It’s almost impossible to escape the notion that data is changing every aspect of our lives. ‘Big’ data is often conflated with ideas like artificial intelligence and machine learning, and also ethical issues like privacy and security. And, of course, these ideas have made their way into education, and that means that data, learning analytics and related ideas are going to have an effect on schools, students and teachers. But what might that effect look like? And what does it mean for the teaching profession? How might teachers be prepared to safely make use of tools provided by learning analytics?
We are already quantified in many aspects of our lives. Often, we choose to do this - tracking the number of steps we take a day, or how long we are reading, or the time we spend on screens. Some people have always done this, in both digital and analog forms. However, we are now entering the era of ‘Big’ data, which is both qualitatively and quantitatively different. Big data is characterised in four main ways. Firstly, the volume of the data - it’s at a scale that is unprecedented. Secondly, the different types of data (or its Variety. Thirdly, the speed at which this data is generated (Velocity) is a key feature. Finally, the veracity or trustworthiness of the data is an important consideration.
Teachers have always used data to inform their teaching. So why should ‘big data’ be any different? The answer might lie less in what data is gathered, but in what is done with it, and it is at this point that machine learning and artificial intelligence (AI) enters the conversation. Informed by big data, it is claimed that AI might have the capacity to either semi or fully automate the teaching and learning design process, and in doing so, do teaching better than teachers themselves. Some recent comments, by the UK Government, suggested that AI might aid social mobility by ensuring quality teaching for all students, regardless of socio-economic status. Such claims about computer assisted instruction are hardly new, and currently, despite some news stories, these promises are yet to be realised.
While fully automated interventions might be some way off, learning analytics might be a useful tool for teachers seeking to enhance their practice. Simply put, learning analytics is the combination of educational theory, learning design and data science. It seeks to turn the wealth of data available into actionable insights that either assist teachers in making decisions, or free them to do the things that are best done by teachers (by doing things that are best done by computers).
This can happen at a number of different levels. Simon Buckingham-Shum (2012) suggested that learning analytics can operate at a macro level, such as the region or state, a mess or institutional level, and also a micro level (related to cohorts or the individual user). For teachers, learning analytics has a role to play at the meso and micro levels. Two possible areas are in automated and semi-automated feedback and adaptive or personalised learning.
In automated feedback, machine learning algorithms provide feedback to students based on the data they have gathered and how well students performance conforms to a model of student success. This is a complicated business, and might include measures related to context, affective, cognitive and behavioural engagement and predictions about expected learning outcomes. Currently, tools like AcaWriter (trialled at UTS) can give automated feedback to students based on their writing, but more complicated uses are yet to be developed. More common is the notion of semi-automated feedback, where the analysis might bring to light specific areas needing attention, at which point the teacher can intervene. For example, a quick quiz might identify that the majority of the class did not understand a specific concept; the teacher might decide to tackle that topic in a different way. Another example is identifying students at risk of failing a course or subject through lack of engagement, allowing teachers to intervene early.
Another approach is adaptive learning. Currently, one of the criticisms of many technology-based approaches to teaching is that they are linear; that is, there is insufficient differentiation for students with different needs, wants or abilities. This is especially the case in online courses, which are often offered in a step-by-step fashion. Such a one-size-fits-all approach can lead to frustration, difficulties in learning, and a high dropout rate. Alternatively, adaptive learning systems, which gathering data about the student at every stage, can change the learning environment and/or learning activities automatically to adjust to the learners’ individual situation, characteristics and needs, and therefore provide personalized learning experiences. This can take the form of offering extension work, or providing extra support, or any of a range of possible interventions.
While automated feedback and adaptive learning might not be fundamentally altering the teaching profession right now, there is a growing clamour for education to start making better use of the opportunities afforded by AI. This means that teachers need to start considering how they might upskill themselves in order to use these technologies when they become available. A good starting point exploring learning analytics, both within a teacher’s current context, and beyond.
Buckingham-Shum, S. (2012). Learning Analytics Policy Brief. UNESCO Institute for Information Technologies in Education
Dr Keith Heggart is an early career researcher with a focus on learning and instructional design, educational technology and civics and citizenship education. He is currently exploring the way that online learning platforms can assist in the formation of active citizenship amongst Australian youth. Keith is a former high school teacher, having worked as a school leader in Australia and overseas, in government and non-government sectors. In addition, he has worked as an Organiser for the Independent Education Union of Australia, and as an independent Learning Designer for a range of organisations.