Analysis of Student Interactions with Browser-Based Interactive Simulations

We have developed open-sourced interactive browser-based simulations that model realistic core engineering systems. Our simulations use JavaScript and HTML-5 to insure that the code is platform-agnostic and functional on all devices with a modern browser, avoiding some of the dissemination hurdles with educational Java applets or mobile apps. For each use of the simulations, we track student mouse movements and clicks, keyboard events, event times, screencast use, correlation with hands-on design project success, and more, leading to a large database that may be mined for pedagogical insights. We have had remarkable success using these simulations while coupling them to collaborative, open-ended, hands-on design projects within the setting of a freshman design laboratory. In this course, students individually conduct experiments with the simulations before they come together as teams to design and build a process or product that relies on related core engineering theory. Pre- and post-course surveys and tests were used to assess the teaching potential and students’ evaluation of the simulations as course materials. Resulting student evaluations are far more positive than those found in a comparable engineering course using traditional pedagogy and static text-book assignments. Student learning was demonstrably improved along with student confidence in a variety of engineering skills. Our findings suggest that the simulations facilitate hands-on active and collaborative learning earlier in our students’ academic career by making complicated engineering theory more accessible. The resulting database of simulation usage data has been effective in detecting and responding to usage patterns of successful and unsuccessful students, allowing for iterative development of educational material. For example, ensemble averages of mouse location for successful and unsuccessful attempts in a spectrophotometer simulation revealed that unsuccessful students did not understand the need to properly calibrate. Student study habits and problem solving strategies also are evident in such data. Finally, we have found usage tracking data to be effective in improving user experience; for example, we detected attempts to interact with non-interactive elements of the simulation, prompting us to add interactive functionality to these elements. By collecting real-time data on how students complete their homework, including both correct and incorrect attempts, we are able to both refocus our in-class discussions to address quantified weaknesses and add automated instructional supports in simulations to address errors at the moment they are detected. We believe, using such data, we will be able to bring some of the benefits of in-person active and collaborative learning to online simulations.