HarvardX Participation and Completion Rates Vary by Discipline, Study Says

The participation and completion rates of online courses offered by the Harvard and MIT branches of online learning platform edX varied across academic areas, according to a recent report based on two years of course data.

The study reviewed data on HarvardX and MITx, subsets of a virtual learning platform founded jointly by the two universities in 2012. Covering data spanning from fall 2012 to summer 2014, the report analysis included 68 courses, 1.7 million participants, and 1.1 billion logged events, or “clicks.”

According to the study, growth in course participation has been steady, with around 1,300 new participants joining a HarvardX or MITx course per day. The report defined “participants” as registrants who accessed any chapter of course content, according to Andrew D. Ho, chair of the HarvardX research committee and a professor at the Harvard Graduate School of Education.

“There is an increasing number of registrants who never visit the course,” Ho said. By not including all registrants over the two-year period, the study excluded 1.3 million users, he added.

Despite the high number of participants, the level of enrollment engagement varied across academic disciplines. According to the report, computer science courses had almost twice as many participants—around 68,000 per course on average—as humanities, social studies, design, and government courses combined. However, the completion rates in STEM courses, including computer science, were only around half the rates in the humanities and social studies.

The report also found that only about  one in five participants in the average STEM/CS course were female, while humanities and social studies courses had around twice that percentage of females.

There was also a disparity in completion rates, an area that has previously been studied by HarvardX research fellows. On average, the report found that only 5 percent of participants who did not pay a fee for ID verification actually completed their online course. However, for those who did pay to verify their identity—a fee that has ranged from $25 to $250 across courses—around 59 percent gained certification.

Ho said these findings show that the success of online courses should not be measured by completion rates.

“If we want to raise completion rates, we could charge everybody money,” he said. “The low certification rates are a function of openness and accessibility, and I think a myopic focus on completion rates is contrary to the mission of open access to educational resources.”

Isaac L. Chuang, professor of physics and senior associate director of digital learning at MIT, offered a similar perspective, adding that a better measure of the effectiveness of online courses would be how they can change the way classes are taught on campus.

“Let’s ask ourselves in year three—how much of this effort on MOOCs is turning around and being used to...experiment with education on campus,” he said. “The more it succeeds in that dialogue and improves learning on campus, the more I think we would be willing...to say that this was a successful experiment.”

—Staff writer Hannah Smati can be reached at hannah.smati@thecrimson.com. Follow her on Twitter @HannahSmati.