#lak12 Side effect or main effect?

I'm enjoying the examples of how visualisation of student data is used to let teachers improve their course...
The software is good at presenting data in a way that makes it easy to get an overview, and the teachers are good at understanding their course content, their students, their context and, thus, good at coming up with measures to improve their education.

A positive side effect is that visualizing student data lures the teachers into thinking more carefully about questions like: what am I trying to achieve? what kind of student behaviour would I like to see? and how can I make that happen? It lures teachers into spending more time at carefully designing their education...

Thinking on further, I'm starting to wonder whether this is a side effect or a main effect. When I teach instructional design this is the behaviour that I would like to see in my students: that they spend time and effort (over and over again) on carefully thinking about what they are trying to achieve, monotoring whether these goals are reached, and keep on trying to improve their education....

In practice, a teacher can never pay attention to all problems and never try out all solutions. I am satisfied with teachers that just pick out one or two points and work on those, and pick one or two other points next time. My experience is, that it is often not that relevant which problems you start with, as long as you keep on trying to improve something.

So once I have lured teachers into spending that time and effort, I don't mind if we switch the software off :-)?

#lak12 contradicting trends...

I have just realized that there are two contradicting trends in educational research:
- analzing big data in pursue of Evidence-Based Education
- small-scale qualitative studies where researchers consciously abolish striving for objectivity and generalizability in favour of deep understanding.

Interesting...

#lak 12 week 5 - part 2

I'm splitting this up, so as not to confuse myself :-)

Six provocations of big data (Boyd & Crawford):
1. Automating Research Changes the Definition of Knowledge.
2. Claims to Objectivity and Accuracy are Misleading - all analysis means interpretation
3. Bigger Data are Not Always Better Data - they can be incomplete, biased, noisy
4. Not All Data Are Equivalent - Context matters!
5. Just Because it is Accessible Doesn’t Make it Ethical
6. Limited Access to Big Data Creates New Digital Divides

This summarizes about all my concerns, I think.
Why didn't we read this earlier on?

#lak12 Week 5

I'm catching up on the readings of week 5 and started of with the big Aspen report.

Nice and down to earth.... in some cases correlations can be useful, but be careful: correlations are not causal relations and they have limited value as predictors. If you want more, you still need to develop theories and test them purposefully. Yep.

In many cases you need to think carefully about what data you need, instead of just collecting everything just because you can... because otherwise you create more confusion and distractions.

I hadn't thought so much about the value of visualization of data (but good to be remembered that visualization techniques contain embedded judgements).

The privacy issues confuse me.... on the one hand I do feel powerless, because I don't know which data about me are collected by whom... and I dread the consequences for the less fortunate people: if you are illegal, you can already not go to a hospital any longer, but in future the police will track you down because of your mobile phone or your Facebook account?

On the other hand, I tend to think this is nothing new. Didn't we hate it when big companies were buying sets of addresses to send (paper) advertisements? well, nowadays I have the right to tell them to stop... And don't we already know much more about our risks to get diseases than 50 years ago? And we still have a reasonable insurance system (at least here in the Netherlands). So, we can expect new legislation to develop. Maybe along the lines of the 'information audit' proposed by Stefaan Verhulst.

#LAK12 Running a week behind...

I'm listening to the recording of the talk by Dragan Gasevic that I missed last week (when I'd missed the fact that it was at a different time :-))

I've been struggling to follow until he started talking about the concrete example of what they do in their own place. I can see how this is useful there, but I cannot link it to my own teaching... I've been thinking why.
It has something to do with the kind of course, which might have to be:
- large (with so many students that the teacher doesn't know them all)
- most interaction with learning material & other students online,
- a domain that can be modelled well (how else can we analyze the content of students' messages or contributions)
- a certain kind of content; some set of concepts or knowledge that is explained or 'delivered' in web resources...
The latter seems in contradiction with Dragan's claim that he does not believe in 'pizza delivery'.

My discovery of the week: I have realized that I do believe that learner analytics can be immensely useful to enable students to self-direct their learning. Big problem with that is that  -even though we know that self-directed learning is good for retention and transfer- most students aren't interested in investing all that time and effort, they are interested in doing as little as possible to make their exam. Of course, there are exceptions... among which the participants of this course :-)

Some things I'm allergic to, though. Evidence-based education... measure whether some treatment works... hey guys, education and medicine cannot be compared so easily. For one thing, in medicine it doesn't matter which doctor prescribes the pills, in education that is not true. And if we know one thing from research it is that the teacher is an important success (or failure) factor. It would be more fair to compare education to psychotherapy really, where the bond between client and therapist has been identiefied as the main success factor. I'll stop moaning about one of my hobby horses now..

#LAK12 listening to John Campbell

I am enjoying this talk a lot more. It is very concrete... using learner analytics to support something that we know that works: a school that cares how their students are doing :-)
Can this help us to make a mentoring system financially feasible?

Half an hour later and John confirmed my ideas: students think that faculty care more about them...
I understand that he is dealing with large numbers of students in a course. I can see this making sense then, because the staff wouldn't know all the students anymore. I would expect that small-scale education where students are not anonymous will do better, both in performance and student satisfaction.

P.S. Relating to my previous post: I particularly like the fact that faculty can change the system's decision. I am sure that this has been an important success factor with regards to acceptance.


Given my previous concerns, I did appreciate one of the last slides :-)

#LAK12 Preparing data for human inspection

This is the sentence that stuck with me from last week, and something that I could see a lot of value in.
We are notoriously bad at keeping an overview over large amounts of data, or for that matter, in filtering data in a sensible way. The latter, of course, requires some careful consideration of which data are important.

Had a long discussion about this last week with one of my colleagues. He advocated the view that we should allow the data to come up with unexpected conclusions and that we would miss those if we always decide beforehand what we want to know. My thoughts: true.... but only adviseable for people who have enough time to do this & look at the stuff that we know is important :-)