How does a fraud scandal impact trust in science?

How to cite

Eliezer, Leon Viscome; Garzon, Luigi; Goulart, Nícolas. 2024. "How does a fraud scandal impact trust in science?." AEA RCT Registry. AEARCTR-0013481. https://doi.org/10.1257/rct.13481-1.1

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

Leon Viscome Eliezer* , Luigi Garzon , Nícolas Goulart

* Primary investigator

Registry ID
AEARCTR-0013481
DOI
https://doi.org/10.1257/rct.13481-1.1
First registered
Last updated
Status
Completed
Trial dates
2024-04-01 – 2024-06-30
Countries
Brazil ()
JEL codes
D73, D83, D84, I23
Registry record
http://www.socialscienceregistry.org/trials/13481

Abstract

This study addresses how unethical behavior in scientific research impacts trust in science, contributing to a literature on the consequences of corruption and unethical behavior. We propose an informational delivery experiment with high school students in Brazil, where we present them with a summary of a suspected fraud scandal in behavioral science. We divide treated subjects into a "fraud" arm, where they only learn about the fraud accusations, and an "accountability'' arm, where they learn about the investigation and punishment processes involved; this allows us to tell apart the effects of learning about cheating from catching cheaters. We measure the effect on beliefs in science, the use of scientific evidence to update world views, real life attitudes towards science, and spillovers to other fields of research. The results of this experiment can enlighten academic institutions about the consequences of fraud, and guide public communication with respect to informing about accountability.

Experimental design

Teenagers who participate in the Brazilian Economics Olympiad can take part in a survey experiment, where they answer a baseline and an endline survey a couple of weeks apart from each other. The informational intervention is in the beginning of the endline survey. We use answers from two surveys to construct a measure of belief updating. After the endline survey, participants are reached with the possibility of engaging in a behavioral economics class for free, and with a third survey wave to collect longer term effects on beliefs.

Design details

Randomization method
Blocks were matched in triplets, following an adaptation of Yuehao Bai's "pairs of pairs" algorithm, and then randomized to different treatment arms.
Randomization unit
Block randomization at the school level, to avoid contamination bias.
Clusters
500 schools.
Observations
1500 students.
Arms
Average between 2 and 3 students per school, with substantial heterogeneity.
Primary outcomes
Trust in science, interest in academic activities, use of scientific evidence to update beliefs (which are constructed in a survey), and real world actions towards scientific-related activities, which include reading and sharing scientific evidence, demanding books related to scientific topics, and attending a class on a science-related topic.

Cited by

No citation data available for this registration yet.

Keywords