8 Things You Should Know About MOOCs, via @chronicle

8 Things You Should Know About MOOCs

Release of edX data provides fresh insights about who takes massive open online courses

8 Things You Should Know About MOOCs

Before Harvard and MIT released data last month on their first 16 edX MOOCs, we already knew a few things: Millions of people register for massive open online courses, though far fewer receive certificates of completion. Most MOOC participants already have a college degree, even those outside the United States. But there was a lot we didn’t know, especially about who took different types of MOOCs and how much of the course content they viewed. This information may be valuable to those looking to design and lead successful MOOCs. Here’s what we’ve learned from this first data release covering more than half a million students.

1. The overwhelming majority of MOOC students are male

A majority of college and university students in the U.S. are female, but MOOCs flip that balance and then some.

And in some courses, like engineering and computer science, participants are almost entirely male.

Percent of students by gender
 
On-Campus vs. MOOCs
By Course Type
2. MOOCs attract students who already have college degrees

Even though the majority of MOOC students already have college degrees, a substantial number of participants have completed no more than a high-school education.

And yet, the classes where they make up the largest proportion of students are the engineering and material science courses, many of which have college-level prerequisites.

Number of students by highest degree attained
 
All
Computer Science
Engineering
Humanities
Material & Biological Sciences
Social Sciences
3. The median age of MOOC participants is 24

Some courses, like computer science, attract students as young as 12.

But other courses, like those in the social sciences, skew toward an older population. Students in humanities courses are four times as likely as those in engineering courses to be 40 years old or older.

Number of students by age
 
All
Computer Science
Engineering
Humanities
Material & Biological Sciences
Social Sciences
4. One-third of MOOC participants are from North America

Given that the MOOCs are based out of Harvard and MIT, it’s not surprising that the largest proportion of students are from North America. Still, the reach is global — and with regional distinctions.

For example, Africans enroll at twice the rate in social science courses than other courses. South Asians are most likely to take engineering and computer science courses.

Percentage of students by geographic region
 
All
Computer Science
Engineering
Humanities
Material & Biological Sciences
Social Sciences
5. Nearly half of registrants never engage with any of the content

Courses are broken into anywhere from 11 to 48 chapters, but few students look at them all. Only 3 percent of participants look at every chapter, and fewer than one in 10 view even half of the material.

In fact, of those who viewed any course material, half looked at 11 percent of the course chapters or less.

Number of students by percent of coursework viewed
 
 
6. Europeans view the most course content

Participants from East Asia (China and Japan) view the least amount of material, with those from Europe viewing the most. This is true both over all and when broken out by course type.

Meanwhile, the social sciences had the highest chapter view rates, and the humanities the lowest, no matter the region.

Average percentage of coursework viewed by region
 
All
Computer Science
Engineering
Humanities
Material & Biological Sciences
Social Sciences
7. Students with a doctorate viewed more course material

Students holding Ph.D.’s were among the most engaged, viewing more course content than any other education level, on average.

However, students without even a high school diploma were also among the most engaged. In computer science, they ran a close second to Ph.D.’s, and in the humanities courses, they completed more course material than any other education level.

Average percentage of coursework viewed by highest degree attained
 
All
Computer Science
Engineering
Humanities
Material & Biological Sciences
Social Sciences
8. Serial students are the most engaged

The overwhelming number of participants signed up for just one course. But those who signed up for multiple MOOCs — and there were tens of thousands — tended to look at more course material.

That is, up to a point: Engagement seems to drop off after six courses.

Five “super-MOOCers” appear in the data to have taken all 16 courses offered by edX (including some that were different sections of the same course) and to have read almost 20 percent of the course material, on average. We’re skeptical, and think it’s likely this is a data fluke.

Average percentage of coursework viewed by number of courses taken
 
What we still don't know

Granted, these data are still a relatively small sample from a limited number of MOOCs. As the number and variety of MOOCs has grown exponentially since these initial courses were offered in 2012-13 — EdX alone has offered more than 200 courses from more than 30 partner institutions — there are certainly more data that can shed light on other interesting questions. What are the motivations and goals of registrants? What kinds of content engage students the most? Do students cherry-pick lessons throughout the course, or tend to drop out as the class progresses?

These are the questions future MOOC data releases can help us answer, so we can learn even more about how such courses are being used and by whom.

Follow @chrondata on Twitter

About the data: The edX MOOC data released by Harvard and MIT covered 16 sections of 13 courses from the first year of edX, from fall 2012 to summer 2013. About 475,000 students and about 640,000 registrations are included. Although there were 841,000 registrations in these courses, 200,000 rows of data were deleted by HarvardX and MITx in the de-identification process. De-identification was most likely to remove outliers and extremely active users, which may affect some of the analysis. You can learn more about the de-identification process from the HarvardX and MITx documentation.