A quantifiable measure of classroom engagement using skin biosensors

By Michael W. Cashman

In the second whole-group training session, my STEP-WISE colleagues and I learned about active learning techniques, their benefits, and why we should incorporate active pedagogy in the biology seminars we are designing and teaching. Based on personal experience as a student, participating in a flipped classroom learning environment that relies on just-in-time pedagogy works! So, I was surprised to learn that reformed-based teaching and learning practices are only gaining traction among academics responsible for teaching undergraduate STEM courses. The flipped classroom actually works for lots of people: a steadily increasing amount of data support active learning techniques and the positive impacts they can have on student learning.

Enter a group of researchers from Auburn University determined to generate data providing insight about what impact reformed-based teaching and learning practices have on student learning outcomes. Their article, “Biosensors show promise as a measure of student engagement in a large introductory biology course,” published in CBE—Life Sciences Education journal (December 2020) explains a masterfully crafted, meticulously designed, and utterly creative test.

Historically, data capturing classroom engagement were based on observations of student behaviors or from pre- and post-assessment surveys. This study uses biosensor data captured from physiological changes on the skin as a proxy for classroom engagement. Skin biosensors use galvanic skin response (GSR) technology to measure unbiased physiologic changes due to external stimuli. GSR reads electrodermal activity (EDA), which is a measure of subtle changes in the amount of perspiration or moisture level on the skin controlled by sweat gland activity. Stimulation of the cutaneous sweat glands implies both psychological and physiological arousal; therefore, skin conductance captured by biosensors using GSR is a useful proxy for engagement. With that said, one minor critique in their methodology relates to the effort to demonstrate that ambient temperature and user movement did not influence skin-conductance data. It would technically be more accurate (and perhaps strengthen their claim), however, to comment on the classroom’s relative humidity as opposed to temperature. Because EDA measures volumetric sweat changes on the skin’s surface, ambient humidity will directly affect its evaporation rate thereby interfering with accurate biosensor data.1

This mixed-methods case study used qualitative and quantitative measures to investigate how teaching approach impacts engagement in an undergraduate introductory biology classroom as measured by skin biosensors, comparing traditional lecture and just-in-time teaching pedagogy in a flipped classroom model. A flipped classroom can involve viewing a recorded lecture for homework followed by brief assessments and exercises to have students reflect on what they understood, what questions they still have on the information presented, and what additional information they would like to know. This outside homework would be used to fuel the in-class group work activities that facilitates thoughtful conversations and encourages group discussions.

Results showed that the more time students spent listening, the less change they had in skin conductance. However, the more time they spent working in groups, the higher the change in skin conductance. In other words, students become less engaged the more time they spent listening and more engaged the more time they spent working in groups. Comparing the average skin conductance measurements in students between the two teaching paradigms did not reach statistical significance; however, the average self-reported engagement (Likert scale 1-5) was statistically significant. The authors argue this is likely due to a limitation of their study by not having enough biosensors, allowing new students to wear the sensors each time they collected the biosensor data most likely contributing to the wide standard deviation seen in the skin conductance measurements. They then correlated skin conductance measurements to traditional observed behaviors of student engagement during times students spent listening and during times students worked in groups, which were both statistically significant.

Interestingly, this study also examined the instructor’s level of engagement. When the instructor lectured, her skin conductance readings stayed constant and relatively high, a reflection of “having to be ‘on’ all the time,” and her skin conductance varied more when she spent her time guiding learning with active techniques.

So, active learning once again seems to promote student engagement. And, although the authors acknowledge that their small sample size demands reproducibility and verification of the data, we may now have a noninvasive instrument that potentially provides instructors with unbiased, objective, and insightful data measuring classroom engagement.


1. Dougherty E. Ask an engineer: why do we sweat more in high humidity? MIT Engineering. Published October 11, 2011. https://engineering.mit.edu/engage/ask-an-engineer/why-do-we-sweat-more-in-high-humidity