Why Behavioral Science contains the Key to Consumer Investment in Energy

January 9, 2015

By Austin Liu

How much do you actually know about energy data analytics? After being on the Bass Connections Data Analytics team for almost half a year, I guess it’s normal for me to reflect on that question as our team begins to move forward in implementing our analysis techniques on the Duke University Smart Home. Retrospectively, I think I have learned a whole lot– everything from machine learning to electronic devices to behavioral science. However, if you asked the general population about their understanding of the usefulness of disaggregation methods, you probably would gather more than a just a few confused looks.

Imagine you were an expert in energy disaggregation and you went on the street and asked a random individual their opinion on installing a smart meter in their home. A good number of people would probably have no idea of what a smart meter is or what it would be useful for. Now, given that most individuals probably don’t have a strong opinion on using disaggregation data to get helpful energy feedback, how many do you think would change their energy behavior if they saw a stream of numbers modeling their energy habits? Probably not a huge number. This poses a huge problem for engineers and environmentalists all around trying to reduce energy usage. Much wasted energy results from a lack of available information on the effects of energy consumption behavior. In fact, a recent ComEd study [1] gives a great example of how even with energy efficient residential lighting, there is still a huge percentage (11%) of energy wasted simply due to occupant behavior!

We know that the average individual probably does not have very much information on their energy consumption aside from a once-a-month bill. Thus, as the Bass Connections in Energy group begins to research and develop anomaly detection algorithms for residential usage, it is important for us to keep in mind that we must make sure that our results actually change the behavior of our consumers. In recognition of the issue of communicating our data, one branch of our data analytics project is dedicated specifically towards conducting a behavioral experiment on the effects on energy usage in the Keohane 4E dormitory. In the experiment, we examine various aspects of views in energy and the effect of introducing different variables, such as social pressure and neighbor-to-neighbor competition, in making our test subjects change their habits.

This pertains particularly to our project, because it is especially important for us to identify the barriers that the average building owner has in investing in lower energy appliances [2]. It would be challenging to convince facilities, managers, and utility companies to implement our anomaly detection algorithms when their consumers don’t find any value in the tool! Part of the solution in changing user behavior lies in the presentation of our data, specifically to mix various results obtained from our behavioral experiment. As a specific example, if consumers respond better to competition, a helpful presentation of our data would be to compare the energy consumption of a user to that of the average user.

Another perspective that we need to address in moving forward is making any of the systems that we create user friendly, for both consumers and facilities. This is precisely the challenge that faces many engineers of our time! As a problem becomes more and more technical (and disaggregating energy consumption is truly a daunting task), the ability to convey the information clearly so that anyone can understand gets increasingly difficult. Thus, the challenge for the engineer is to create an algorithm that is both specific to the problem and can be easily understood. For the Bass Connections Data Analytics team, it is our goal to make the results of our project not only clear to our stakeholders so that they may find value in our analysis, but also clear to the end-users [3] who may use the analysis to change their own behaviors.

Particularly, our team has thought a lot about this problem, and we have determined a few steps to convey our information clearly and effectively. In creating a product that is user friendly, we aim to:

  1. Make sure to give the user an understanding of relevant properties that may increase energy consumption.
  2. Create a electronically accessible dashboard of energy usage that can be easily understood.
  3. Give feedback to the user related to way to increase energy efficiency that is relevant to their needs.
Now, how to accomplish all of these criteria is a huge task in itself! Hopefully, with the results of our behavioral experiment in the next semester, we will gain a better understanding of individual behaviors and how to properly present our data. Personally, I am quite optimistic that the presentation of our results will result in positive changes in individual energy behavior. However, our team still has a lot to accomplish in the following semester. All in all, in order make our project successful, we must choose the proper techniques and media to transmit our results effectively—whether they be through social norms or other behavioral techniques.
 

Citations:
1. “Behavior-Based Energy Efficiency.” State and Local Energy Efficiency Action (SEE) Network. N.p., n.d. Web. 17 Nov. 2014.
2. Laskey, Alex, and Ogi Kavazovic. “Energy Efficiency through Behavioral Science and Technology.” OPOWER (n.d.): n. pag. Web.
3. Sullivan, Natural Resources Defense Council, Dylan, Carrie Armel, Precourt Energy Efficiency Center, and Annika Todd, Lawrence Berkeley National Laboratory. “When “Not Losing” Is Better Than “Winning:” Using Behavioral Science to Drive Customer Investment in Energy Efficiency.”When “Not Losing” Is Better Than “Winning:” Using Behavioral Science to Drive Customer Investment in Energy Efficiency (n.d.): n. pag. Web.