Environmental self-identity and the impact of social information on energy use: experimental evidence

How to cite

Bonan, Jacopo; Tavoni, Massimo; Cattaneo, Cristina; D'Adda, Giovanna. 2018. "Environmental self-identity and the impact of social information on energy use: experimental evidence." AEA RCT Registry. AEARCTR-0002699. https://doi.org/10.1257/rct.2699-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Jacopo Bonan* , Massimo Tavoni , Cristina Cattaneo , Giovanna D'Adda

* Primary investigator

Registry ID
AEARCTR-0002699
DOI
https://doi.org/10.1257/rct.2699-1.0
First registered
Last updated
Status
On going
Trial dates
2016-07-01 – 2018-04-30
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/2699

Abstract

We document a field experiment on the impact of a social information program on energy consumption and digital engagement, conducted with about 450,000 customers of a large Italian electric utility. The program consists in the delivery of a bi-monthly Home Energy Report (eHER), via email, presenting a comparison of one's own previous month consumption with that of similar efficient homes nearby. A random sample is assigned to a control group, receiving nothing.
The main focus of the present project is the question of a possible mechanism behind the effect of the program. Environmental values are strong predictors of environmental behavior, because people who strongly endorse pro- environmental values are more likely to engage in various pro-environmental behaviors. While values tend to be stable in everyone’s life, environmental values are correlated with environmental self-identity, which, on the contrary, can be primed (van der Werff 2013). Therefore, we designed a treatment message aimed at priming environmental self-identity. The treatment message is included in the November/December eHER. Comparison of outcomes for customers exposed to the environmental self-identity priming relative to control subjects, will reveal the role of environmental values on the program impact.

Experimental design

The project is centered around a customer engagement and energy efficiency program conducted with utility's customers, which kicked-off in July 2016. The program targets roughly 450,000 existing customers. The main goal of the program is to increase loyalty, digitalization and engagement of customers. Energy efficiency goals are secondary. The intervention follows the typical design of Opower programs, already described and evaluated by several papers. The program consists in the delivery of a bi-monthly Home Energy Report (eHER), via email, presenting a comparison of one's own previous month consumption with that of similar efficient homes nearby. A random sample is assigned to a control group, receiving nothing.

The main focus of the present project is the question of a possible mechanism behind the effect of the program. Environmental values are strong predictors of environmental behavior, because people who strongly endorse pro- environmental values are more likely to engage in various pro-environmental behaviors. While values tend to be stable in everyone’s life, environmental values are correlated with environmental self-identity, which, on the contrary, can be primed (van der Werff 2013). Therefore, we designed a treatment message aimed at priming environmental self-identity. The treatment message is included in the November/December eHER. To identify the most effective messages to prime environmental identity to include in the RCT, we pre-tested a battery of different messages through an online experiment conducted on Prolific Academic. Subjects in the experiment were exposed to one of different messages, priming environmental self-identity, and then were asked to donate to an environmental NGO and to read a set of energy saving tips. The pre-test also included manipulation checks for the primes, so as to make sure that their impact on pro-environmental behavior worked through their effect on environmental self-identity. The most effective message, in terms of both donation levels and perceived self-identity measures in the pre-test, was one that reminded individuals of the ways in which they already saved energy in their daily lives, boosting their self-image as pro-environmental people. We thus selected this message to be included in the eHER marketing module, followed by a request to find more ways to save energy by consulting the energy saving tips contained in the program portal. The control message simply asks customers to find ways to save energy by reading the tips. In the analysis, we will compare engagement with the eHER and energy saving tips, and energy consumption, between customers exposed to the eHER enriched by the environmental self-identity message, customers exposed to the standard eHER, and control customers receiving no eHER.

The assignment of customers in the program to the group receiving the eHER augmented with the environmental identity prime and the control message was done through a randomization algorithm (minmax t-statistic) which matches customers on the basis of baseline consumption and geographic location (Bruhn and McKenzie, 2009). The sample of customers still enrolled in the program and who actually received the eHER containing the experimental messages is 278,252 customers (according to the data we currently dispose). About half of the group (49.9%) received the environmental priming message and the remaining 50,1% received the control one.

Design details

Randomization method
The assignment of customers in the program to the group receiving the eHER augmented with the environmental identity prime and the control message was done through a randomization algorithm (minmax t-statistic) which matches customers on the basis of baseline consumption and geographic location (Bruhn and McKenzie, 2009).
Randomization unit
Randomization is at the individual level
Clusters
No clusters are involved
Observations
The experiment involves a total sample size of 278,252 customers still enrolled in the program by the time of the experiment and actually receiving the eHER in the November-December 2017 wave. For the manipulation check (see pre-analysis plan for more details) we use the second wave of survey collected in December 2017- January 2018. We expect to get about 1000 observations.
Arms
For the analysis of outcomes such as electricity consumption and digital engagement, we use the sample of customers enrolled in the program who received either the treatment (N=138,829) or the control message (N=139,423) incorporated in the eHER. For the manipulation check (see pre-analysis plan for more details) we use the second wave of survey collected in December 2017- January 2018. We expect to get similar shares of respondents to the survey from the treated and control group.
Primary outcomes
The main outcomes are individual electricity consumption and digital engagment (see PAP for more details)

Cited by

No citation data available for this registration yet.

Keywords