Friday, March 9, 2012

Two distinct course formats in the delivery of MOOCs.

MOOCs have been carried out with great success during the last years. Examples are CCK08, PLENK2010, MobiMOOC (2011), EduMOOC (2011), Change11, and LAK12. Their implementation requires conceptual changes in perspective from both “facilitators” (tutors) and learners. They are so novel that much research needs to be done for their understanding. 

Basically 2 very distinct delivery formats have been used:

  • Those that use what’s called an aggregator: a newsletter called “The Daily”.
  • Those where all events go through a “centralizing” web page and discussions happen with the use of a mailing list: in most cases using Google Groups.

Each of these will of course have a different impact on the behavior of learner’s experience and the outcome of the course.

In this blog I expose some ideas on studying this problem.

How many participate in MOOCs

For our discussion it is important to understand how many participate in MOOCs and of these how many are active and how many take a passive role behind the scenes. Lurker is the term used for the latter.

In a recent post Geoge Siemens (see reference 1) gave the following numbers and pattern of participant’s behavior for “change11” but these could be considered representative of all MOOCs.

“The Change MOOC has about 2400 participants, yet we typically get about 40 participants per live sessions, 5-10 blog posts a day, and 20+ daily tweets related to the course. Some are active throughout the course (though when I did an analysis on CCK08, only a few of the most active participants in week 1 were still in the top ten by week 12), some have spurts of activity, and others subscribe to the daily but don’t engage in ways that are visible to us as facilitators. Consistently, as the course progresses, active participation declines.”

Delivery formats

Format 1:

Many MOOCs have utilized a daily newsletter named “The Daily” which basically aggregates contributions from all blogs (or other resources) from participants tagged in a certain way. Examples are CCK08, PLEN2010, Change11 and LAK12.
In these cases it is nearly impossible to track learner’s behavior except for those that are active participants. The work of Rita Kop in Plenk2010 represented an exception since an effort to track lurkers behavior was done by implementing surveys and other strategies.

Those that participate as “active”, have a certain degree of expertise in the course domain and confidence in exposing their writings.
Lurkers seem to restrain to make any kind of appearance but will burst into a blog if that post (announced through The Daily) is of interest. The number of lurkers at any time can be as high as 50% of registered participants.
These MOOCs follow in part a pattern similar to that described in Reference 2: “the MOOC mirrors a discussion at a conference, in a research lab, or in a workshop”.

Format 2:

Some MOOCs employ a “centralizing “web page used by the facilitators for announcing all activities and a mailing list open to contributions from participants(mostly Google Groups). Threads on different subjects are opened and continuously a participant receives the new contributions to the different threads.
Examples are MobiMOOC (2011) (556 registered) and EduMOOC (2011) (2700 registered).
Since adding an opinion or just a comment to some discussion thread does not need to show expertise, a dormant lurker becomes active just in those occasions. Participants get to know each other more since these occasional appearances of lurkers makes them visible.
This second format is closer to the idea of “eventedness” described in Reference 2: “The course members resemble the people in a corner having an in-depth discussion that others can choose to enter. Enough structure is provided by the course that if a learner is interested in the topic, he or she can build sufficient language and expertise to participate peripherally or directly. The more people who walk over to talk, the better the chance will be that people will contribute to the conversation”.

Dropout rate.

Finally, let me introduce a small comment on the dropout rate.
In reference 2, Cormiere and Siemens write:
 “The most disconcerting issue for many educators running an open course is the drop-out rate”. 
And in reference 1:
“While active participation in our courses declines as the course progresses, subscribers to the Daily increase. I’m not sure what to make of that. If I was getting five emails a week on something I wasn’t interested in, I would unsubscribe. Does that mean we can view Daily subscribers as a) people are still engaged, b) people can’t find the unsubscribe link, or c) that we’ve subjected over 15,000 people to guilt about not being active in MOOCs?”

The answer to the last question is a) (people are still engaged) and in reality the most disconcerting issue to those running a course comes from not realizing that lurkers might conform a high percentage (difficult to quantify precisely) of those registered.

Tag for #change11 and  #lak12

  2. Cormier, D., & Siemens, G. (2010). Through the open door: Open courses as research, learning, and engagement. Educause, 45 (4), 30-39. Retrieved October 20th, 2010 from: 

Wednesday, March 7, 2012

MOOCs, the CYNEFIN framework and understanding the “basics”.

In a recent post Dave Cormiere proposed that the CYNEFIN framework as developed by Dave Snowden could help describing rhizomatic learning (MOOCs). Broadly CYNEFIN offers 5 “categories” for separating kinds of decisions that can be made: simple issues, complicated issues, complex issues, chaotic issues and disorder.

In this post I would like to contribute some thoughts to the subject and see if I can get some feedback from those in LAK12 and change11, so as to understand better the ideas proposed.

