Leveraging digital technologies for fostering resource conservation among the general public
Do people change their behavior when they receive feedback on their energy use in real time? Does the approach work among a broader public... and even in settings where people have zero financial incentives for resource conservation?
The paper in Nature Energy is available here: https://go.nature.com/2qSngnk
After decades of engineering efforts for improving energy efficiency of power plants, engines, building materials, appliances etc., consumer behavior has moved into the focus as key factor influencing energy consumption. As more and more sensors collect data on various domains of our daily lives, it is nowadays possible to monitor how our lifestyle and daily decisions influence our energy consumption – and to make that information transparent to the individual user. Thus, feedback may help people to identify relevant and effective target domains for conservation and enable people to take concrete steps for reducing their energy consumption.
A few years ago, together with colleagues, we have been able to show that real-time feedback on a specific, energy-intensive behavior can indeed trigger substantial behavior change and large and meaningful resource conservation effects (Tiefenbeck et al. 2018). Yet, almost every time we presented the results of that experiment, people in the audience asked the same two valid questions: A) What role do financial savings play for the conservation efforts (or: Why does the smart shower meter not display monetary savings…do people really want to save energy or is it all about the money)? and B) How representative was your sample (or sometimes a bit less politely: Maybe your participants were just a weird bunch of environmentalists… or technology enthusiasts… or both)?
After debating several ideas, we identified hotels as the perfect setting for tackling both questions: Guests have no financial incentive for conserving resources during their stay at the hotel (room rates are fixed and if you pay $200 or so per night, a hot shower should be included, after all…). In addition, we would be able to randomly install the measurement and feedback devices ourselves in dozens of similar showers. Even better, thousands of ordinary hotel guests would use those showers, simply encountering the device as integral part of their room’s bathroom. To run the experiment, we joined forces with Amphiro AG, a startup company manufacturing the smart shower meters. For them, hotels were a promising new sales channel for their products, so they reached out to several hotels to implement the study. It turned out that many hotel managers were very open to the idea. Overall, the deployment, field phase, and data collection went very smooth (a big “Thank You!” goes to Susanne and Tom for their great work and efforts).
One aspect that really surprised us in the dataset was that hotel guest who did not receive feedback (i.e., the control group) used LESS energy (and water) per shower than the participants in the household samples we had studied. The technical infrastructure (low-flow showerheads in particular) probably plays a role here – flow rates (liters/minute) in the participating hotels were on average lower than in the households we had previously studied. Consequently, if we control for the lower consumption at the hotels in the absence of feedback, the savings effects among the hotel guests and among the volunteer household sample are in fact comparably large.
In the first version of the article submitted to Nature Energy, we had put the conservation effects in the absence of monetary savings in the center of the story; the aspects related to volunteer selection bias played a minor role as a nice feature of the study setting. All three reviewers had various concerns related to this framing of the article - in fact, we have rarely received reviewer reports that were so well aligned. Once we had digested those points and read the related work recommended by one reviewer, we came to realize that we should flip the framing of the article upside down, as the experimental setting eliminating volunteer selection bias was in fact the more relevant and more straightforward contribution that we should focus on. Thanks to the helpful reviews the revision gave the article a completely different spin, which we are very grateful for.