Daily chronicle – 4th day of the Leadership School

By Julie Bu Daher & Laura Infante Blanco

Session 1 of day 4 of the leadership school

Yannick Toussaint (University of Lorraine), Mandla Makhanya (University of South Africa) and Gard Titlestad provided an introduction for the Learning Analytics session. They launched the discussion on digital transformation of Higher Education, open distance e-learning and online educational systems and the importance of learning analytics.

After the short introduction, there was an address by Wim Van Petegem (eden) and George Ubachs (EADTU).

Wim Van Petegem gave a description of the European Distance and E-Learning Network – EDEN. Its mission is to support endeavours to modernise education, recognize excellence, facilitate knowledge and practice exchange, improve understanding among professionals and promote policy and practice across Europe and beyond. Their activities include conferences, open classroom conferences and research workshops, and they are involved in many projects like openmed and elene4work.

George Ubachs presented EADTU that is now appointed as a focal point on quality assurance in online and distance education for Europe. They have many European as well as global partners, and they are involved in research publications in the domain. He highlighted that LA is not just collecting and analyzing data, it also involves using this data for efficient problem solving, study progress and student retention.

The first keynote was presented by Barbara Wasson (SLATE University of Bergen).

She gave a description of SLATE: center of the science of learning and technology. Its aim is to clarify and explore concepts such as learning analytics, data in education, assessment for learning, and learning & technology. She then talked about learning analytics and developed a search string of keywords from different conferences on learning analytics such as learning, retention, prediction, modeling and decision support. She detailed the history of LA in the field of research, some examples are SoLAR, EDM and LAK. She also detailed some research projects in this field, some of them are visualizing activities on dashboards to show how students are active on it, providing dashboards for instructors, getting emotions of students on MOOC and ethics and privacy detailed in a JISC report.

She concluded that LA is still in its infancy and the definition of learning analytics is still under discussion. She also believes that although data is rich, the theory remains fairly poor in the field.

After this interesting keynote, there was a talk about learning analytics and policy by Anne Boyer (Université de Lorraine).

Anne discussed the impact and potential of learning analytics. She first described the university of Lorraine, and she discussed from the HE landscape that there is growing number of non-traditional students, redefinition of the role of HEI in society and presented applied examples of LA and its impact on students. Some of the described examples were about behavior of students on early alert systems, using tools to compare their activities, detecting change of students behavior as a response to the data presented on LA dashboards and many others. She also presented the results of a study showing that a high percentage of students prefer to know early if they are at risk and to be contacted through university email. She then described the METAL project that aims to improve the quality of learning, provide dashboards for school students as well as for instructors and help students to improve foreign language pronunciation. She concluded that students need transparency of algorithms and explanations on the results provided to them, and she also stressed on the importance of ethics and privacy in this domain.

After Anne’s presentation, Dai Griffiths discussed learning analytics and policy.

He presented LA and its long history of research, and he provided the scientific terms used in this field like distance education, educational data mining, technology enhanced learning, e-learning and adaptive learning. He also discussed the correlation of operations research and LA methods, and the ethical and privacy issues concerning the data used in the domain of LA. He concluded that institutional and governmental policies for LA are important, but do not operate in a vacuum and that LA policy is enmeshed with major policy debates.

After this talk, a workshop of related questions and answers was conducted where the all members of the conference  shared ideas and raised some interesting questions mainly on  personal data, ethical and privacy aspects in the domain of LA.

Keywords: Educational data mining, data transparency, adaptive learning, LA, data ethics

Session 2 of day 4 of the leadership school

After lunch, Rafael Molina (Loma Linda University) was in charge of presenting the following panel: organizational issues of a LA initiative – towards a methodology.

The first speaker of this panel was I. Dolphin from APEREO foundation. The mission of this foundation is to support learning, teaching and research. APEREO provides a wide variety of open source software, such as the famous CAS or the Sakai LMS, but the main goal of his presence was to present the LA Initiative, which is a community of researchers interested in Learning Analytics who are actively developing an open LA platform with an objective of globalization of LA. The platform provides already a Data Warehouse and some predictive functionality for intervention. Evidence of learning outcomes will be coming soon, as well as predictions for learning pathways (adaptive learning).

