Data Analytics for Interactive Virtual Laboratories
Previously we have reported on the development of a set of Interactive Virtual Laboratories (IVLs) that help students understand key concepts in thermodynamics. These laboratories guide students through approximately twenty frames using an inquiry-based “predict, observe, explain” pedagogy. Use of IVLs in class provides a copious amount of data from every student participant. This paper reports initial processes and strategies to make student thinking and learning visible to instructors and researchers through the use of data analytics. One outcome from this work is an automatic grading system to provide instructors detailed assessment data from the IVLs.
We analyze data from the implementation of three IVLs in a junior level thermodynamics course at a public university. Two hundred and forty seven students majoring in chemical, biological, or environmental engineering participated in the study. Amongst the participants, there were 12,350 total answers collected. Data gathered are used to develop an algorithm to help assess student responses, and this capability will be integrated into a web-based tool and available to faculty. As part of the process, each question within the IIVLs is coded in relation to student engagement. Questions are coded as procedural, conceptual, procedural, and reflection. Different question types are weighed differently in the assessment algorithm. For example, a question coded as “procedural” might involve interpreting a numerical value from a graph; while a question coded as “prediction” requires students to recall previous problems and predict how their answers will change due to a change in the system. Thus, a question coded as “prediction” might be worth more than a question coded as “procedural.” Using cluster analysis, we explore relations between how sets of students answer different type of questions. This analysis can provide information on what type of thinking the student is struggling with, and is the first step towards automated adaptive instruction.
Finally, we have audio-recorded 17 students as they completed the IVLs. The audio data provide a connection between reasoning employed by students and their submitted answers. These data will be used to confirm the coding scheme and verify the analytics’ prediction of where the students struggle.
Are you a researcher? Would you like to cite this paper?
Visit the ASEE document repository at
for more tools and easy citations.