Topic Resources

Tools Used
Initiated By
  • ETH Zurich (Department of Management, Technology, and Economics)
Partners
  • Amphiro AG (an ETH Zurich spin-off company)
  • ewz
  • Swiss Federal Office of Energy
  • University of Lausanne (Faculty of Business and Economics)
Results
  • Shower times, energy consumption and water consumption dropped by 20-22%
Landmark Case Study

Shower Feedback in Switzerland

This pilot program provided Swiss households with real-time feedback on one specific, energy-intensive behavior: showering. Participants received smart shower meters that displayed feedback on the individual’s energy and water consumption in the shower in real time. A randomized controlled trial with 697 households in Zurich found that the treatment group who received feedback in real time reduced their energy and water consumption, as well as time spent in the shower by 20-22% over the control group. The effects were stable throughout the two-month study, resulting in average savings of 1.2 kWh per day and household.  This program was designated a Tools of Change Landmark case study in 2016. Listen to the program manager, Verena Tiefenbeck of ETH Zurich, and ask questions during our webinar on March 6, 2017.

Background

Water heating is the second largest residential energy end use and showering accounts for more than 80% of hot water demand.

Few programs focus on hot water conservation in general and shower behavior in particular. This is noteworthy, given a) the large impact it has on household energy use, b) the high level of user influence on the energy consumption, and c) the large variance in shower behavior. This program was designed to promote showering behaviors that reduce energy and water use, by providing engaging non-judgemental feedback.

Programs using feedback (i.e., providing information about one’s own or other people’s behavior) have been rolled out to millions of households over the past years. Among feedback programs, the most widespread approach consists in monthly or quarterly mailed reports that compare household’s utility consumption with similar homes as a frameof reference. This implies that the majority of these interventions are still delivered in the form of paper-based utility bills, thus with a substantial time lag and aggregated over long time periods and to the entire household. These programs generally yield an average conservation impact between 1% and 3% (Ayres et al. (2009); Allcott and Rogers (2014)). While this may seem low at first glance, these programs are highly cost-effective and scalable (1% opt-out rate). While existing paper-based behavioral interventions demonstrate that feedback can cost-effectively influence consumer behavior on a large scale, it has also been shown that feedback works best when it is delivered frequently, timely, clearly, and on specific actions which individuals can easily influence.

The ewz-Amphiro-study was carried out under the lead of researchers of ETH Zurich (Department of Management, Technology, and Economics (D-MTEC)) in close collaboration with ewz, researchers from the University of Lausanne (Faculty of Business and Economics (HEC)) and the ETH Zurich spin-off company Amphiro AG. The Swiss Federal Office of Energy supported the research activities of this study, while ewz funded the study devices.

Getting Informed

Selecting the Behavior

Water heating is the second-largest energy end use in Swiss households, after space heating. As advances in technology, codes and standards have led to more efficient space heating, water heating has increased in relative importance. While the majority of hot water is consumed in the shower, the general public is not aware of the energy dimension of showering.

Overcoming Barriers

A key obstacle to behavior change in this particular area is that individuals had a very poor understanding of how much energy and water they used in the shower. It was technically challenging to deploy measurement equipment in wet-humid environments. Further, not everyone was willing to be reminded of their resource consumption in the shower. Yet we found that the effects are largest among individuals with a high initial level of consumption. In contrast to flow restrictors that reduce comfort and individuals’ freedom of choice by reducing the water flow to a pre-defined level, the feedback intervention allowed individuals to act on the feedback as they deem it necessary - or to ignore it.

Delivering the Program

Customers of a local utility company were invited to participate in this opt-in program. Participants received a smart water meter that displayed real-time feedback on the energy and water use of the ongoing shower. Full details will be posted here after the Landmark case study webinar on March 6, 2016.

Financing the Program

The Swiss Federal Office of Energy supported the research activities of this study, while ewz funded the study devices.

Measuring Achievements

The study collected both detailed survey data (before and after the intervention) and detailed resource consumption measurements in the shower. Altogether, 697 participating households were recruited from a larger sample of 5,000 ewz-customers who had previously completed the ewz Studie Smart Metering. Households were randomly assigned to three experimental conditions, each of which received a different version of the smart shower meter amphiro a1. During the first ten showers (baseline period) all of the shower meters displayed water temperature only. After the baseline period, the three device versions displayed different feedback content in the shower. The devices used by both treatment groups then provided real-time information from the current shower. The devices used by one of the treatment groups also provided information on the past shower. The devices used by the control group displayed only water temperature (to indicate that the device was measuring), but not feedback on resource consumption. 

