solving exercises
  • Number of solutions used while solving exercises
  • Number of theory pages used while solving exercises
  • Solutions called for before attempt

SOWISO’s software already personalizes the exercises for students. The constantly changing variables enable unlimited learning for students, while the platform automatically analyzes the student’s’ progress and mastery level, adapting the level of the exercise to optimize their individual learning path.

reporting zoomed in on exercise (package) level.png

However, teachers can also access this data to alter the course or subsections of the course in order to maximize the effectivity of the program. SOWISO enables educators to personalize and adapt the course almost effortlessly, without compromising the overall level or content, creating the comfort and personal feel of a small classroom for every conceivable group size with students from all over the world.

Intrigued? Feel free to check out our free demo to see for yourself!

I would love to hear any questions or comments you have,


One Comment

  1. Pingback: Visualization in Math Education - SOWISO Blog

Comments are closed.