As two primary types of energy consumption, water and electricity have been extensively investigated in the literature on conservation behaviors. But rarely do the researchers compare the intervention effects between water and electricity consumptions. Since the daily activities consuming water and electricity are to a large extent different, their intervention elasticities may differ. With access to the daily records of both the water and electricity usage of the dormitory rooms in a Chinese university, we are able to study both the water and electricity conservation behaviors of the residents. In the experiment, we periodically provide the residents with their past usage of either water and electricity and examine the effects of the different interventions. The experiment setting also allows us to study the spillover effect of pro-environmental interventions. We estimate the direction and magnitude of the spillover effects.
External Link(s)
Citation
Guo, Shiqi and Chengzheng Hang. 2017. "Pro-environmental intervention and its spillover effect: evidence from Chinese dormitories." AEA RCT Registry. November 09. https://doi.org/10.1257/rct.2573-1.0.
In the experiment, the dormitory residents will receive a letter at the beginning of every week, revealing the information on their electricity or water usages during the last week. We vary the information in the letters in different treatments. Overall, we have a 2×2 factorial design. In the treatments of water, only the usage of water is shown in the letter, while in the treatments of electricity, only the usage of electricity is revealed. In the treatments of feedback, we only disclose their own energy usage, while in the treatments of social comparison, we show the usage of themselves and their efficient neighbors. As a result, we have four treatment groups. In Group W-F, the water usage of the same room during the last week is reported in the letter. In Group W-S, the water usage of both the same room and the efficient rooms are reported. In Group E-F, the electricity usage of the same room is reported. In Group E-S, the electricity usage of both the same room and the efficient rooms are reported.
For comparison, we also have two baseline groups. In Group A, we announce to the subjects that the researchers will investigate their daily water and electricity usage records. This is to reveal the experimenter demand effects. In Group C, we make no interventions or announcements.
The observation is on room level. To reduce the spillover effects among treatments, the rooms on the same floor are included in the same treatment. So we cluster the treatment on the floor level. The sample size calculation with the power of 0.8 yields a sample size of about 50 rooms. Thus, to guarantee that we have enough well-functioning dormitory rooms, we randomly select two floors for each treatment and baseline group. At the beginning of each week, we will send letters to every room with the room number on the envelop, showing the relevant information depending on the treatment groups. The experiment will last for about four weeks.
Experimental Design Details
Randomization Method
Randomization done in office by a computer.
Randomization Unit
Randomization is on the building floor level.
Was the treatment clustered?
Yes
Sample size: planned number of clusters
12 building floors (each has a cluster of about 50 rooms)
Sample size: planned number of observations
600 rooms
Sample size (or number of clusters) by treatment arms
2 floors in treatment arm A, 2 floors in treatment arm B, 2 floors in treatment arm C, 2 floors in treatment arm D, 2 floors in treatment arm E, 2 floors in control.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)