Mechanisms Driving Social Comparison Nudges

Last registered on May 26, 2021

View Trial History

Pre-Trial

Trial Information

General Information

Title

Mechanisms Driving Social Comparison Nudges

RCT ID

AEARCTR-0007726

Initial registration date

May 26, 2021

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published

May 26, 2021, 10:54 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Country

United States of America

Region

New York

Primary Investigator

Name

Erica Myers

Affiliation

University of Calgary

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Maya Papineau

PI Affiliation

Carleton University

Contact Investigator

PI Name

Nicholas Rivers

PI Affiliation

University of Ottawa

Contact Investigator

Additional Trial Information

Status

On going

Start date

2020-11-15

End date

2021-11-15

Keywords

Environment & Energy

Additional Keywords

Home Energy Reports, Behavioral Nudges, Social Norms, Moral Suasion

JEL code(s)

D90

Secondary IDs

C93, L94, Q41

Abstract

We explore mechanisms driving the effectiveness of the behavioral nudge known as Home Energy Reports (HERs). HERs typically provide information about a household’s own energy usage, how that compares with neighbors’ usage, and estimated monetary savings from several suggested conservation actions. A household’s energy consumption is a function of both their behavioral choices, such as thermostat settings and appliance usage and the characteristics of their home that affect the rate at which it retains heat. Therefore, a typical HER nudge is a signal about a combination of household behavior and the structural efficiency of the home. The purpose of this study is to disentangle the individual effects of each of the two signals. We will estimate the effects of a signal based either solely on the physical characteristics of the home and a signal based solely on household behavior. Further, we will determine home or household characteristics that drive treatment effect heterogeneity for each type of signal, which could be used for targeting purposes.

We take advantage of a new technology that assesses the structural efficiency of a home based on images of heat loss taken from an aircraft-mounted infrared sensor during the winter heating season. The technology measures the heat flow from different parts of the roof, comparing the flow of heat coming through the furnace vent to other components of the roof. Importantly, since the measure comes from a comparison across the roof of a single home, it is driven solely by the building envelope and not by household choices about thermostat settings.

External Link(s)

Registration Citation

Citation

Myers, Erica, Maya Papineau and Nicholas Rivers. 2021. "Mechanisms Driving Social Comparison Nudges." AEA RCT Registry. May 26. https://doi.org/10.1257/rct.7726-1.0

Sponsors & Partners

Experimental Details

Interventions

Intervention(s)

Home Energy Reports: A widely-used behavioral intervention that reduces energy consumption by repeatedly mailing social comparison-based home energy reports (HERs) to household

Intervention Start Date

2020-11-15

Intervention End Date

2021-11-15

Primary Outcomes

Primary Outcomes (end points)

Energy Consumption, Participation in utility sponsored energy efficiency programs

Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)

home or household characteristics that drive heterogeneity in either of the primary outcomes

Secondary Outcomes (explanation)

Experimental Design

We take advantage of a new technology that assesses the structural efficiency of a home based on images of heat loss taken from an aircraft-mounted infrared sensor during the winter heating season. The technology measures the heat flow from different parts of the roof, comparing the flow of heat coming through the furnace vent to other components of the roof. Importantly, since the measure comes from a comparison across the roof of a single home, it is driven solely by the building envelope and not by household choices about thermostat settings. We use deciles of these raw, propriety, measures of efficiency to create a 1 through 10 scale of heat loss, which we use in our study.

To create a 1 through 10 scale of energy consumption based on behavior, we use deciles of energy consumption conditional on house vintage (year built), square footage, and the structural heat loss rating described above. The proposed impact evaluation seeks to answer 4 research questions.

In each of two treatment arms, consumers receive a Home Energy Report comparing either: 1) their heat loss score or 2) their behavioral score to their neighbors. Along with the comparison, consumers also receive information about energy efficiency investment and conservation opportunities. We deliver this to some customers through e-mail and others through the mail.

Experimental Design Details

Randomization Method

The groups were created by randomly selecting individual households into two groups each containing 33.8 percent of the sample (forming treatments 1 and 2), and one group containing 32.4% of the sample for the control group. The sample was randomly selected within each billing cycle using this approach.

We used Stata’s iebaltab command to calculate balance statistics, using the following variables: average household consumption over the previous 8 months (February to September 2020), MyHEAT heat rating, municipal assessed value, year built, home square footage, bill cycle, property class (i.e. home type), receipt of an e-bill and paper bill, receipt of e-bill exclusively, receipt of paper bill exclusively. We repeated the randomization 40 times and selected the sample with the maximum-minimum p-value among for the t-test of the difference in means between each treatment and the contro

Randomization Unit

Household

Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters

60129 Households in owner-occupied homes

Sample size: planned number of observations

For each household, we will have billing and investment data before and after treatment

Sample size (or number of clusters) by treatment arms

treatment groups 1 and 2 with 20,598 and 20,608 observations, respectively, and a control group with 18,923 observations.

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

Using pre-trial billing data, we have simulated that roughly 20,000 treatment and 20,000 control households will be needed to identify an effect size of 1.5 percent

Supporting Documents and Materials

IRB