Ethics, Learning Analytics and Disability

Today I have been writing a contribution for a paper requested by the Open University’s Ethics Committee about ethics in Learning Analytics.  This blog post is adapted from that.

There are two broad use case scenarios where learning analytics approaches may benefit disabled students:

  1. Targeting support to disabled students or their tutors (Support)
  2. Identifying online activities that seem to be problematic for some disabled students (Accessibility)

As far as we are aware these approaches are yet to be deployed anywhere world-wide but we are actively researching them here at the Open University where we have approximately 20,000 disabled students.  We envisage that if the early promise of this research holds up, deployment on about a 3 year horizon.  These approaches, especially the accessibility one, are reported in more detail in Section 5. of Cooper et. al. 2012.

Firstly, a few definitions:

IMS Global Learning Consortium offered education-specific definitions of both disability and accessibility when introducing its work on the development of technical standards for accessibility in e-learning:

[…] the term disability has been re-defined as a mismatch between the needs of the learner and the education offered. It is therefore not a personal trait but an artifact of the relationship between the learner and the learning environment or education delivery. Accessibility, given this re-definition, is the ability of the learning environment to adjust to the needs of all learners. Accessibility is determined by the flexibility of the education environment (with respect to presentation, control methods, access modality, and learner supports) and the availability of adequate alternative-but-equivalent content and activities. The needs and preferences of a user may arise from the context or environment the user is in, the tools available (e.g., mobile devices, assistive technologies such as Braille devices, voice recognition systems, or alternative keyboards, etc.), their background, or a disability in the traditional sense. Accessible systems adjust the user interface of the learning environment, locate needed resources and adjust the properties of the resources to match the needs and preferences of the user. (IMS Global 2004)

Thus disability is not an attribute of a person, but an attribute of the relationship between that person and the tools they are using to meet their goals; in this case online learning.  And, accessibility is a property of the learning resources that makes is usable by all, including those traditionally labelled as disabled.

The principle ethical dilemma when approaching learning analytics and learners who might experience a disability in the context of online learning is:

  • For what purpose has the individual students declared their disability to the university or other educational establishment, and is this consistent with how that information is to be used in the learning analytics approaches?

No other literature has been found explicitly addressing this issue.  So this blog post might represent the first public statement of the problem.

At the Open University students who declare a disability so that they can be provided with support in their studies.  This is consistent with the first use case scenario (Support).  It is a moot point if it is consistent with the second use case scenario (Accessibility).  More critically at this stage of development of these approaches it is not obvious that it is consistent with research into these approaches.  Is it ethical to use historic or current data relating to students with disabilities to undertake research into future approaches of applying learning analytics?


Cooper, M,Sloan, D., Kelly, B.,  and Laithwaite, S. (2012) A Challenge to Web Accessibility Metrics and Guidelines: Putting People and Processes First, Proc. W4A2012, April 16-17, 2012, Lyon, France. Co-Located with the 21st International World Wide Web Conference.

IMS Global Learning Consortium (2004), IMS AccessForAll Meta-data Overview. Available online at: (accessed 17/02/14)

Learning Analytics for STEM – disabled student support/accessibility LA4STEM (#la4stem)

Today I submitted and internal Open University project bid to a programme called eSTEeM.

I post here the project description.  N.B. at this stage this is just a proposal. However we should hear by 31 October 2012 if this has been supported as an eSTEeM project and funded. If so I might be blogging much more about this work and its findings.

May I remind readers I set up a LinkedIn Group to try and tease out if there was anyone worldwide doing anything in the area of Learning Analytics and Accessibility. There has been some interest (the group currently has 75 members) but no one has yet shared that they are doing substantive work.  So you never know LA4STEM but in the future be seen as seminal. 😉

If you are interested in this field may I commend to you SoLAR – The Society of Learning Analytics Research:

I will be giving a 30 min presentation about this work at this event  – it’s a long way to travel for me 😉 – it’s OU main campus where I work :

SoLAR Flare UK (19 Nov 2012) #flareUK

Mon 19 Nov 2012, The Open University
Jennie Lee Building, Walton Hall, Milton Keynes, MK7 6AA [

Feel free to post comments or questions!

LA4STEM Project Description

The LA4STEM project will review the potential of Learning Analytics in higher education, specifically in STEM, and with an emphasis on supporting disabled students and facilitating accessibility enhancements.

