The impact of personalized information on individual decisions to mitigate forever chemicals: Evidence from an online randomized evaluation

How to cite

Refulio-Coronado, Sonia; Uchida, Emi; Messer, Kent; Noblet, Caroline. 2025. "The impact of personalized information on individual decisions to mitigate forever chemicals: Evidence from an online randomized evaluation." AEA RCT Registry. AEARCTR-0014470. https://doi.org/10.1257/rct.14470-1.1

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

Sonia Refulio-Coronado* , Emi Uchida , Kent Messer , Caroline Noblet

* Primary investigator

Registry ID
AEARCTR-0014470
DOI
https://doi.org/10.1257/rct.14470-1.1
First registered
Last updated
Status
Completed
Trial dates
2024-09-09 – 2025-03-31
Countries
United States of America (), United States of America (Rhode Island), United States of America (Delaware), United States of America (Maine)
JEL codes
Q25, Q52, Q53
Registry record
http://www.socialscienceregistry.org/trials/14470

Abstract

Adopting technologies to mitigate pollution exposure, such as water filters certified to remove PFAS, requires individuals to overcome significant knowledge and decision-making barriers. These include search costs, misinformation, and the challenge of connecting current adoption decisions with uncertain future health benefits. In behavioral science, information programs tailored to specific consumers have been effective across multiple domains. These programs intend to shape individual behavior in a manner considered welfare-improving from the perspective of society and the individual. However, to date, tailored messaging has not been tested for consumers’ decisions in pollution aversion behavior.

This study uses an online randomized experiment to evaluate the impact of personalized information on consumer adoption of water filters certified to remove PFAS. Participants are randomly assigned to one of three groups: a control group with no information, treatment group 1 receiving generic information about PFAS, or treatment group 2 receiving both generic and personalized information tailored to household composition and medical history. The survey assesses participants' water usage behaviors, PFAS knowledge, and willingness to purchase a certified water filter, with follow-up data collected to measure changes over time.

The experiment plans to recruit 1,500 participants from a nationally representative sample, supplemented by 350 participants from Rhode Island, Delaware, and Maine, and a pretest of 300 participants. Results will provide insights into the role of tailored messaging in reducing knowledge barriers and encouraging the adoption of health-protective technologies. This research offers evidence to inform policies promoting equitable engagement in pollution aversion behaviors.

Intervention

This study conducts an information intervention using an online survey to measure the causal impact of personalized information on people's averting behavior, measured as the individuals' water filter uptake to protect themselves against PFAS pollution of drinking water at home. PFAS information changes from no information for the control group to generic information for treatment 1 and personalized information for treatment 2. Following this intervention, the survey introduces three PFAS-removing water filters already available in the market. It asks participants about their interest in purchasing a certified water filter on a scale of 1 to 10. After that, the survey requests them to upload proof of purchase if they bought a PFAS-removing water filter.

The experiment is conducted in two rounds, and participants are compensated after they complete each survey.

Experimental design

We construct demographic groups based on whether there is a pregnant woman, children at or under the age of 13, and individuals suffering from PFAS-related diseases in the household. Individuals in each demographic group will be randomly assigned to the control or treatment groups in equal proportions.

Design details

Randomization method
We set up the randomization using the survey flow in Qualtrics.
Randomization unit
This study randomly allocates participants to the control and treatment groups within each of the eight demographic categories based on the presence of a pregnant woman, children aged 13 or under, and individuals in the household suffering from PFAS-related diseases.
Clusters
Eight
Observations
Pretest: 300 participants. Nationally representative sample:1,500 participants. Additional samples: 350 participants.
Arms
Since we lack prior information on the number of individuals living in households with children aged 13 or younger, pregnant women, or people with PFAS-related diseases, we cannot pre-estimate the distribution of participants across the control and treatment groups. However, the survey collects data on these characteristics to categorize participants into eight demographic groups. These groups will then be distributed equally and randomly among the control and treatment groups.
Primary outcomes
Submitting proof of purchase.
Secondary outcomes
Level of interest in purchasing a water filter and willingness to pay (WTP) for a water filter certified to remove PFAS.

Cited by

No citation data available for this registration yet.

Keywords