The context is missing but is important. When I say MOOCs have failed I mean something very specific: asynchronous online learning has proven to be so dramatically less effective than synchronous, guided learning. Better than nothing existing, but we’re learning that only a small subset of people are able to effectively learn from them.
In other words, take a random sample of 100 people with the same level of motivation. 90% will complete a live, guided course. 5% will complete a MOOC of the same quality. Which is unfortunate because the live/synchronous part is what is expensive.
There was a time when we thought all education was is just a bunch of video lectures, and that if we got a bunch of professors to record their lectures college would be free.
We now know that doesn’t work, at least for the vast, vast majority of people, both because we’re unable to discipline ourselves and because most true learning happens in a multi-way environment with accountability measures.
From my anecdotal experience, the people taking the "live, guided course" have committed to:
- synchronous attendance at a particular time, probably at a particular place
- probably a dollar investment
- face-to-face contact with instructor
- near-traditional learning environment
People taking a MOOC have committed to: - browsing on an electronic device
There's generally a really high barrier to entry to the synchronous, guided learning, so it seems like comparing students who have paid tuition and enrolled in a class vs prospective students expressing interest on a university recruitment page.Internally we have taken sets of students that apply to our free “mini code bootcamp” and sent half to an archived version published each night and send the other half a live link that becomes an archive each night. So we didn’t control scientifically, jsut an AB test, but the shift has been breathtaking.
We can also take students that are struggling learning from our MOOCs and they finish the live course almost every time. I know the latter is unscientific, but it becomes obvious over time.
https://www.theatlantic.com/education/archive/2014/01/the-tr...
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2381263
"In these course[sic], “dropping out” is not a breach of expectations but the natural result of an open, free, and asynchronous registration process, where students get just as much as they wish out of a course and registering for a course does not imply a commitment to completing it."
...
Also the discussion of the "Colbert Bump", where increased publicity to less-motivated (just curious) students doubled completion but tripled registration (thus reducing completion rate).
Also the point about whether to allow or disallow registration after the course begins (late registrants are ineligible for certification, so they are automatically "failed").
"HarvardX could boost its certification rate by closing those courses to new registrants or restrict courses to those most likely to complete them. Instead, it keeps courses open to maximize the number of students who are learning something new."
I absolutely agree that (at least some) users can be helped by live instruction or supplemented online instruction, but I disagree that commonly cited, low completion rates measure students with the "same level of motivation".
For instance, are MOOC students more likely to not complete the course because they have already found work in the meantime? Perhaps because they have more freedom to look for work during business hours, for example, rather than sitting in on a live class during the optimal employment search time?
If the goal is to find work, there is no incentive to keep going in the course after you have met your goal.
If you want the stats for our full computer science academy I can give you those; they're good. 50% hired within one month, median $90k salary.
Beyond broadening access, if you have to be somewhere in person you have to make it priority at that time - and you are not interrupted while you are there. In home study is very easily interrupted and easy to postpone due to other very real duties until it is too late.
The motivation hit from meeting other students is also missing.
This first couple of things I'd check would be:
- Double-blind: What if you tell students it is live, but it is actually pre-recorded?
- Are the live+archive and archive-only videos the same?
- Are the live+archive students actually watching the video when it is live, or are they using the archive feature?
- Do archive-only students who consistently watch at the same time do better?
- Does time/day of watching affect results? (archive-only students get the link at least [duration of class] later.)
You also need to control other interactions with the participants, track the results faithfully, avoid bias because instructors are more familiar with students from "live" interactions, ensure cohort size and result size are large enough to be meaningful, that group composition isn't biased by confounding factors, etc.But random assignment is a good start!
When I take a MOOC, I'm pledging at least 5 hours a week. That covers 2 hours of videos, 2 hours of homework and 1 hour of additional research / review.
If I pledge 5 hours /week to a traditional class, I don't get nearly as much learning. I have to commute to the class, which is at least 30 minutes each way. Now my 5 hours buys me 2 hours of class, 2 hours of commuting and 1 hour of homework. I've lost an hour of homework and an hour of research.
I know which one is better for me.
With the side issue that 100% of everyone in my 20 person automata theory class back in ninety-something is a much smaller number than 2% of 24000 people signing up for Ullman's automata theory MOOC.
The fundamental problem is authority along with excessive broadness. If I want to learn Y and the MOOC teaches X Y and Z, I will fail the MOOC. There are also Skinner box conditioning issues where my assignment for something like using bootstrap framework, will be scratching my own itch of project or doing some project at work, such that again I'll fail the MOOC because I'll refuse the assignment; I already have a more ambitious self assigned assignment and am not doing both my assignment and someone elses much less interesting assignment. Whereas if I sit in class I'll have peer pressure and financial pressure to do the class assignment regardless of work or hobby interests.
Sort of a traditional in-person or online class is a goal in and of itself; MOOCs are a means to an end for work or hobby for most.
What does complete mean here? Complete within a given timeframe? Doesn't that miss the entire gain that MOOCs bring to the table? Which is that education becomes a lifelong endeavour instead of something you try and fit into a few years and, for most, never return to again?
Sounds like MOOCs need to learn from educational institutions that actually have a track record in distance learning.
Saying MOOCS are a failure because only 2% complete the course is like saying the OU is a failure because only 2% of their web site visitors sign up.
I suspect that a MOOC offered with the same incentives and structure as a live class - ie, as part of a degree-granting program with mandatory tests at set dates - would have significantly higher completion rates.
If it is the former, then it is definitely a success. If it is the latter, then I think that not only percentage rates of success are important, but also the nominal completion rate itself.
A MOOC supplies the opportunity for someone getting a degree in communications, business, or some other not-entirely-technical degree, that might be wonder what a computer science or high level mathematics course is like to go and find out. Does their failure in such a course mean that the class itself is a failure, or can the fact that they had access and were able to put a toe in the water mark success?
I would suggest the latter.
In the old days if you wanted to learn something, often you'd pay $20 to $50 for a book. For example, Amazon order search claims that somewhere I have a copy of "Paradigms of Artificial Intelligence Programming" by Norvig, which is supposed to be a decent book in the field, cool. Ironically maybe that IS the textbook for the famous AI MOOC class, I don't know, maybe it should be if it isn't. However I never finished the class, only glossed thru the book, and barely remember either.
I was curious about the topic, not trying to meet someone elses arbitrary goalpost, and the "graduates" metric isn't a useful number.
Kinda the difference between running because its a beautiful day out and I need exercise, and running a competitive marathon against live athletes in person solely for the goal of getting a number when I run thru a gate. I would guess in the general public the former outnumbers the latter 50 to 1, so the MOOC stats aren't unusual at all.
Another interesting analogy is "The Mona Lisa" painting is a failure if you measure its worth solely based on number of PHD students graduating who give it credit for them selecting "Art History" as a major. That doesn't mean it needs to be burned as a failure.