Hard garbling methods for monitoring harassment at scale

How to cite

Gonzalez-Torres, Ada; Boudreau, Laura; Chassang, Sylvain; Heath, Rachel. 2025. "Hard garbling methods for monitoring harassment at scale." AEA RCT Registry. AEARCTR-0014286. https://doi.org/10.1257/rct.14286-1.2

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

Ada Gonzalez-Torres* , Laura Boudreau , Sylvain Chassang , Rachel Heath

* Primary investigator

Registry ID
AEARCTR-0014286
DOI
https://doi.org/10.1257/rct.14286-1.2
First registered
Last updated
Status
In development
Trial dates
2015-12-18 – 2026-03-31
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/14286

Abstract

This research studies how survey design affects transmission of sensitive information within organizations. We partner with a large, international apparel buyer and conduct a phone-based survey experiment with workers across several of its Bangladeshi supplier factories to study mechanisms behind garbling methods shown to increase workers' willingness to report misbehavior by managers. Boudreau et al. (2023) was the first test of hard garbling (HG) outside of a lab, building on Chassang and Padró i Miquel (2018); Chassang and Zehnder (2024). We extend our prior work to test for differences in reporting between higher and low HG rates, to assess workers' preferences over different HG rates, and to assess worker welfare consequences of the ability to report harassment in a way that allows possible deniability.

Intervention

We evaluate secure survey methods to elicit information about misbehavior in organizations. We randomly assign workers to direct elicitation (DE) or hard garbling (HG) with different garbling rates, and also evaluate preferences for different elicitation methods.

Experimental design

We experimentally vary direct and indirect survey methods (i.e. hard garbling at different garbling rates) to ask sensitive questions on management misbehavior. We also randomly assign a subset of people in the DE arm to an explanation of HG, after the module on reporting experienced misbehavior, to learn about preferences for different elicitation methods.

In the survey, we also elicit information about workers’ primary deterrents to reporting, their comprehension and preferences for different reporting systems.

Design details

Randomization method
The randomization is done by a computer.
Randomization unit
The unit of randomization is a worker, stratified by factory, production team or section, and sex.
Clusters
Not applicable
Observations
We aim for a sample size of 5,377 workers from across 3-5 factories.
Arms
Our target sample size is 1793 per treatment arm (with the DE arm having about 592 randomly assigned to the explanation of HG).
Primary outcomes
The key outcome variables are reports of physical harassment and sexual harassment from a supervisor or manager in charge of their team.
Secondary outcomes
Additionally, as secondary outcomes, we study how increased reporting affects worker welfare in the short run. We also include a set of questions to study mechanisms through which HG affects workers’ responses relative to DE. In addition, we study whether garbling affects the reported incidence of fatal accidents (expecting no effect).

Cited by

No citation data available for this registration yet.

Keywords