Understanding Uncertainty

How to cite

Mcmahon, Michael; Rickards, Peter; Rholes, Ryan; Naylor, Matthew. 2025. "Understanding Uncertainty." AEA RCT Registry. AEARCTR-0016245. https://doi.org/10.1257/rct.16245-2.0

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

Michael Mcmahon* , Peter Rickards , Ryan Rholes , Matthew Naylor

* Primary investigator

Registry ID
AEARCTR-0016245
DOI
https://doi.org/10.1257/rct.16245-2.0
First registered
Last updated
Status
On going
Trial dates
2025-06-16 – 2025-12-31
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/16245

Abstract

We are interested in the effect that communicating uncertainty has on the public's understanding of key messages, as well as how they use those messages going forward. Do they believe these messages more or less when uncertainty is communicated, and does communicating that uncertainty affect credibility in the future in the face of data outturns that do not align with forecasts.

Intervention

We first provide to participants varying types of uncertainty communication mediums in the form of a survey. We will identify if there are considerable differences in how the public understands the uncertainty being communicated across these mediums. We run a survey of the public and then one of experts all anonymously recruited.

Next, we will run an experiment where we ask people to engage in a forecasting experiment where they are provided different types of communication about the future, some involving uncertainty and some not. We will compare how people use the information in the different treatments, as well as how credibility is influenced.

Experimental design

In the survey, we provide different charts and images to participants and ask a slew of questions about understanding. This gives us some summary statistics and baseline data around which types of medium convey uncertainty.

In the experiment, we randomly allocate participants to one of many treatments first. They will continue through this treatment and then will be allocated to one other treatment afterwards. We are therefore using a between-groups identification method although some participants will see two types of treatment to increase the effective sample size.

Design details

Randomization method
Randomisation done by a computer
Randomization unit
Individual
Clusters
No clusters
Observations
1800 individuals
Arms
400-control, 400-treatment1, 600-treatment2 400-treatment3 Within treatments there are 4 variations so 100 per variation within treatment arm.
Primary outcomes
In the survey, self-reported understanding, uncertainty and objective measures of median values and uncertainty ranges to be taken from data in the chart. In the experiment, we will use the variable that they are forecasting to assess (relative to a Bayesian benchmark and the control group) how they use information to update their incentivised point estimate beliefs and their own incentivised uncertainty beliefs.

Cited by

No citation data available for this registration yet.

Keywords