Uncertainty and Rounding in Expectation Surveys

How to cite

Dovern, Jonas. 2026. "Uncertainty and Rounding in Expectation Surveys." AEA RCT Registry. AEARCTR-0017646. https://doi.org/10.1257/rct.17646-1.1

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

Jonas Dovern*

* Primary investigator

Registry ID
AEARCTR-0017646
DOI
https://doi.org/10.1257/rct.17646-1.1
First registered
Last updated
Status
Completed
Trial dates
2026-01-14 – 2026-02-28
Countries
Germany ()
Registry record
http://www.socialscienceregistry.org/trials/17646

Abstract

Evidence from previous research on survey-based inflation expectations of private households suggests that rounded point forecasts signal heightened subjective expectation uncertainty. This association is typically established through simple correlations between rounding indicators and measures of uncertainty derived from probabilistic expectations. What is missing from the literature is a causal investigation of the effects of changes in subjective uncertainty on rounding behavior. This project aims to fill this gap by conducting a randomized controlled trial in which respondents' subjective uncertainty is exogenously shifted through the provision of external information about the range of inflation or experts' inflation forecasts. We then analyze changes in the frequency and intensity of rounding in respondents' point forecasts caused by these shifts in expectation uncertainty.

Intervention

In the RCT, we vary the information about US inflation that survey participants see during participation.

Experimental design

We provide the treatment groups with the information treatment after we have elicited prior expectation and expectation uncertainty.

Following the treatment, we elicit some socioeconomic information and then the posterior expectation and expectation uncertainty.

The brief survey concludes with a question about the reasons for providing rounded inflation expectations that is shown to all participants that stated integer inflation expectations.

Design details

Randomization method
Randomization is implemented via random number generation within the implementation of the survey (in Limesurvey).
Randomization unit
Randomization is done at the level of private households.
Clusters
NA
Observations
The survey will cover n=4,000 households. n=400 observations are from a pilot study. The main study is planned to cover n=2600 households.
Arms
Control group: 15%. SPF disagreement treatment: 35% Inflation history treatment (verbal): 25% Inflation history treatment (graphical): 25%
Primary outcomes
Indicators of whether households round their inflation expectations to integer values or even multiples of five.
Secondary outcomes
Inflation expectations themselves.

Cited by

No citation data available for this registration yet.

Keywords