Instructional Design, Learning Analytics, and Educational Data Mining (EDM): Towards Personalized Learning
The goal of Instructional Design is that everyone can learn provided you teach them the way they can learn. Although this seems like a dream, soon this will become a reality. Learning Analytics will make it possible through Personalized Learning and with the help of Educational Data Mining.
We will look at how each of these areas - instructional design, learning analytics, educational data mining and personalized learning, work together to help us learn better and to use our mind and body to our maximum potential.
Have you ever wondered how the process of learning happens? Learning as many of us know is a natural and lifelong process. It is also a joyful process. But unfortunately somewhere in the education system, something went wrong to make learning boring.
As we all know, human beings are the most sophisticated gadget on earth. The kind of software (brain), within each gadget determines its capability to perform. Again, whether the gadget uses its capability to its fullest potential is based on various factors. These factors are determined from the time the child is born and even before that.
Now, imagine a child born into this world. By default, the child’s brain is programmed with minimal data/information for basic survival (drink milk, cry when hungry or in pain, etc.). But as the child grows, it comes in contact with its parents, siblings, relatives, friends, society, etc., and sooner or later it also receives information through books, journals, television, computers, mobiles, and various other means. The information and experience received from all such sources play an important role in developing the human software (brain). The thoughts, beliefs, attitudes, and behavior are all molded with the help of this software, which acts as a storehouse of all kinds of useful and not so useful information.
Now that we can look at ourselves as a great technology, can we not engineer it the way we want it to be? Yes, it is possible through mentored learning or personalized learning.
What is personalized learning? It is defined as the tailoring of pedagogy, curriculum and learning environments by learners or for learners in order to meet their different learning needs and aspirations. Typically technology is used to facilitate personalized learning environments.
The personalized learning model is highly flexible and adaptable to each individual’s needs. It encourages the collaborative involvement of all stakeholders to create the best learning path. However, this is possible only with the help of emerging technologies such as learning analytics and educational data mining that act like multiple human brains. But without the help of good instructional design and continuous human (stakeholder) intervention, personalized learning will fail.
We need to remember that personalized learning would keep evolving constantly to build an intelligent curriculum. Due to its unique nature, it cannot be based on ‘one size fits all’ model. Consequently, some of the latest learning trends such as blended learning and next Gen learning may all fall into this category.
Personalized learning has been defined in different ways by different people since the 19th century, when Helen Parkhurst created the Dalton Plan stating that each learner can program his or her curriculum in order to meet his or her needs, interests and abilities; to promote both independence and dependability; to enhance the learner’s social skills and sense of responsibility toward others.
Although the term personalization, in the context of education was first coined by Victor Garcia Hoz in the 1970s, it has been practiced by many successful people knowingly or unknowingly for ages. Researchers have found that personalized learning helps learners use their ability to the maximum potential. Even the wisdom traditions say that when humans use their body and mind to their maximum potential, they feel truly happy and successful.
Chris Watkins a reader at the Institute of Education, University of London, demonstrates in his research “learners in the driving seat” that when learners drive the learning it leads to:
- Greater engagement and intrinsic motivation
- Learners setting higher challenges
- Learners evaluating their work
- Better problem-solving
Sir Kenneth Robinson, an educationalist, in his TED talk, talks about the need to move from the industrial model of linearity and conformity to the agricultural model, where the right conditions are set for the learners. This, he says in support of personalized learning.
Many researches prove that personalized learning is the way to empower learners. Personalized learning is not just about individual success. It has the power to bring about a social change. It can raise the level of people who are at the bottom of the socioeconomic ladder, reduce learner dropout rates, include more girls and people of color into STEM (science, technology, engineering, and mathematics) fields, and also address the lack of suitable candidates for high-tech jobs. However, we need to always keep in mind that all this is not possible without the right spirit, right intention and right intervention from the part of educational stakeholders.
The first step that needs to be taken to implement the personalized learning model is to create perfect learner profiles. For this, regular and constant monitoring of learner data in educational contexts is essential. The measurement, collection, analysis and reporting of such data for purposes of understanding and optimizing learning and the environments in which it occurs is called learning analytics. This was defined at the first international conference on learning analytics and knowledge (LAK11), 2011.
Data collected through learning analytics in a computer based education system such as a learning management system (LMS) or intelligent tutoring system (ITS), can be interpreted by teachers, instructional designers and other educational stakeholders to understand - learner engagement, progression, and achievement. Furthermore, this learner-centric data will help formulate future pedagogical decisions on how to modify the learning environment to improve student learning. This must be an iterative process until the best learner outcome is reached.
