The Habit-Forming Effects of Feedback: Evidence from a Large-Scale Field Experiment

Last registered on December 22, 2021

View Trial History

Pre-Trial

Trial Information

General Information

Title

The Habit-Forming Effects of Feedback: Evidence from a Large-Scale Field Experiment

RCT ID

AEARCTR-0008720

Initial registration date

December 19, 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

December 22, 2021, 11:24 AM EST

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

Locations

Country

Australia

Region

Melbourne

Primary Investigator

Name

David Byrne

Affiliation

University of Melbourne

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Leslie Martin

PI Affiliation

University of Melbourne

Contact Investigator

PI Name

Lorenz Goette

PI Affiliation

National University of Singapore

Contact Investigator

Additional Trial Information

Status

Completed

Start date

2017-04-10

End date

2017-12-21

Keywords

Behavior, Environment & Energy

Additional Keywords

Habit formation, Attention, Realtime feedback, Water usage, Randomized Control Trial, Structural Estimation

JEL code(s)

C93, D12, D83, Q25

Secondary IDs

HREC Approval ID 1544989 from University of Melbourne

Prior work

This trial is based on or builds upon one or more prior RCTs.

Abstract

We propose an attention-based mechanism for habit formation and contrast its predictions with those from Stigler and Becker (1977)’s consumption-based mechanism. We develop a habit formation model that nests both mechanisms and structurally estimate it using real-time feedback and consumption data from a large-scale field experiment in shower water consumption. Identification comes from a novel experimental design that varies real-time feedback cycles that temporarily draw attention to individu- als’ water consumption costs. Using the model, we validate an attention-based mechanism, show that attention-based habits form quickly and decay slowly, and uncover a new solution concept for optimizing feedback interventions to create sustained behavioral change.

External Link(s)

SSRN Page

Registration Citation

Citation

Byrne, David, Lorenz Goette and Leslie Martin. 2021. "The Habit-Forming Effects of Feedback: Evidence from a Large-Scale Field Experiment." AEA RCT Registry. December 22. https://doi.org/10.1257/rct.8720-1.0

Sponsors & Partners

Partner

Name

South East Water

Type

government

URL

https://southeastwater.com.au/

Experimental Details

Interventions

Intervention(s)

We explore a new behavioral mechanism for habit that we call attentional habit formation. To this end, we develop a model of attention-based habit formation and design and implement a large-scale field experiment to test the model’s predictions and structurally estimate its parameters. With the estimated model, we examine how attentional habits build up and decay and study the design of policy interventions with personalized feedback for creating robust habits. Throughout, Stigler and Becker (1977)'s model provides a natural benchmark for our analysis. We document stark contrasts between their model and ours in terms of empirical performance and implications for policy intervention design.

Intervention Start Date

2017-05-21

Intervention End Date

2017-10-21

Primary Outcomes

Primary Outcomes (end points)

Individual-level per-shower water usage

Primary Outcomes (explanation)

Directly measured on a per-shower basis from the Amphiro B1 smart meter

Secondary Outcomes

Secondary Outcomes (end points)

Secondary Outcomes (explanation)

Experimental Design

Sample structure
- 7 experimental conditions (T1, T2, T3, T4, T5, T6, T7)
- 100 households per experimental condition
- 130 showers per individual

Experimental Design
"noFB" denotes no real-time feedback on shower water usage from an Amphiro B1 smart shower meter
"FB" denotes real-time feedback on shower water usage from an Amphiro B1 smart shower meter
- T1: 10 noFB, 120 noFB
- T2: 10 noFB, 120 FB
- T3: 10 noFB, 48 FB, 72 noFB
- T4: 10 noFB, 24 FB, 48 noFB, 24 FB, 24 noFB
- T5: 10 noFB, 12 FB, 24 noFB, 12 FB, 24 noFB, 12 FB, 24 noFB, 12 FB
- T6: 10 noFB, 6 FB, 12 noFB, 6 FB, 12 noFB, 6 FB, 12 noFB, 6 FB, 12 noFB, 6 FB, 12 noFB, 6 FB, 12 noFB, 6 FB, 12 noFB
- T7: 10 noFB, 3 FB, 15 noFB, 3 FB, 15 noFB, 3 FB, 15 noFB, 3 FB, 15 noFB, 3 FB, 15 noFB, 3 FB, 15 noFB, 3 FB, 15 noFB

Experimental Design Details

Randomization Method

Random number generator in office by a computer

Randomization Unit

Individual

Was the treatment clustered?

Yes

Experiment Characteristics

Sample size: planned number of clusters

700 individuals

Sample size: planned number of observations

136,500 Details - 700 households, 130 showers per individual in each household - Assume 50% of households have 1 individual, 50% have 2 individuals - Program Amphiro B1s to account for multi-person households with shower noFB/FB sequences described above - Implies 700*130+350*130=136,500 planned observations - Assumes full compliance with trial instructions and no within-trial attrition

Sample size (or number of clusters) by treatment arms

100

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

Powered to detect a 7 L/shower effect of providing real-time feedback based on Tiefenbeck et al. (2018 Management Science) which also uses an Amphiro B1 to implement conditions T1 and T2 (only) described above to estimate effect of real-time feedback on consumption.

Supporting Documents and Materials

IRB