In the post Dave states:

  1. “If you are looking for ‘best practices’ in a given domain, the MOOC is a fantastically inefficient way of acquiring them”. 
  2.  “If you are looking for ‘good practices’ a MOOC is probably a better option than for simple practices, but it’s still not exactly designed for that.”
  3. “If you are looking for a ‘chaotic experience’ MOOCs are probably a little tied tight for you.”
  4. “The complex domain is where the MOOC really shines. If you want to try things, see how it goes, and build from that response, a MOOC is just the ecosystem you need”.
As I understand 1 and 2 refer to “what we learn” (best/good practices in a certain domain) while 3 and 4 on “how we learn”.

All 4 statements are correct when applied, for example, to MOOCs like CCK12, EduMOOC, Change11, LAK12. No learner in these MOOCs was looking for best/good practices.

Counter example:

MobiMOOC (2011) was a very successful MOOC on mobile learning, definitely rhizozomic, that filled all requirements spelled out in the literature for being a MOOC (educause (2010) paper by Cormiere and Siemens). But:
  • If you were looking to learn on “best practices” in mobile learning this was the place.
  •  “Good practices” were also part of the learning space. Not in one expert, but in many distributed through the network. Mentorship was spread through the web.

MobiMOOC, being a MOOC, satisfied points 3 and 4.

A few thoughts on the question on understanding the “basics”

In his post Dave states: By basic here i mean ‘turn on the computer’ rather than define a computer”

I agree and here are some my thoughts:
  • MOOCs are not suited for teaching/learning efficiently “the basics” in any domain.
  • It’s fundamental for learners in a MOOC to have a certain degree of preparation. If someone participates as a lurker and is unprepared he will not understand. If someone wants to be an active participant he can contribute nothing. The learners are nodes of the network and must contribute in part with their knowledge.
  • The 101 running course in Python programming of would be impossible to carry out in the MOOC format.
 Tag for #lak12 and #change11

Saturday, March 3, 2012

Vast Lurker and No-lurker Participation in Open Online Courses: MOCCs and the AI Stanford like courses respectively.

Open online courses with a massive number of students have represented an interesting development for online education in the past years.
They have basically followed two very different formats: MOOCs and courses similar in spirit to the AI-Stanford course.
In this post I analyze the behavior (both in number and pattern), for both types of the massive courses, of what are described in the research literature  as lurker participants (see Rita Kop, 2011).
MOOCs represent an emerging methodology of online teaching with a structure inspired by the philosophy of connectivism. During the last years they  have been carried out with great success. Examples are CCK08, PLENK2010, MobiMOOC (2011), EduMOOC (2011), Change11, and LAK12. Their implementation requires conceptual changes in perspective from both “facilitators” (tutors) and learners.
These courses can be classified within the connectivist pedagogy (Dron and Anderson 2011, see also a previous post).

Figure 1 (extracted using google analytics to the home page of EduMOOC 2011) represents a typical behavior pattern of those participating in a MOOC. A big number register (2700 in this case) but after a few weeks the active participants reduce to less than 100. Activities like online meetings do not register more than a few tens. Participation in surveys is also small.
Figure 1. shows the number of visits from new visitors (dots) and returning visitors (squares) as defined in the Google Analytics analysis of  the main web site in EduMOOC for the period extending a week before the start until one week after.

Then an important question emerges: have more than 90% of registered participants dropped the course?
Lurker is a term used to define a participant that just follows the course, looks at the recordings, and browses the available course resources. He is mostly behind the scenes waiting for some interesting event as can be seen in Figure 1 and quantified in Table 1. A successful blog post or a particular debate posted to the Google group mailing list may obtain responses that could be 50% of those registered.

Table 1. shows the number of new and returning  daily visitors to EduMOOCs main web page. W0 is the week before the start and W8 the ending week. Thursday was chosen as the sampling day. The total number of unique visitors during the 8 weeks was around 10.000.

The AI-Stanford like courses and
The 2011 AI-Stanford class on Artificial Intelligence taught by Sebastian Thrun and Peter Norvig was also a massive open online course with 160,000 registered enrollees of which 20,000 completed all coursework. It was offered free and online to students worldwide from October 10th to December 18th 2011. A very similar pattern is followed by courses released and still in progress by
The Ai-stanford course included feedback on progress and a statement of accomplishment. The curriculum drew from that used in Stanford's introductory Artificial Intelligence course. The instructors offered similar materials, assignments, and exams.

These course can be classified within the cognitive-behaviorist pedagogy

Figure 2 shows the number of participants through the duration of the AI-course course expressed as daily reach (analytics extracted using A huge peak surges to nearly 100.000 (the daily reach of around October 10th (the beginning of the course). Very rapidly it stabilized at 25.000 active participants. The smaller peaks are linked to the weekly obligatory exams. Practically no lurkers participate and the change from 160.000 to 25.000 simply represents dropouts. 

Figure 2. Number of active participants in the Stanford AI-class.

Two very different course formats.

From previous studies it has become evident (George Siemmens 2012) that we are in the presence of different formats:

  • the AI-Stanford participants have totally different learners goals and preparation than those in MOOCs.
  • there exists a very different nature of the subjects studied: engineering  and  educational theory.
  • the AI-Stanford course falls into the cognitive-behaviorist pedagogy category and the MOOCs  into the connectivist.
The retention and lurker behavior described above adds another differentiation to the previous list.

tag for: #lak12 and #change11