Monday morning 1 July 2013

George Siemens was first up and posted what he called "Really big freaking ideas" - here they are, together with my own views and questions (in italics).

1. Increase access, rethink learning, redesign structures of learning
Yes - but, that has been holy grail of almost every field involving students for as along as I can remember. Qs: is the goal too broad to get any meaningful traction? Can we redefine the question and then take advantage of the multiple disciplines here to take a transdisciplinary approach

2. Track and improve how people connect/ collaborate and grow knowledge
This follows on from 1.

3. Prepare for a climate of continual complex problems
Is this also too broad? From where I see the field (as an administrator - Vice-President and Deputy Vice-Chancellor) the big complex problem on the horizon is how to do more in universities with diminishing funding. Therefore, my rephrasing of this question would be "How can we make use of academic and learning analytics to make better decisions about where to direct funding"

4. Find the balance between social and technical sensemaking systems
Yes, but that follows the big questions

5. LA as a means of deciding if we are making the right decisions as a system

Not convinced these ideas are as BIG as they might be given the challenges we are facing. Asked this as a question and a suggestion that we use the week to come up with bigger and more focused ideas that we then take a transdisciplinary approach to solving

NOTE: A google doc has been created to gather ideas on bigger questions:

+-------------

Why Big Data & Analytics could transform the Learning Sciences and Education?

Panel - Philip Winne, Dan Suthers, George Siemens, Draga Gasevice

Dragan – should start reviewing our definitions –

learning analytics – what? Measurement, collection, analysis, and reporting of data about learners and their contexts.

learning analytics – why? understanding and optimising learning and the environments in which learning occurs

What values to promote? Targeting “learners on average”, reporting and knowledge transmission or …

 Don't think we necessarily need to target 'learners on average' - see for example the whole field of phenomenography - these researchers work on uncovering the finite number of qualitatively different ways that people understand phenomena. These descriptors could serve as the perfect starting point.

For example, I did a project many years ago to develop a 'Computer-Based Learning' (anyone else remember that term?) program to help students better understand electric current. Our starting point was that students know the formula: voltage = current * resistance but have no idea for example, what current IS. Phenomenographic research told us that there are three common conceptions of current:
1. current is used up in a circuit
2. current clashes in a circuit and that's where the energy comes from
3. current is conserved in a circuit
the CBL program uncovered the student's conception and took them through a different path so they could see why their existing conception did not work, and then help them rebuild their personal theory.

We need to be able to do more of this kind of work!

George – What will LA do for education?

• Add a new research layer
• Personalisation
• optimisation
• organisational insight
• improved decision making
• new models for learning

Dan - challenges for learning analytics

Had a slide showing 3 epistemologies:

1. acquisition
2. intersubjective meaning making eg Scardamalia
3. participatory eg Communities of Practice (CoP), Apprenticeships (Legitimate Peripheral Participation --> centre of CoP)

my question is whether this is hierarchical in the sense for example, that meaning making requires some kind of acquisition 

Fabulous talk by Phil Winne

Started out by saying that unlike other speakers, he does not read from his powerpoint slides. Wondered what other speakers thought about this comment?

Then, close to my heart, he talked about the fact that most studies are way too small to get any scientifically meaningful data from. eg if asking the question "does treatment X boost learning compared to a control group" If you have 6 factors at vary levels eg

• sex (2 levels)
• interest (3)
• prior knowledge (3)
• need for cognition (2)
• goal orientation (2)
• self efficacy (3)

 requires N=12000 at ~ 30 learners per cell.
Reference he gave:

Winne, P. (2006) How Software Technologies Can Improve Research on Learning and Bolster School Reform Educational Psychologist

Said there are 2 origins of possible effects (to lapse into statistical speak): 

1. the way things ARE - usually not under learner control
2. the way learners MAKE THINGS - under learner control - not often studied. (I have seen lots of examples where students just want to pass for example and try to get through with a minimum of learning)

The Psychology of Academic Achievement 

Annual Review of Psychology

Vol. 61: 653-678 (Volume publication date January 2010)

First published online as a Review in Advance on October 19, 2009

DOI: 10.1146/annurev.psych.093008.100348

Talked about data I*3:

• incomplete
• impressise
• impermanent

+-----
Then there was a panel session on Ethics and Privacy but Doug Clow's blog says it all