Avoiding Negative Effects of Negative Information

How to cite

Delecourt, Solene; Fong, Jessica. 2019. "Avoiding Negative Effects of Negative Information." AEA RCT Registry. AEARCTR-0004609. https://doi.org/10.1257/rct.4609-1.0

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

Solene Delecourt* , Jessica Fong

* Primary investigator

Registry ID
AEARCTR-0004609
DOI
https://doi.org/10.1257/rct.4609-1.0
First registered
Last updated
Status
In development
Trial dates
2019-09-01 – 2022-02-15
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/4609

Abstract

More information should always be better, but it can have negative consequences, especially when the information is negative. For example, bad news can be demotivating. Relatedly, people may avoid freely available information, especially if it contains bad news. In this paper, we ask: How can we mitigate the negative effects of bad news? We test ways to mitigate potential negative effects of negative information using a randomized field experiment. We collaborate with a large company selling wearable activity trackers that provide users with personal information on the duration and quality of their sleep. We conduct an experiment with 1 million users for 4 weeks that randomizes the content of messages that goes along with information when the quality of their sleep has dropped. The types of messaging we test not only measures the most effective content at preventing negative consequences of bad news, but also information avoidance and boosting effort/motivation. They also provide insight into the mechanisms of how types of information can affect behavior. Therefore, we can apply these types of messaging to improve the deliverance of other information in general. It has clear implications on how to help firms and other organizations deliver crucial negative information to their users in a more effective manner.

Intervention

We conduct an experiment with 1 million users for 4 weeks that randomizes the content of messages that goes along with information when the quality of their sleep has dropped.

Experimental design

We conduct an experiment with 1 million users for 4 weeks that randomizes the content of messages that goes along with information when the quality of their sleep has dropped.

Design details

Randomization method
Randomization done in office by a computer.
Randomization unit
Individual level.
Clusters
NA
Observations
1 million active users.
Arms
We have 1 control group and 5 treatment groups, so that is about ~165,000 individuals per group.
Primary outcomes
Effect on engagement: are users looking at sleep data the next day? Did the user wear their device to bed the next day? Effect on motivation: Are users following (reinforcing) good sleep habits? (bedtime, sleep duration, wake up time) Effects on other health behaviors the next day: exercise (steps, active min)
Secondary outcomes
We plan to look at heterogeneity of the treatment effect by demographics, gender, age, baseline activity level (step count and active minutes), baseline sleep quality (prior to the experiment).

Cited by

No citation data available for this registration yet.

Keywords