Outreach to Domestic Abuse Victims in Times of Quarantine

How to cite

Grogger, Jeff; Ivandic, Ria; Kirchmaier, Tom. 2020. "Outreach to Domestic Abuse Victims in Times of Quarantine." AEA RCT Registry. AEARCTR-0005723. https://doi.org/10.1257/rct.5723-1.1

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

Jeff Grogger* , Ria Ivandic , Tom Kirchmaier

* Primary investigator

Registry ID
AEARCTR-0005723
DOI
https://doi.org/10.1257/rct.5723-1.1
First registered
Last updated
Status
In development
Trial dates
2020-04-17 – 2021-11-17
Countries
Private
JEL codes
K42
Registry record
http://www.socialscienceregistry.org/trials/5723

Abstract

There is concern worldwide that the coronavirus epidemic will lead to an increase in domestic abuse (DA), as stay-at-home orders force victims to spend more time in at home with their abusers. At the same time, there is concern that the reporting of domestic abuse may fall. The result is a greater number of victims, increasingly isolated and without recourse. We are working with Thames Valley Police (TVP) to devise an outreach campaign targeted toward women at high risk of domestic abuse, to inform them how they may safely contact police.

To measure the impact of the outreach campaign, we are conducting an RCT that randomly assigns high-risk DA victims to treatment and control status. The treatment group is targeted for extra social media advertising that bears the campaign’s message; the control group is exposed only to various national campaigns that are ongoing at the same time. Outcomes to be analyzed include calls for service and calls for service by method during the ad campaign, and post-campaign measures of calls for service and incidents involving violence as measures of recidivism.

Intervention

We devised materials for a Facebook/Instagram (FB) advertising campaign and an associated website. These provide readers with instructions for making reports to police via the Silent Solution, an approach that allows callers to make reports with little if any speaking. To test the effectiveness of the campaign, we will identify a group of high-risk potential DA victims based on their previous victimization history. We will randomly assign half of this high-risk pool to treatment and half to control. Those in the treatment condition will be targeted for the ad campaign via FB.

Experimental design

Our main analysis will be based on the RCT, which involves assigning half the high-risk group to treatment and half to control. We will use the RCT design to identify the intent-to-treat effect of the ad campaign on contemporary calls for service and calls for service by type. We will also use the RCT design to measure the ITT of the ad campaign on recidivism. Depending on the strength of the first stage, we may also use an IV strategy to estimate how calls for service during quarantine affect recidivism. Finally, we will use a before-after approach to estimate the extent to which the quarantine affected DA. This analysis will compare the number of calls before quarantine to the number of calls after.

Design details

Randomization method
Computer
Randomization unit
Individial
Clusters
TBD
Observations
TBD
Arms
TBD
Primary outcomes
To measure the contemporary effect of the campaign, we will analyze calls for service regarding DA incidents and calls for service by method (e.g., Silent Solution v. conventional calls). We will track these both during the campaign and for one month afterward. To measure recidivism, we will track calls for service and incidents involving violence for one year after the end of this period.
Secondary outcomes
We will analyze data from FB regarding the reach of the ad campaign, including number of impressions, clicks, geography, and demographics. We will analyze similar data for our website from Google Analytics.
Partners
Name: Thames Valley Police Partner_type: government Url: https://www.thamesvalley.police.uk/

Cited by

No citation data available for this registration yet.

Keywords