The meters stored data from every shower taken throughout the two-month study period. At the end of the study, participants were asked to ship their devices back for the data readout and to fill out the final survey. Almost all (685, 98.3%) shipped their devices back and 636 (95.5%) could be read successfully. This resulted in a dataset of 45,664 showers. Nearly as many (666 households, 95.5%) filled out the final survey. In all, 626 households (90%) provided all of the required information.

The first shower taken by each household was excluded from the study, because preliminary analysis of the data found that these first showers deviated significantly from typical patterns for temperature and volume. Probably, a rather large fraction of participants simply turned on the water after the installation of the device to check its functionality and display content, without taking an actual shower.

Regression analysis showed little difference in shower energy use at baseline between the three groups. This indicated that the sample had been successfully randomized.

Outside temperatures remained relatively stable throughout the study period; there was no particular trend upwards or downwards that might explain a drift towards higher or lower water consumption or temperature over time.

The treatment groups’ energy and water consumption was compared to the control group to control for seasonal effects and ensure internal validity of the study. Furthermore, a few survey questions were included to test representativeness of the sample; the responses showed for instance that the sample was not more environmentally friendly than the average population.

All devices were shipped on November 29/30, 2012. The packages contained a return envelope with prepaid postage and shipping address for the readout at the end of the study. Subsequent to the two-month field deployment phase, all study participants were asked by email to return their shower meter and to fill out a final survey (approx. 20 minutes). Participants who had not shipped back their device or not completed the survey received one or two additional reminders in the course of the following weeks. The data readout, device reconfiguration and reshipping procedure were completed in April 2013. Thereafter, the individual datasets were merged, anonymized and analyzed and prepared for dissemination in peer-reviewed journals, international conferences, and local workshops.

Feedback

 ... to come....

Results

Participants barely adjusted the flow rate or the water temperature once they received feedback; instead, they cut their shower time short by an average of roughly 20%. 

The two program / treatment groups did not differ significantly in outcomes. Relative to the control group, they both reduced energy consumption in the shower by 22%, which amounts to yearly savings of 452 kWh for the average 2.1-person household. The savings are equivalent to the energy consumption of two modern European refrigerators. There was no indication of a weakening of the effect size over the two-month study period. 

Regarding a comparison with electricity smart metering studies, participants from the same pool of households had previously taken part in an electricity smart metering study where they reduced their electricity use by only 86 kWh per year. 

The intervention also reduced water consumption by 21%, thus serving a dual purpose in water-stressed regions of the world.

In addition to its direct impact on behavior, real-time consumption feedback also appears to substantially increase knowledge about resource consumption

The survey data indicate that the participants were not more pro-environmental than the national average person. Information is not available on shower energy and water use among the general population (in fact, this study is one of the first ones to quantify that in detail), but the results show that the intervention is particularly effective for participants with a high baseline level.

Cost effectiveness (for details please refer to the working paper “Overcoming salience bias: how real-time feedback fosters resource conservation”): For an individual household who reaps savings both on their energy and on their water bill, the device paid itself off within 9 months. For the case of a large-scale rollout by a utility company, there would be a negative abatement cost of 234 USD per ton of CO2 abated.

Furthermore, profiling, e.g., targeting households with above-average baseline consumption, can double the treatment effect, raising the cost-effectiveness of future deployments even further.

Our fixed-effects model does not indicate any evidence of a weakening of the effect size over the two-month study period. Follow-up studies with longer time spans (3-6 months) in Singapore and in the Netherlands did also not exhibit any decay in the persistence of the treatment during those studies.

This program has already been replicated with additional households across Europe. Impact studies with participants from Switzerland (N=800) and the Netherlands (N=630) found similar effect sizes to those reported for the case study. In addition, a long-term study with 50 households that spanned over 16 months indicated that the effects are also stable in the long run. The impact evaluations showed that the savings were not driven by a small group of environmentalists; the net conservation effect was independent of environmental attitudes. 

Notes

Landmark Designation

The panel that designated this case study consisted of:

  • Arien Korteland from BC Hydro
  • Doug McKenzie-Mohr of McKenzie-Mohr Associates
  • Brian Smith from the Pacific Gas and Electric Company
  • Marsha Walton from the New York State Energy, Research and Development Authority
  • Dan York of ACEEE

Search the Case Studies

Click for Advanced Search »