Please fill out this short user survey of only 3 questions in order to help us improve the site. We appreciate your feedback!
Randomized Impact Evaluation of the Community Auxiliary Police, Bougainville, Papua New Guinea
Last registered on April 11, 2018


Trial Information
General Information
Randomized Impact Evaluation of the Community Auxiliary Police, Bougainville, Papua New Guinea
Initial registration date
August 20, 2016
Last updated
April 11, 2018 11:39 AM EDT
Primary Investigator
University of California San Diego
Other Primary Investigator(s)
Additional Trial Information
Start date
End date
Secondary IDs
In developing countries, the state is often unable to deliver the protections of the law to all its citizens. When the state’s reach is limited, the few interactions citizens have with police are often marked by corruption, absenteeism, or abuse, contributing to dissatisfaction and distrust. These citizens sometimes instead rely on informal security providers, such as chiefs, families, or mobs, whose punitive and remedial procedures may be biased toward certain social groups at the expense of other groups. For example, men may receive systematically more beneficial outcomes from the informal justice system than women.

This study seeks to measure the effectiveness of a novel project implemented in Papua New Guinea by the New Zealand and Bougainville Police Services that seeks to overcome the challenges of policing in weak state environments, called the Community Auxiliary Police (CAP). The CAP project transfers the state’s constitutional police powers to carefully selected community members in villages located in remote parts of the country, and trains those officers in investigative techniques and legal norms, particularly surrounding gender-based violence. CAP officers are unarmed but do have powers of arrest, detention and investigation in the communities that they come from.

In partnership with the Bougainville and New Zealand Police, researchers are conducting the first randomized evaluation to study the impact of the Community Auxiliary Police program on crime, norms and citizens’ relationship to the state.
External Link(s)
Registration Citation
Cooper, Jasper. 2018. "Randomized Impact Evaluation of the Community Auxiliary Police, Bougainville, Papua New Guinea." AEA RCT Registry. April 11. https://doi.org/10.1257/rct.1505-3.0.
Former Citation
Cooper, Jasper. 2018. "Randomized Impact Evaluation of the Community Auxiliary Police, Bougainville, Papua New Guinea." AEA RCT Registry. April 11. http://www.socialscienceregistry.org/trials/1505/history/28030.
Sponsors & Partners

There are documents in this trial unavailable to the public. Use the button below to request access to this information.

Request Information
Experimental Details
Within the communities in which they live, CAP officers are authorized by the Police Act of 1988 to use the full powers of the police for offenses whose punishment comprises a prison term of no more than 12 months, and the power of detention and referral to the central police for all other offenses. They are uniformed officers and in practice work in their communities as full time police, conducting arrests, dispelling tension, investigating crimes and mediating conflicts. Importantly, they are not armed. The study randomizes the areas in which the CAP program is implemented. Treatment villages have a uniformed CAP officer work and live there as full time agents of the state police. Villages in the comparison group will maintain the status quo, which often means very poor access to state policing, or none at all.
Intervention Start Date
Intervention End Date
Primary Outcomes
Primary Outcomes (end points)
Among households exposed to the treatment: victimization incidents; reporting behavior; perceptions of security; perceptions of formal and informal procedural justice; gender norms and attitudes; trust in the state; and knowledge of the state. Among candidates who are hired or not hired: behavioral measures of honesty; perceptions of self-efficacy, entitlement, power and respect; beliefs about justice; attitudes and legal knowledge; and alignment with community and state norms. See pre-analysis plan for detailed description.
Primary Outcomes (explanation)
The pre-analysis plan gives a detailed description in English and pseudo-code of how each outcome is constructed.
Secondary Outcomes
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
In December 2015 and January 2016, the Bougainville Police Service and the New Zealand police service expanded the serving CAP force in the North and Central regions. Candidates cannot apply directly, but must be nominated by the council of paramount chiefs in their Council of Elders (local government unit). Each COE receives an allotted share of the new recruits - usually 1-2. From the shortlist of candidates that the police had interviewed, tested and among whom they were indifferent about hiring, some candidates were hired through a COE-level lottery, and others were not. Although the Columbia Univerity research team advised on the randomization procedure, it was designed and implemented entirely by the police. Candidates police their own communities if they are hired. Thus, for households in villages that are part of the study ("households"), there are two treatment conditions: having a candidate to the CAP recruited in one’s village (treatment); and not having a candidate to the CAP recruited in one’s village (control). The study also analyzes outcomes among those who applied and were hired or almost hired as CAP officers ("candidates"), thus the two treatment conditions for those subjects are: being hired as a CAP officer (treatment), and not being hired as a CAP officer (control).
Experimental Design Details
Randomization Method
Lottery done in each block of the experiment in the police station by project partners, resulting in cluster-level assignment for households and individual-level assignment for candidates.
Randomization Unit
Blocks are COEs, clusters are villages of households. Units of analysis are households and individual candidates.
Was the treatment clustered?
Experiment Characteristics
Sample size: planned number of clusters
39 villages
Sample size: planned number of observations
1950 households and 45 candidates
Sample size (or number of clusters) by treatment arms
850 households in treatment 1,100 in control. 17 candidates in treatment, 22 in control.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB Name
Columbia University Morningside IRB
IRB Approval Date
IRB Approval Number
Analysis Plan
Analysis Plan Documents



Uploaded At: August 20, 2016

Post Trial Information
Study Withdrawal
Is the intervention completed?
Is data collection complete?
Data Publication
Data Publication
Is public data available?
Program Files
Program Files
Reports, Papers & Other Materials
Relevant Paper(s)