I think the learning analytic research should move from the current practice of doing quantitative data analyses to include in it qualitative analyses. The quantified self should be expanded to be qualified self.
In learning analytics research we should consider use of mixed methods that are combining quantitative and qualitative data analyses.
Today the learning analytic research builds strongly on the quantified self idea. The idea of quantified self is simple and powerful. With help of technology we can collect data on our daily life, such as our physical activity (mobility, walking, running etc.), surrounding environment (weather, air quality etc.), our performance (work, study etc.) and social relations (emails, phone calls etc.). The reason to gather and analyze data is to increase awareness on ones own life and ultimately, I assume, to have a chance to change things in it.
The idea of quantified self raises some questions. Like, how much data on their behavior and analyses people really need to get to the right conclusion? For instance, people who have never tracked or record their jogging can still tell pretty accurate information on it (for instance: I run 0, 1 or 2 times / week / 3-5 kilometers). Whatever they run a lot, little or not at all they must be aware about the fact. People also can tell relatively good description of their diet. Most of us do not have a clue about the amount of calories we eat, but most of us know whatever our diet is healthy or not. Because of knowing all this (without any numbers) people may also pay attention on their diet and may have an attempt to run more (or less). On the other hand many people rarely enjoy running and often enjoy unhealthy food. In some aspect jogging and eating healthy food are decreasing the quality of their life.
A different thing is when someone is training, for instance, to run a marathon. In it exact data and a plan helping to reach the objective is for sure useful. Most of us, however, are not interested in this kind of training. Doing some training is still important.
The idea behind the learning analytics is that collection and analysis of data about learners and their context will provide opportunities to optimize learning and the learning environment (compare to training to run a marathon). In practical implementation of the learning analytics, learners and teachers are provided visualizations on their interactions and progress in some study course. The visualizations can be things like performance in assignments and tests compared to other students or social network analyses.
At some level this probably makes sense, but I think often in study work one can reach good conclusion simply by observing, self-reflecting and using common sense. I think most students know, from various small hints, how they are doing in a class. It is a bit like knowing that I do not run enough or knowing that I should eat healthier food — just by knowing it without any accurate data. In this case people are doing qualitative analyses that is not based on the limited accurate data from the course but from various sources of fuzzy information.
Getting back to the issue of running and diet, however, we must remember that without tens of years of scientific research on the topics — health, physical exercise and diet — people wouldn’t be able to come up with the “right” conclusions of these things importance in their own well-being. I assume this is the case with learning and learning research, too. We should study how people learn, because that will help individuals to monitor, reflect and self-regulate their own behavior. Even if numbers and visualizations on individuals’ behavior may help students to be aware of some things related to their learning, I think we should get beyond it — to the quality of learning.
For many years in social science there has been two methodological camps — you may call them paradigms —fighting on their relevance. These are quantitative research and qualitative research. Recently there has been some advance of bringing them together. The mixed methods have become popular. Often quantitative research can provide interesting research questions for qualitative research and other way around. To get a good picture on some complex social phenomena (e.g. learning) one must use both.
The mixed method (also called multimethodology) approach could be used in learning analytics research, too. What then would be qualitative learning analytics? Could this approach lead to qualified self?
With some latest prototypes we have somehow touch the topic. We call the new learning tools reflection tools. Here is a video of the three latest prototypes.
The idea with the tools is not to collect quantitative data (there is some, like how many reflections one have made), but to provide a space for student to do reflection in natural language. With the tools students are asked to think and ponder questions, like: what I have learned? What I have done? What I am planning to do next? Have I faced any problems to implement my plans?
The reflection tools are also calm technology. They are designed not to be distributive in aa learning situationtaking place in social interaction. They are not central, they are peripheral, but can be brought to the center when needed.
What I would like to see in future in the learning analytic research is a move to the direction of machine learning and natural language analyzes. I am imaging that one day we could automatically or semi-automatically analyze content people create as part of their learning activities (or everyday life) and based on that provide them hints on directions they could explore more. The picture build out of the qualitative data (the content produced) could be something that could be called “Qualified Self”.
As a final (meta) note I want to explain how this idea of qualified self and qualitative learning analytics idea came to life. Why? Because it is a nice story and demonstrates how research happens.
A couple of weeks ago I met with Erik Duval when he was giving a keynote in a conference in Finland. Erik is doing right now a lot of research on learning analytics. His talk and discussions we had were very inspiring. At some point we also discussed about quantitative and qualitative analyses – actually in the context of research evaluation.
Next week I was in Copenhagen and was lucky to have dinner with Timo Honkela – a colleague who happens to be visiting fellow in Copenhagen right now. Timo’s area of research is computational cognitive systems — “adaptive, autonomous and socio-culturally grounded cognitive systems that are able to learn and use language“. Some years ago with Timo we did some theoretical research around the idea of using self-organizing maps (SOM) in learning. During the dinner I explained to Timo the idea of qualified self. He liked it and brought in it the idea of machine learning. I hope in a near future we will do some writing on it.
In Copenhagen I also met Jonas Löwgren, one of the leading figures in interaction design. He made some more interesting comments on the idea of qualified self.
Thank you all!