Zusha! A randomized behavior change road safety intervention

How to cite

Jack, William; Habyarimana, James. 2014. "Zusha! A randomized behavior change road safety intervention." AEA RCT Registry. AEARCTR-0000554. https://doi.org/10.1257/rct.554-1.0

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

William Jack* , James Habyarimana

* Primary investigator

Registry ID
AEARCTR-0000554
DOI
https://doi.org/10.1257/rct.554-1.0
First registered
Last updated
Status
Completed
Trial dates
2011-03-01 – 2013-12-31
Countries
Kenya (National)
Registry record
http://www.socialscienceregistry.org/trials/554

Abstract

Road accidents kill 1.3 million people each year, most in the developing world. We test the efficacy of evocative messages, delivered on stickers placed inside Kenyan matatus, or minibuses, in reducing road accidents. We randomize the intervention – which nudges passengers to complain to their drivers directly – across 12,000 vehicles. We use insurance claims data and vehicle speeds to assess the impact of the interventions. We also evaluate a complementary radio campaign aired during randomly selected weeks over a 5 month period.

Intervention

Our intervention is aimed at promoting agency among matatu (minibus) passengers, empowering them to challenge the driver to slow down and drive less recklessly if they felt their safety was compromised. To this end, we post stickers with evocative messages inside the vehicles, exhorting the passengers to act, with phrases like “Don’t let a reckless driver get away with murder,” and others. We test the efficacy of different kinds of messages, including the role of images in general in eliciting a response, and the relative merits of using fear versus reason to motivate behavior change. We also assess the impact of messages promoting collective action compared with those aimed simply at alleviating individual-level constraints.

Experimental design

We randomly assign vehicles across 8 treatment groups, one placebo, and a control. Assignment and distribution of stickers occurs at the point of sale of insurance coverage.

Design details

Randomization method
Randomization conducted via computer generation of a random number at the point of recruitment.
Randomization unit
Randomized at the individual level to 14-seater minibuses.
Clusters
12,000 vehicles
Observations
12,000 vehicles
Arms
1,000 vehicles in each treatment arm, 2,000 in the placebo arm, 2,000 in the control.
Primary outcomes
Outcomes are insurance claims and speeds recorded by GPS on study vehicles.

Cited by

No citation data available for this registration yet.

Keywords