Send the link below via email or IMCopy
Present to your audienceStart remote presentation
- Invited audience members will follow you as you navigate and present
- People invited to a presentation do not need a Prezi account
- This link expires 10 minutes after you close the presentation
- A maximum of 30 users can follow your presentation
- Learn more about this feature in our knowledge base article
Do you really want to delete this prezi?
Neither you, nor the coeditors you shared it with will be able to recover it again.
Make your likes visible on Facebook?
You can change this under Settings & Account at any time.
The Promises & Pitfalls of Big Data in Higher Education
Transcript of The Promises & Pitfalls of Big Data in Higher Education
for the course Education & Digital Culture at the University of Edinburgh
'A new era of Big Data is emerging, and the implications for business, government, democracy and culture are enormous.'
(Bollier 2010: 1)
What is Big Data?
Big Data is a widely used but frequently ill-defined concept (boyd & Crawford 2012) and, owing to its shared history between academia, industry and the media, no universally-accepted definition of the concept exists (Ward & Barker 2013). Gartner (2001) however, identifies 3 V's associated with the concept of Big Data:
. In other words, rapid advances in the development of ubiquitous computing, coupled with advances in sophistication (and reduction in cost) of computing technology, has led to vast, fast and diverse quantities of digital information being generated (online) on a daily basis; data which due to their sheer size, are beyond the management of conventional (i.e. human-controlled) storage and analytical systems. Yet, Big Data is more than a tangible set of data (however big it may be). Big Data is also a
that reflects both an approach to knowledge as well as a technique for achieving socio-political objectives. As such, I propose a tripartite approach (PEI) to Big Data,which is discussed below.
Promises & Pitfalls
The Promises & Pitfalls of 'Big Data' in Higher Education
Big Data in Higher Education
Big Data as
Big Data as Practice refers to the existence of
qualitatively new forms of data
, that require new (specialised) analytical techniques. According to Daniel (2014), for example, the existence of vast amounts of online data provides new opportunities for Knowledge Discovery (KDD) methodologies, capable of mapping and identifying systematically, patterns in human behaviour. This can be achieved through the development of algorithms that use both mathematical formula (e.g. cluster, factor and regression analysis) and visualisation techniques, to reveal patterns in online activity.
Big Data as
A final approach to understanding Big Data is to see it as an
, linked to the achievement of socio-political goals. As boyd and Crawford (2012) argue, Big Data does not merely relate to sets of digital information and corresponding analytical procedures, but to deep seated value and belief systems located within societies. As such, the concept of Big Data can trigger both utopian, as well as dystopian, visions of the future.
Big Data as
The second approach to understanding Big Data is to view it as a series of claims surrounding the nature and purpose of knowledge (i.e. an epistemology). Authors such as Kitchin (2014), for example, have argued that proponents of Big Data have heralded its development as signalling the beginning of a
(Kuhn 1962) in scientific enquiry. Reasons for this include the belief that Big Data can offer insights into the 'true' nature of human behaviour; insights unhindered by the (deleterious) role of human bias and interpretation. Thus, proponents argue, Big Data will lead to a 21st century scientific revolution that will render the practice of theorising largely obsolete (Kitchin 2014).
The Signals Project,
Perdue University: A traffic light system designed to indicate to students when they are doing well (green), if there is cause for concern (amber) or if the student is at high risk (red). See: http://www.itap.purdue.edu/studio/signals/
Examples of Learning Analytics in Higher Education
, Open University: Currently under development, this software seeks to provide automated feedback to students when preparing summative assignments (see Labeck et al. 2013).
Learning Analytics as Practice
A multi-epistemological framework for interpreting Big Data (Erwin, Bond & Jane 2015: 58)
Summary & Reflections
'What different sorting algorithms sound like'by Andrut
This essay explores the promises and pitfalls associated with the rising influence of
in higher education, both in the UK and internationally. It begins with an exploration of what is meant by Big Data and the ways in which Big Data is influencing activities and priorities within higher educational institutions. In so doing, the essay introduces a new (social science) definition of Big Data, as:
). The PIE approach is then applied a specific form of Big Data in higher education, learning analytics. Based on the PIE approach, the essay concludes with a warning against techno-deterministic constructions of Big Data, as a pathway to a utopian (or dystopian) educational future.
