Emotions and Policy

How to cite

Davoine, Eva; Stantcheva, Stefanie; Algan, Yann; Renault, Thomas. 2025. "Emotions and Policy." AEA RCT Registry. AEARCTR-0015432. https://doi.org/10.1257/rct.15432-3.1

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

Eva Davoine* , Stefanie Stantcheva , Yann Algan , Thomas Renault

* Primary investigator

Registry ID
AEARCTR-0015432
DOI
https://doi.org/10.1257/rct.15432-3.1
First registered
Last updated
Status
Completed
Trial dates
2024-11-01 – 2024-11-05
Countries
United States of America ()
Registry record
http://www.socialscienceregistry.org/trials/15432

Abstract

Individual positive or negative emotional states seem to play a critical role in the support or opposition to policies. In this project, we aim to unpack this relationship. We design and implement an experimental survey to understand the impact of positive and negative emotions on people's attitudes towards policy in the United States. We further disentangle the specific roles of distinct negative emotions, notably fear and anger, examining their differential impacts in the context of climate change. Additionally, we investigate whether these emotions have real-world implications by testing whether the emotional tone of a message affects people's willingness to amplify it via retweeting.

Intervention

We conduct three experimental surveys.

In Survey A, respondents are randomly assigned to one of two main treatment groups or a control group, designed to assess the impact of emotions on policy attitudes:

Positive emotion treatment branch: Respondents watch two one-minute relaxation videos (before the trade and wealth \& taxation blocks) designed to induce calm, and the open-ended policy questions in each policy block are framed in a positive way, prompting optimism. The first video features serene landscapes—green fields, a lake, an eagle in flight, a cyclist by the water, lavender fields, and snow-capped mountains—accompanied by peaceful music. The second video showcases tranquil mountain scenery, a blue sky, a flowing river, and a lush green forest from above, also with peaceful music.

Negative emotion treatment branch: Respondents watch two one-minute anger-inducing videos (before the trade and wealth \& taxation blocks) designed to evoke anger, and the open-ended policy questions in each policy block are framed negatively, further prompting anger. The first video is a one-minute video highlighting government corruption and corporate greed as causes of worsening disasters. It focuses on a wildfire ignited by PG\&E’s failure to replace aging electrical line hooks, which resulted in 84 corporate homicide charges, showing the victims and the destruction caused. The second video is a one-minute video highlighting government favoritism toward corporations and the lack of real accountability. It details how PG&E, despite making \$17 billion annually, faced only a \$3.5 million fine—the maximum penalty under California law—for the wildfire caused by its negligence. The video also exposes how the California Public Utility Commission refused to assist in prosecuting PG\&E, withholding evidence that could have strengthened the case against the company.

Control branch: The open-ended policy questions at the beginning of each policy block are framed in a neutral way.

To be able to disentangle the effect of the video from the effect of the framed question, we also added two secondary treatment groups:

Effect of the framed question only: The open-ended policy questions at the beginning of each policy block are framed in a negative way, prompting anger.

Effect of the video only: Respondents watch the same two relaxation videos and the open-ended policy questions are framed neutrally.

In Survey B, respondents are randomly assigned to one of three main treatment groups or a control group, designed to dig further the impact of negative emotions distinguishing fear from anger.

Anger treatment branch: Respondents watch the same two one-minute anger-inducing videos as in Survey A.

Fear treatment branch: Respondents watch two one-minute fear-inducing.

Positive emotion treatment branch: Respondents watch the same two one-minute relaxation videos as in Survey A but are not asked the positively framed open-ended question.

In Survey C, respondents are randomly shown a Tweet with an emotional tone (anger, fear or positive) on a specific policy topic and are asked whether they'd like to retweet it.

Experimental design

We randomize respondents into one of the treatment groups or control group. We then compare the policy views or retweeting behavior of the respondents of the treatment groups to the policy views of the respondents of the control group.

Design details

Randomization method
The randomization is done by the survey software (Qualtrics).
Randomization unit
The unit of randomization is the individual respondent.
Clusters
The planned number of clusters is around 4,000 individuals.
Observations
Approximately 4,000 individuals in each survey.
Arms
Survey A: The sample is divided into 5 arms: 820 respondents receive the main negative treatment (video + negatively framed question), 720 respondents receive the main positive treatment (video + negatively framed question), 820 respondents constitue the control group, 820 respondents receive the secondary treatment (negative question only) and 820 respondents receive the other secondary treatment (video relax + neutrally framed question). Survey B: The sample is divided into 4 arms: 1, 000 respondents receive the anger treatment, 1, 000 respondents receive the fear treatment, 1, 000 respondents receive the positive treatment, 1, 000 respondents are assigned to the control group. Survey C: The sample is divided into 8 arms: 500 respondents see the anger pro-policy Tweet, 500 respondents see the fear pro-policy Tweet, 500 respondents see the positive emotion pro-policy Tweet, 500 respondents see the neutral pro-policy Tweet, 500 respondents see the anger anti-policy Tweet, 500 respondents see the fear anti-policy Tweet, 500 respondents see the positive emotion anti-policy Tweet, 500 respondents see the neutral anti-policy Tweet.
Primary outcomes
The key outcome variables of these experiments are policy views on five topics: trade, immigration, taxes and redistribution, democracy/governance, society issues, climate change and whether people are willing to retweet a message.
Secondary outcomes
We are also interested in the heterogeneity of treatment effects by political affiliation and hence ask respondents about their vote during the 2024 presidential elections. Moreover, we will investigate whether background characteristics of respondents play a role. We will thus look at outcomes by income, age, gender, and education levels.

Cited by

No citation data available for this registration yet.

Keywords