How Do Policymakers Update?

How to cite

Vivalt, Eva. 2016. "How Do Policymakers Update?." AEA RCT Registry. AEARCTR-0001237. https://doi.org/10.1257/rct.1237-4.0

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

Eva Vivalt*

* Primary investigator

Registry ID
AEARCTR-0001237
DOI
https://doi.org/10.1257/rct.1237-4.0
First registered
Last updated
Status
In development
Trial dates
2016-05-19 – 2019-12-31
Countries
Private
JEL codes
O1, O2
Registry record
http://www.socialscienceregistry.org/trials/1237

Abstract

In order for policymakers to make evidence-based decisions, three things must happen. First, there must be evidence on which they can base their decisions. Second, they must accurately update their beliefs based on that evidence. Finally, they must have the incentive to make their decisions according to what the evidence shows. Most research focuses on generating new evidence or on the last question. We focus on the second step. It is well-known that there are a number of behavioural reasons why people may not accurately update their beliefs. We posit that some of them may be more relevant to policymakers, either because policymakers tend to have slightly different biases than other populations or because people are particularly susceptible to some biases when presented with the kind of information that policymakers typically receive. We design an experiment to explore whether updating biases exist and whether the type of information provided may be able to reduce the impact of these biases if they do exist. We also consider other factors that may affect how policymakers update in response to study results.

Experimental design

Policymakers, practitioners and researchers are surveyed at conferences and by e-mail. We elicit priors, present data, and elicit posteriors. Two information treatments are provided to help mitigate the biases we expect to observe. Participants are asked to make real-world allocations, and we also conduct a discrete choice experiment to estimate how they weight various types of evidence (e.g. evidence from RCTs vs. quasi-experimental studies, evidence from their country vs. others, etc.). Respondents' willingness-to-pay for information is elicited. The MTurk sample focuses on estimating the behavioural biases and their mitigation as a comparison group.

Design details

Randomization method
Randomization by computer as respondent participates.
Randomization unit
Individual.
Clusters
400 policymakers, practitioners and researchers 1,600 MTurk participants
Observations
2,000
Arms
200 policymakers, practitioners and researchers and 800 MTurk participants per each of two information treatment arms 300 policymakers, practitioners and researchers and 1,200 MTurk participants and separately 100 policymakers, practitioners and researchers and 400 MTurk participants for two other information treatment arms
Primary outcomes
- k and mu' in the updating model - real-world allocations - estimated coefficients in the discrete choice model Further details provided in pre-analysis plan.

Cited by

No citation data available for this registration yet.

Keywords