Recent years have witnessed rapid growth in the application of Big Data (as practice, epistemology and ideology) in higher education, both in the UK and internationally. One of the most important manifestations of Big Data in this respect has been the increasing influence of
'Learning analytics is an emerging field in which sophisticated analytic tools are used to improve learning and education. It draws from, and is closely tied to, a series of other fields of study including business intelligence, web analytics, academic analytics, educational data mining, and action analytics.'
(Elias 2011: 2)
According to Bichsel (2010) learning analytics seek to combine practices of data mining and visualisation with insights developed through pedagogical models. As Elias (2011) highlights, the increasing use of virtual learning environments (VLEs) in both distance and campus-based degree programmes has led to increasing amounts of data on students' learning behaviours being generated. This in turn, is leading to new opportunities for data mining and algorithmic-based analysis. Here are some examples of learning analytics that are being developed and implemented in universities across the world:
The remainder of this essay explores the implications of the growing use of learning analytics in higher education, based on the
approach to Big Data.
This section considers the promises and pitfalls, as highlighted by those seeking to develop learning analytics in higher education.
According to Bischel (2012) , one of the key promises of Big Data in general (and the use of learning analytics in particular) is the ability to better understand students' learning behaviour. This, it is argued, provides important opportunities for educational interventions on two levels:
Predicting educational outcomes:
By using data generated from students’ engagement with educational content, learning analytics can be deployed to identify students who are ‘at risk’ of achieving sub-optimal outcomes. Early identification thus provides opportunities for early intervention; for example, through offering additional support to at risk students, and/or addressing key barriers to learning before they impact on students' assessed work and final degree classifications.
Creating individualised learning experiences:
According to Dietz-Uhler and Hurn (2013), learning analytics can be used to challenge widely-held assumptions held by educators, that all students begin courses at the same stage and that they will go on to learn at roughly the same pace. By generating extensive data on the activities of each individual learner, learning analytics will help reveal the different paces and processes through which students acheive. As Dietz-Uhler and Hurn (2013:18) argue:
'Just as Amazon.com uses the data from our purchase history to make suggestions about future purchases, so can learning analytics allow us to suggest new learning opportunities or different courses of action to our students.'
Whilst universities collect substantial amounts of data pertaining to students' learning behaviours, these data often go un-analysed (Bischel 2012). Following Dietz-Uhler and Hurn (2013), gathering extensive amounts of data that could be used to enhance educational outcomes, and not analysing it, presents
for universities which, in a growing climate of consumer rights and student redress, could provide damaging; if, for example, students withhold payment of fees as a result of what they perceive to be sub-standard education (as has happened recently at the University of Worcester). In addition, objections have been raised that excessive tracking of students’ learning activities may
encroach upon their civil rights
, especially their right to privacy (see for example, Scalter 2014). Managing these ethical complexities in an era of learning analytics is likely to prove highly challenging for higher education institutions.
This section explores the promises and pitfalls of learning analytics in higher education, from an epistemological perspective.
Learning Analytics as Epistemology
Drawing on writers such as Prensky (2009) and Anderson (2008), Kitchin (2014) links the development of Big Data to the re-emergence of empiricism in social science. In line with the effective removal of theory (a key tenet of realism), the drive towards empiricism is the assertion that, simply put, 'correlation is enough' (Anderson 2008 quoted in Kitchin 2014: 3). In other words, that algorithms are capable of identifying patterns in students' learning that humans simply cannot. In this respect, the potential for algorithms to trawl the vast quantities of learner data in order to establish correlations between student behaviour and educational outcomes has the potential to render much of contemporary pedagogy (i.e. the theory of teaching) redundant.
Thus, in a university where decision-making is governed by reliance on learning analytics, educational interventions are based on robust data (i.e. evidence), as opposed to educators' theoretical suppositions.
