Desktop version

Home arrow Philosophy

  • Increase font
  • Decrease font

<<   CONTENTS   >>

Discussion and challenges of big data and the future of strategy research

Although commercial industries do not specifically examine the strategic processing of learners, certain non-educational sectors have been effectively leveraging Big Data in recent years in ways that help them maximize services to their customers. To achieve this, multiple sources of data that reflect patterns of both users’ behaviors and ambient environmental conditions are used to create detailed profiles for users. Industries that have taken advantage of Big Data have used such comprehensive models of users’ unique digital fingerprints to nudge users toward actions that are best predicted to help users achieve both corporate and users’ personal interests. For educational researchers, such real-world examples provide design inspiration for those who are seeking to develop EDM/LA, educational interventions that can both readily identify the types of strategies and other behaviors that learners are using to meet their goals, and dynamically adapt and personalize learning environments based on strategy and usage patterns.

There is no question that EDM/LA presents huge potential for research on student and instructor strategic processing and knowledge development in the pursuit of developing robust and adaptive online learning environments. At this point, it should be clear that there is great potential in the harnessing of Big Data for the purposes of facilitating evidenced-based and automated instructional personalization. What should also be clear is that right now, much of what is occurring in the education research sector through EMD/LA is at the “buzz word level.” What we mean by this is, much of what is being explored in the EDM/LA literature has either conflated that notion that a data set on a large number of participants is equivalent to the broader context of “Big Data” research, or has simply appropriated research on single-source data forms such as click streams or log files and renamed it EDM/LA. Getting serious about Big Data in education means we need to move past the term de jour. The field must understand that Big Data involves not only volumes of data but also the real-time capture and analysis of data from a variety of sources that can then be operationalized to tailor instruction based on a user’s interactions, strategies, and learning trajectories. This argument is not meant to slight the extant research in its current nascent stage; rather, it is intended to push research forward to advance our knowledge and current pedagogies in ways that reshape education in the way industry has reformed the online market experience for consumers.

As the field moves forward, however, there are a number of issues that need to be resolved to fully embrace the power of Big Data in education. The prospects of collecting and using information about students also raises a number of ethical and logistical questions, including selection of data streams, system interoperability, issues pertaining to student privacy, and the need to create interdisciplinary research teams (Rubel & Jones, 2016). In the following section, we discuss each of these challenges that the field will need to address as the use of EDM/LA becomes more prevalent and more individuals will be impacted by the data collection protocols and resultant environmental adaptations that will be more increasingly conducted by automated, machine-learning algorithms.

Data Selection and Interoperability

Because it is still in its infancy, most of the research reviewed in this chapter has a limited view of the utilization of EDM/LA and Big Data techniques. Within the ITS literature, for example, while there is the use of learner data modeling, and, in some cases, machine learning to iterate these models, research has inadequately explored the integration of multiple streams of data outside of user interactions (Beck & Mostow, 2008). Similarly, while a great deal of rhetoric has expounded on the potential of online learning and MOOCS to use data as a tool to push forward the future of tailored, personalized learning, in reality the majority of this work has been constrained to user log data derived within the LMS platform used to deliver the courses (Mangaroska & Giannakos, 2017). EDM/LA approaches are only as good as the corpus of data available to extract, aggregate, and analyze. If the data pool is restricted to a single stream or service, this is not as insightful as having access to a wider data pool that is pulled from a variety of sources. Such multi-sourced data streams have the unique potential to reveal a more realistic range of learner strategy use and other behaviors that account for both learners’ prior interactions and various environmental or structural factors that can influence learner activity. With so many forms of data available, the questions of which data to collect and which sources best provide this data are critical to answer (Slade & Prinsloo, 2013). Educational research must then begin to integrate and aggregate data across a multitude of sources (Di Mitri, Schneider, Specht, & Drachsler, 2018). By examining a more robust corpus of data streams, researchers interested in understanding the strategies that learners employ can expand beyond navigation trace data and include information on why certain users select these strategies, what determines prompt changes in strategy, and how learning environments can dynamically respond to particular strategies to shape more efficient learning practices. For example, at the post-secondary level, such data sources could include admissions information, geo-tagging of student online habits, capture of online activities in which the student is engaged concurrently with their access to online course materials, bio data provided by wearable technologies, and the list goes on. At the elementary and high school levels, collecting such robust data sets on students could facilitate the transfer of knowledge about a student from teacher to teacher or across grades, creating student educational histories that mirror processes used with medical records.

A big challenge for this comprehensive vision of multi-source Big Data integration in educational research is system interoperability. While the ability to capture data in each of these independent streams is currently available (and widely used in industry), communicating these data across the diverse set of systems is a bit more problematic. Essentially, each system has its own data collection and storage protocols. To port data out of individual systems and to aggregate them into a single repository is typically no small feat. The enterprise of education is going to need to put concerted effort into addressing the issue of interoperability if EDM/LA is to realize its value to research. Despite this challenge, the potential benefits of solving the issue of interoperability far outweigh the required investment of time and resources.

<<   CONTENTS   >>

Related topics