This is a RAPID to study Massive Open Online Courses as to how and in what context learning best occurs. This is a collaboration between MIT and Harvard on their edX initiative. The course studied is the MIT Circuits course "Circuits and Electronics" first offered in the Spring of 2012 as a MOOC. Data was collected from the 154,763 students enrolled in the course and this award will fund analyses of that data and additional data on students that needs to be collected quickly before too much attrition takes place. Additionally, the learning outcomes and experiences of students in the classroom version of the course will be compared to the outcomes and experiences of students in the MOOC to best determine both the effectiveness of MOOCs and the contexts in which MOOCs work well.
MOOCs are a recent phenomenon and MOOCs have not been studied as to their effectiveness in a learning environment. This study will answer many such questions and provide guidance for improvement in MOOCs and in their deployment.
Massive open online courses (MOOCs) have garnered a tremendous amount of publicity since first launched in 2011. MOOCs allow tens of thousands of students to take one course together with lectures, assignments, and exams all online. This project analyzed data from "Circuits and Electronics" (6.002x), the first MOOC developed by edX. The course started with almost 155,000 registrants and just over 7,100 received a certificate of completion. Goals of the Project This research was undertaken to analyze three kinds of data from 6.002x: (1) data logs that captured 230 million interactions students had with the course; (2) over 90,000 posts on a discussion forum; and (3) an end-of-course survey to which over 7,000 students responded. The objectives of the study were to explore the methodologies that could be used to analyze the huge amount of data 6.002x created, and to identify both factors in the students’ backgrounds and their use of instructional materials that contributed to their persistence and success in the course. Findings We were able to draw a fairly detailed picture of the 6.002x students. They came from 194 countries with the U.S., India, and the U.K the top three. Two-thirds who initially registered for 6.002x reported English as their first language. The percentage of students who completed 6.002x was under 5%; most students who registered left after the first week. Although 6.002x students ranged from teenagers to in their seventies, most students who answered a question about age on the end-of-course survey were in their 20s or 30s. (This statistic and the statistics below are based on a smaller number of responses than the total for the survey as not every student received every question.) Not surprisingly, 88% of the students were male. 6.002x students were highly educated: 37% had a bachelor’s degree, 28% had a master’s or professional degree, and 27% completed high school. Over three-quarters of the respondents had a strong background in calculus. Just over half reported their reason for enrolling was for the "knowledge and skills gained from the course"; just over a quarter enrolled for the "personal challenge." After identifying basic characteristics of the students, our next step was to carry out more sophisticated analyses. First, we looked for relationships between students’ background and their achievement in the course, defined as earning a certificate. Using this definition, we found no relationship between achievement and age, gender, reason for enrolling, or home background. Rather, the strongest positive correlation for achievement was whether students reported they worked offline with another person—either another student in 6.002x (18%) or "someone who teaches or has expertise in the area" (3%). The second strongest predicator of achievement was a background in calculus. When we explored the patterns between uses of course components and achievement and persistence both for all students and for certificate earners, the findings were particularly interesting but puzzling. Some resources were consistent in helping students persist and achieve in 6.002x, but others were not. For example, for all students, time spent with homework and on the discussion forum was consistently positive. However, the correlation with time spent with the e-textbook was negatively predictive of scores for both populations. When we looked at time spent on labs or on the problems embedded in the lecture videos, the correlation with achievement was reversed for the two populations. For all students, the hours on lab assignments were predictive of a higher score, but for certificate earners, the correlation was negative. We found homework and labs most useful for all users, but the embedded lecture problems and homework were most helpful for certificate earners. We also see the effects of resources are different if we look at "achievement" versus "persistence." For example, time spent with the e-textbook was a negative predicator for achievement, but it predicted a slightly lower probability of stopping out. We believe these results, while not generalizable to all MOOCs, are "tantalizing" and could point the way for more experimentation. From our work with the discussion forum data, we know only a small number of students posted. Certificate earners participated more in the discussion forum than other students, and we saw a mild, but significant, correlation between the number of posts per student and total scores. In classifying the posts, we saw approximately one-third could be categorized as "social-affective." Next Steps We hope to access data from other edX MOOCs to see if the findings from this study hold true. We are also using MOOC data to partner with MIT instructors to improve blended learning, which combines online resources with activities in the classroom. We believe online learning holds great potential both to meet the challenges of accessibility and affordability in higher education and to strengthen teaching and learning whether online or on campus.
