PhD thesis on learning object reuse, and some ponderings

June 9, 2009, [MD]

I'm stuck at Dubai airport for 12 hours, coming from Beijing, and transiting to Milan, where I will be attending ElPub 2009 (and later OAI6 in Geneva). I could go downtown, but I'll be spending four days in Dubai on the way back, so it didn't seem worth it. The airport isn't that exciting, but it does have free wifi and occasional power-outlets, which is a huge plus.

Via David Wiley's blog, I came across Sean Duncan's just completed dissertation on the (lack of) reuse of learning objects, where he did a study of Connexions. I grabbed the .doc from, and since I didn't have any good movies left on my laptop, I settled down to read it in the airport.

Digression about document formats A little digression here about document formats. For things that I am just going to be reading on screen, I find it annoying to download large .doc or .odt files. I seldom have Word of open, because I rarely use them (preferring tools like Scrivener for authoring, and for the final polish). Just starting Word slows my computer down, let alone opening a several-hundred page Word document. In these cases, when I am not expected to be editing the document, I much prefer PDFs. In fact, the first thing I did was to open the document in Word, convert it to PDF, open it in my PDF reader, and close Word. My PDF reader quickly opens many hundred page-PDF documents without even hesitating.

On the other hand, PDFs are only good for one thing - seeing the document on the screen (or printing it), exactly as it were. Which excludes two important things: one is reformatting the document for nicer vieweing (or use on other devices), and the other is reuse. For example, I hate reading double-spaced documents (1.5 line spacing is tolerable), but in a PDF, there is no simple way to change this (in Word, it's extremely easy). I also remember borrowing my friend's CyBook for a week, and got very frustrated when trying to read the book Opening up education on it, which was released as a Creative Commons book, but only in PDF. Trying to display PDFs directly on the ebook is possible, but extremely awkward. Trying to extract the text gives poor results, with mixing of columns etc.

And of course, for reuse, the situation is even worse... What's the point of applying a Creative Commons license that allows for reuse, if the text is "trapped" in a PDF, especially with a multi-column layout? And I realize that I do exactly this myself. I try to put many of my papers, reports, etc. online, with an open license. However, I always upload a PDF... So what's the alternative? .Doc and .odt? Both? I thought the idea of embedding an ODT into a "hybrid PDF" sounded quite promising. Could we do the same, but embed an XML? DocBook? Markdown?

Back to the dissertationSo back to the actual contents. I have been very interested in the question of reuse (or even use) for a long time, since it often seemed like the open ed community was more focused on production than anything else, and especially reuse seemed to be something that most people wanted, but that was not very successful. Thus, Sean's topic is extremely timely.

He starts out with a literature review, done in an interesting fashion. He uses a quite explicit methodology for gathering data: "In the literature search for this study, the original search string [(SU "learning objects" or KW "learning objects") and (TX reuse OR reusable OR reusing OR "re using")], in the Academic Search Premier, Psychology and Behavioral Sciences Collection, and PsychINFO databases, resulted in 43 records."

He then drills further down, discarding non-peer reviewed papers and other non-relevant papers, and ends up with reviewing 25 articles in depth. I am curious about why he choose only these databases, and did not expand his search wider. For example, I ran the search "learning objects" AND ("reuse" OR "reusable" OR "reusing") in Google Scholar, and it returned 9,280 hits. Granted, most of these will likely not be usable, but it seems probably that it would turn out more than 25 that were relevant to the topic in question.

Where I really get confused however, is in the actual methodology and data collection part. He chooses to limit his study to Connexions, which has the advantage that all its content uses a free license, the system has built-in support for translation, modification and collaboration on modules, and the usage data is openly available. Modules in Connexions are basically articles, single text pieces, which can incorporate graphics etc. Collections can contain many modules.

Sean defines the inclusion in a collection as use, in several collections as reuse, and also talks about translation and modification. He doesn't make a very strong case for why this would be valid, however. The first question is how Connexions is actually used - is it used mainly by self-learners, who wish to find useful material for their own studying? Or who want to study an entire "collection"? Or is it used more by educators, who "pre-package" content for their students, into collections? Or are the collections made by some self-learners, who package stuff they find neat for other possible future users?

I don't have the answer, but it seems like we would have to know a lot more about how Connexions is being used, to see if this mapping of statistical indicators to conceptual ideas "work"... Sean has not referred to any of the literature on Connexions, although there are several articles out there. One weakness for example, is that he does not take into consideration reuse of specific modules from outside of Connexions. A simple example is this, where a university curriculum explicitly refers to a specific Connexions module. The way I found this was through the Google search

How long should a PhD dissertation be? One of the things that struck me when opening the file, was how short the dissertation was. The PDF is 74 pages, and at first I thought that maybe it was single-spaced. But no, it is double-spaced, and in fact, there are a lot of appendices, so if we only count the text before the reference list, there's only 45 double-spaced pages in this PhD dissertation. Immediately that seemed very little to me. It might sound very superficial to focus on such an indicator, but it is striking, because it is so different... I recently looked over some literature from the Australian National University, because my wife was thinking about attending, and in their guidelines, they specify that a PhD dissertation should be between 80,000-100,000 words. According to Wolfram Alpha, that gives about 400 double-spaced pages, ie. 10 times as much.

Part of the reason that I am point this out, is that I think that this dissertation would have been much stronger if it had had a wider literature review, and perhaps even a design that combined qualitative and quantitative methods. It could first use exploratory qualitative methods to understand what constitutes reuse in the context of Connexions, and then use statistics to gauge the extent to which that is actually happening. It would also be great to look at, or even test, some different theories about why this might be the case or not. As it stands, I feel like the conclusion is not strong enough - it's telling us that there isn't that much reuse of learning objects within Connexions, but we're not sure why.

All in all, it's an important topic, and it's great to see people picking up the gauntlet. I also know myself how hard it can be to venture into a research area that hasn't been much explored. I congratulate Sean on his achievements and hope he takes my thoughts in good spirit.


Stian HĂ„klev June 9, 2009 Toronto, Canada
comments powered by Disqus