C. Koh followed presenting the case of the University of Singapore. They have set an Institutional Research and Analytics Unit (IRAU) with the aim of building up the analytics capability, facilitating analytics projects, developing warehouses and platforms for information sharing and decision making about learning activities (training and learning, institutional and unit projects, dashboards and deployment, integrated informations systems). He highlighted the challenges they need to overcome for the development of the LA strategy, like convincing the actors to use LA in their decisions, issues concerning data integrity like security and confidentiality, and training people in developing LA skills. They are already capable of detecting students at risk, and future directions include developing strategies and frameworks, demonstrations of usefulness, mentors awarding, etc.
Then it was the time for Mark Brown to present the example of Dublin City University.

He spoke about one of the major projects in LA on his institution, PredictEd. This project aimed at using VLE data to provide weekly automated feedback on student engagement based on the learner’s activity. The project was a success as average scores for participants were higher in 8 of the 10 modules analyzed, according to an evaluation considering 2 groups, those who used the system and those who didn’t. Mark Brown was concerned about viewing education as a simple process that can be easily modelled or making decisions based on rich data but with weak theories or based on wrong data or even invalid variables.

After the questions for the panel members, T. Belawati (Universitas Terbuka) introduced the session entitled: Impact of a LA strategy on universities management

Niall Sclater (JISC) was in charge of the keynote for the session. He started enunciating the main reasons why an institution should engage a LA plan such as data-informed decision making, understanding and quantifying educational processes, meeting government requirements or and pressure from students. LA can deal with many university issues including enhancing student learning experience,  improving retention, providing students with better information on their progress, improving national student survey scores, enhancing teaching, building better relationships between students and staff, or providing additional support to underachieving groups. He then presented some successful use cases like Nottingham Trent University, University of Technology Sidney, University of South Australia and he gave the public some tips for developing a good LA Strategy, for instance to get Interested in better reporting and dashboards, rather than simply predictive analytics at this stage of maturation of technology.

He finally recommended starting on a small scale, supporting and empowering the key stakeholders and distribute the learning analytics governance between structures.

After the keynote, Wei Shunping (Open University of China) made an introduction to Big Data and LA, and the use of this concepts in the core of his institution. UOC is an internet-based university, with more than 1 billion cumulative log data in the last 8 years coming from basic data, resource data and behavior data. They apply techniques of Data Mining and LA to achieve decision (policy making), learning (personalized learning), management (fine management), research (learning rules discovering) and continuous optimization. Their LA models are based in learning behavior (who, what, where, when, results, actions) and they have implemented many application cases such as data-driven evaluation of online instruction and learning process and a width-depth-persistence model for learning process evaluation.

Next was N. Fassina (Athabasca University), who highlighted the main motivations of LA in decision making, especially in the most relevant ones, the long term ones, and accountability pressure related to the turbulent public economic situation around the world. He explained that decisions have to be taken according to the needs of the institution, and not by the mere fact of adapting to new technologies. For instance, having a unique Data Warehouse containing all data produced by the multiple technological services of the university … but for what purpose? Ok, we have an amount of data, but, shouldn’t we wonder before what do we want to do with it? He also threw some other questions to ponder: How to have a feedback loop? How to influence governments to make the right policies  to permit doing LA correctly at university level?

After the break, there was an interactive workshop where participants worked in groups in order to answer the following questions: what are the main issues at your institution that you think could be addressed with LA? what resources do you need to get this work accomplished (or at least, how will you establish what resources you’ll need)? what do you perceive as the main barriers to success of the initiative and how will you overcome them? The interesting answers were compiled in a document … coming soon!

Keywords: LA, Data Warehouse, Data Mining, personalized learning, predictive learning

 Tips of the day:

  • LA is still in its infancy and the definition of learning analytics is still under discussion
  • Students need algorithm transparency and explanations to the provided results
  • Data ethics and privacy in LA are critical and challenging aspects
  • Absolutely vital to success is having a leader with a deep scholarly understanding of LA principles and practices and the mechanics of creating predictive models
  • Start on a small scale
  • Support and empower the key stakeholders
  • Transparency and openness are vital for success
  • Distribute learning analytics governance power structures.
  • Design good feedback loops

Comments are currently closed.