Imagine the purpose of gauges such as a speedometer, fuel gauge, and odometer, and indicators such as gearshift position, and seat belt warning light, on a car dashboard. Learning analytics works like these gauges and indicators, collecting data from multiple sources and presenting the information clearly to the driver (learner), so that he or she can drive (learn) without any problem and reach his or her destination (career goal or learning goal) successfully.
Learning analytics cannot be considered as a single discipline. It is an amalgamation of many academic disciplines like pedagogy, educational psychology, computer science, statistics and information technology. It is also related to fields like academic analytics, web analytics, action analytics, and educational data mining. There are many challenges ahead in this area of research. Because, in the current educational scenario most learning management systems or virtual learning environments lack an efficient student tracking system that can provide easy to understand learner activity reports. The raw data available through data mining can be interpreted only by technically savvy educators. And that involves a lot of time. Therefore, development of new processes and tools for improved learner profiling is necessary.
Many new technologies like eye-tracking, content analysis, sentiment analysis, video analysis, and interaction analysis are on the way. These technologies will aid in understanding each learner in depth, including their learning style, their interactions, their networking, and other learning-related activities. Based on these analysis, instructional designers, teachers, and academicians can build courses and lessons most suitable for learners. This will bring about a revolutionary change in the field of education. However, we need to take care that learner privacy is maintained and confidential user information is protected from abuse.
Learning analytics can be broadly classified into four types - descriptive, diagnostic, predictive, and prescriptive. Descriptive analytics focuses on what is happening and what happened. Diagnostic analytics focuses on why something happened. Predictive analytics focuses on what will happen. And prescriptive analytics focuses on what should be done to achieve a specific outcome. Each of these types can be used based on the requirement in a particular educational context.
Learning analytics cannot survive without its related field - educational data mining (EDM). EDM is the backbone of learning analytics. The technology and methodology by which they work are different, but their goals and interests are the same. They both work toward understanding the learner and improving their performance. The International Educational Data Mining Society defines EDM as “an emerging discipline, concerned with developing methods for exploring the unique types of data that come from educational settings, and using those methods to better understand learners, and the settings which they learn in.”
Imagine you are working on a course in a computer based learning system. Every interaction of yours from a mouse click to text chat, from your administrative to demographic data, and even your emotional states will all be recorded as large amounts of unstructured or raw data. This raw data is not easy to understand. It requires special skills to be interpreted. EDM has its origins in statistics, machine learning, visualization and computational modeling. It focuses on automated discovery patterns and models from extensive datasets. It creates applications and methods for easy understanding of such raw data and learning analytics makes use of these applications and methods to understand learner activity in a particular learning context. It is working on visualization and modeling techniques that maximize comprehension for educators to look and make sense out of the data.
There are several data mining methods in the educational environment. Some of the popular methods are - prediction, clustering, outlier detecting, SNA, and text mining. Prediction is used to predict learner performance and behavior. Clustering is used to group similar course materials and learning patterns. Outlier detection is used to identify learner difficulties and any irregular learning model. Social network analysis understands relationships like friendship among learners and text mining is used to analyze the content in discussion boards and chat rooms. However, the purpose of all these methods and techniques are to create personalized learning environments for the learners.
Personalized learning is an instructional design strategy. Instructional design is defined as the systematic design, development, and delivery of instruction. It generally uses the ADDIE (Analysis, Design, Develop, Implement and Evaluate) model. A good instructional design model works like a cars GPS system, driving the learner towards his or her destination (learning goal) and alerting them when they go off- track. The learning systems of the future will contain data dashboards and personalized learning maps that will show the learners how long they have travelled and how far they need to go to reach their destination. And what they need to do. The learning path need not be linear. They can take multiple pathways based on their interest. Instructional design should continuously monitor their progress and fill/address the gaps, if any, as the learners move along their learning path. Good instructional design also takes care that learners don’t get lost on their way to progress, through regular feedback and intervention. A competency-based instructional design makes sure that the learner demonstrates mastery over a lesson or subject area through various ways such as showcasing of projects, writings, demonstrations, e-portfolios, etc.. Learning analytics and educational data mining play a very vital role in every phase of the instructional design process. Because the learner reports they provide help in learner profiling, which further supports - in finding out the gaps, to address the gaps, to keep learners on track, to bring them back on track, if lost, in finding how quickly or slowly the learner can tread the path, and much more.
All of this process, helps us understand, how brilliantly they (instructional design, personalized learning, learning analytics, and educational data mining) coordinate and collaborate and work hand in hand to help us learn.
