In this talk we will share experiences from the development, delivery, and assessment of the Data Analysis and Statistical Inference course both on Coursera and on-campus at Duke University, with an emphasis on interactivity. For the MOOC we will discuss the motivation and development of computational data analysis labs in DataCamp, a web-based interactive platform for learning R. This platform was presented to MOOC students as an alternative to following static instructions posted on Coursera to emulate the brick-and-mortar classroom experience where the instructor can help the student identify errors in their code in real time. Having provided both options, and having collected data from both platforms, allows us to investigate questions such as what types of student chooses which approach to learning R, how well do they persist and learn, etc. During this talk we will also present findings from this data analyses that aim to answer these questions. In addition, having click-through data from DataCamp has allowed us to evaluate our learning materials and identify those that might be particularly difficult or confusing for students. We will also discuss findings from this analysis as well as how this information has been used to improve the learning materials for the next iteration of the course. In addition, we will discuss how the materials created for the MOOC (including the videos) have been used to flip the on-campus class and make that course more interactive and engaging for the students as well. We will also present findings on comparisons of student attitude, engagement, and learning before and after flipping the class.