The project uses a longitudinal, comparative, and cross-sectional research design and will generate evidence that is descriptive [observational], associative/correlational [quasi-experimental] and causal [quasi-experimental, statistical modeling, difference in differences, survival analysis]. Original data are being collected on participants in MIT's first MOOC using observation [web logs] and survey research [self-completion questionnaire and semi-structured or informal interviews]. Instruments or measures being used include achievement and activity information, discussion board posting from web use of MOOC; student background data from computer log (e.g., IP address) and background survey (e.g., educational experience); interviews with residential experience students.
We will use a multi-level model to estimate the predictive power of student background and learning resources for achievement and course completion. We will then create a hazard model using student-level demographics and longitudinal behaviors to predict dropout from the course. We will use social network analysis to understand the user groups that connected online, and we will use interview methods and propensity score matching to understand the experience of residential MIT students who took the online course and were offered in-person supplemental activities.