Working Paper: CEPR ID: DP16417
Authors: Ana Beatriz Galvo; James Mitchell
Abstract: Economic statistics are commonly published without any explicit indication of their uncertainty. To assess if and how the UK public interpret and understand data uncertainty, we conduct two waves of a randomized controlled online experiment. A control group is presented with the headline point estimate of GDP, as emphasized by the statistical office. Treatment groups are then presented with alternative qualitative and quantitative communications of GDP data uncertainty. We find that most of the public understand there is uncertainty inherent in official GDP numbers. But communicating uncertainty information improves understanding. It encourages the public not to take estimates at face-value, but does not decrease trust in the data. Quantitative tools to communicate data uncertainty - notably intervals, density strips and bell curves - are especially beneficial. They reduce dispersion of the public’s subjective probabilistic expectations of data uncertainty, improving alignment with objective estimates.
Keywords: Macroeconomic Data; Uncertainty; Uncertainty Communication; Data Revision; Randomized Experiments
JEL Codes: C83; E01
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| communicating uncertainty information (D80) | public understanding of GDP estimates (E20) |
| quantitative tools (C89) | dispersion of subjective probabilistic expectations about data uncertainty (D80) |
| individual characteristics (Z13) | treatment effects (C22) |
| uncertainty communication (D80) | trust in data (Y10) |