Learning Analytics is defined as the measurement, collection, analysis, and reporting of data about learners and their contexts, for purposes of understanding and optimising learning and the environments in which it occurs. Learning analytics is a “hot topic” in eLearning and was the second headline topic in the 2-3 year time to adoption section in the 2012 NMC Horizon Report on Higher Education[1]:

“The larger promise of learning analytics, however, is that when correctly applied and interpreted, it will enable faculty to more precisely understand students’ learning needs and to tailor instruction appropriately far more accurately and far sooner than is possible today.”

The LA4STEM project will specifically explore the following STEM application areas for Learning Analytics:

  • Student support (with an emphasis on support for disabled students)
  • Tutor support (facilitating their support of disabled learners)
  • Module review (identifying accessibility enhancements)
  • Retention and attainment (focussing on where disabled students appear disadvantaged)
  • Learning analytics in remote labs (because of their potential for enhancing access to STEM)
  • Recommender systems (the timely direction of disabled students to support and study skills aids; including scaffolding of STEM specific learning activities)

A key output of the project will be an external funding bid for a larger-scale collaborative project.  The work of LA4ALL will inform pilots in this project. Provide envisaged benefits are confirmed, this should lead to enterprise level implementation within the OU and across HE.

The findings of the LA4STEM project will be disseminated, firstly throughout the Science and MCT faculties, then to the wider university. External dissemination will highlight the OU’s lead in this field.

[1] Johnson, L., Adams, S. and Cummins, M. (2012) The NMC Horizon Report: 2012 Higher Education Edition. The New Media Consortium, Austin, Texas:

Use Cases and Scenarios for Learning Analytics Supporting Disabled Students / Accessibility Enhancements

This blog post makes available publicly simple use cases and scenarios we are using in various internal projects and discussions at the Open University on the application of Learning Analytics to support disabled students and to identify accessibility deficits in online teaching and learning. Scenarios based on potential use cases are useful as design tools and in communicating ideas around complex systems with multiple actors. Three scenarios are presented here that were adapted from those created for a concept paper [1] for the Society for Learning Analytics Research (SoLAR – Credit goes to my colleagues Rebecca Ferguson and Annika Wolff for writing them. Discussion of these use case scenarios is deferred to subsequent blog posts but comments and questions on them would be most welcome.

Scenario 1

Kris is a student who has access to a dashboard of analytics that provide him with feedback when he is at his computer or using a mobile device. He has set the dashboard to send him a weekly summary of his activity on university sites and on a set of external sites where he has chosen to share his data with the analytics system. He receives basic statistics on attendance, participation and marks on his formal assignments and exams. He receives personalised recommendations suggesting resources and contacts available at his location and relevant to his range of learning interests. However, what he finds most useful for reflection are the visual ‘mirrors’ that the system presents to him, plus suggestions of ways in which he might become a more effective, strategic learner.

Scenario 2

Jenny teaches one of Kris’ courses. Her dashboard is designed for educators, and can be configured to illuminate problems and progress on the course. Visualisations provide an overview of the course’s social network and indicate students ‘at risk’, as defined by a range of algorithms that match online behaviour to predictive models based on past cohorts. There are different categories of risk, so Jenny can easily filter to see more details behind each student’s classification. One risk category flags students who have a disability and who are nearing a point in the course that previous analysis has identified as being a potential problem area for such students. Kris has a visual impairment and Jenny sees that this is a good time to enquire whether he needs any additional support. She logs all contact (including the form it takes), generating data that can be used to evaluate the effectiveness of the predictive methods. Jenny has also chosen to view aggregates of all the personal Learning Analytics on her students’ dashboards, providing a deeper level of insight into how they are self-reporting and evidencing their progress.

Scenario 3

Natalie is a module manager at the OU. She is exploring statistics from a previous module presentation in order to discuss changes for the following presentation. Natalie can see from the visualisations of VLE data combined with assessment data that students who performed poorly on the second assessment had engaged significantly less with quizzes than those who did well. The system flags up that a significant proportion of these students have declared a disability. Natalie compares datasets and finds a similar pattern on other modules, indicating a potential accessibility issue with the quizzes presented alongside module materials. She undertakes some follow-up activities, contacting a percentage of students from several modules with a questionnaire to try to identify the cause of the problem before passing her findings to the university’s accessibility experts for detailed evaluation and remedial action.


[1] George Siemens, Dragan Gasevic, Caroline Haythornthwaite, Shane Dawson, Simon Buckingham Shum, Rebecca Ferguson, Erik Duval, Katrien Verbert, Ryan S. J. d. Baker, (2011), Open Learning Analytics: an integrated & modularized platform – Proposal to design, implement and evaluate an open platform to integrate heterogeneous learning analytics techniques. Available at: (accessed 12 October 2012)