The limitations of this approach lie very much in critiques of empiricism in general as a method of social enquiry. In particular, advocates of constructivist epistemologies would argue that data can never be pure (i.e. free from human bias). The generation of data is inevitably led by human assumptions (e.g. about which forms of information should be captured and which should not). In this respect,
the social construction of learning analytics as pure information creates a false sense of neutrality.
As, authors such as Erwin, Bond and Jane (2015) argue, it is essential to acknowledge and utilise the role of human interpretation and sense making in Big Data. This, they argue, involves combining both quantitative methodologies (such as algorithms employing cluster, factor and regression modelling) with qualitative methodologies (such as those based on constructivist grounded theory). From this perspective, approaches to incorporating learning analytics in higher education decision-making need to fully consider the epistemological assumptions underpinning this move, and avoid naively relying on classical empiricist beliefs that learning analytics will render obsolete, the need for (critical) pedagogy.
Learning analytics as Ideology
This final section considers the promises and pitfalls of learning analytics as an ideological force.
It is difficult (if not impossible) to separate the development of learning analytics in higher education from broader trends in the political economy of higher education, both in the UK and globally. As several authors have highlighted, there is a growing international move towards the marketization of higher education (see Levidow 2002 for example). In the UK context, the introduction of tuition fees following the Dearing Report (1997), and the coalition governments' recent lifting of students caps in England (Osborne, 2013), can be associated with a broader policy move to positioning students as
(as opposed to passive recipients) of higher education. In this context, the development of learning analytics has been linked to the potential to
and parents, by enabling them to make better, more informed, choices about where and what to study. As Murray (2013:1), for example, argues:
'When it comes to individual courses, students could identify which ones – based on their personalities – they would be more likely to succeed in, or those that would do most to secure them the jobs they want.'
From a counter-ideological position, one of the dangers of expanding the use of learning analytics is the unconscious reproduction of regulatory technologies that serve to perpetuate neoliberal approaches (Hope 2015). Neoliberalism, advocates of more socially progressive models argue, does not empower students as active citizens. Rather, learning analytics in the service of marketisation is seen as leading to furthering reductionism, whereby the value of education is linked solely to students' employability and their market value (Levidow 2002). Thus, from this perpsective,
the rise of learning analytics risks obscuring educational outcomes that cannot be measured through Big Data
, such as the promotion of students' social citizenship (Marshall 1950/2009).
This essay has sought to explore the promises and pitfalls of the Big Data Revolution for higher education. In attempting to do so, however, one thing becomes immediately apparent; that it is not possible to talk about Big Data as a single entity. Rather,
Big Data needs to be understood on a series of fronts, including as practices of data collection and analysis (practice), as techniques that reflect basic assumptions about the nature and value of knowledge (epistemology) and as instruments employed to further socio-political goals (ideology).
The promises and pitfalls of Big Data cannot be evaluated outside of these divisions.
Drawing on this tripartite approach to understanding Big Data, discussion focused on the rise of learning analytics in higher education. Whilst, arguably, the existence of learning analytics pre-dates discussions of Big Data, interest in the process has been fuelled by the growing quantities of online data that exists about the learning habits and behaviours of students. We've seen that learning analytics have been associated with many promises; from offering
to 'at-risk' students and providing
new possibilities for evidence-based decision making
students as active consumers
of higher education. Equally, a series of pitfalls have been identified, that range from
potential infringement of students' civil rights
, to the
perpetuation of neoliberal agendas
that seek to transform higher education into a market-based system.
, it is worth emphasising that Big Data does not, in and of itself, offer a route to a utopian promised land, or a technological hand-cart to a dystopian future. Instead, following Fenwick (2010), Big Data may be more fruitfully understood as systems of material and techniques that are continually constructed (and re-constructed) through human-machine relationships. Thus, the promises which Big Data makes to higher education, and the pitfalls that may accompany such promises, are not pre-determined but are, instead, available to us, to make of what we will.
Thank you! Please post your thoughts below or tweet me @NicholasJenkin8