x

We are happy to announce that all trial registrations will now be issued DOIs (digital object identifiers). For more information, see here.
Strengthening Professionalism and Accountability within the Ghana Police Service, using Identities, Norms and Narratives
Last registered on September 23, 2019

Pre-Trial

Trial Information
General Information
Title
Strengthening Professionalism and Accountability within the Ghana Police Service, using Identities, Norms and Narratives
RCT ID
AEARCTR-0003631
Initial registration date
September 20, 2019
Last updated
September 23, 2019 4:17 PM EDT
Location(s)

This section is unavailable to the public. Use the button below to request access to this information.

Request Information
Primary Investigator
Affiliation
University of Oxford
Other Primary Investigator(s)
PI Affiliation
Southern Methodist University (SMU)
PI Affiliation
University of East Anglia
PI Affiliation
University of Ghana, Legon
PI Affiliation
University of Oxford
PI Affiliation
University of Oxford
Additional Trial Information
Status
On going
Start date
2018-10-01
End date
2020-01-31
Secondary IDs
Abstract
Corruption has long been one of the major obstacles to improving economic efficiency and reducing poverty in developing countries. In many places, corruption has become a norm - a way of life - something that is generally accepted as a behavioural standard. How do we change a corrupt norm? The aim of this project is to devise an innovative policy intervention to address this question. We focus on Ghanaian traffic police because it is relatively understudied and the relationship between the traffic police and citizens is a complex one. However, the high frequency of interactions between citizens and traffic police officers means that it is more visible than other forms of corruption. In Ghana, traffic officers are known to either purposely create roadblocks to extort bribes from drivers or to reduce the number and/or the frequency of checks on smugglers/traffickers in exchange for various sums of money (National Road Safety Commission, 2010, 2015). However, ethnographic research has also revealed that citizens who violate the traffic laws often initiate the payment to avoid a court order which is typically a very cumbersome process and time consuming. The aim of our paper is to first gain a better understanding of the organisational culture of the police service, the working environment, and obstacles that police officers face. In particular, we are interested in the role of intrinsic motivations in driving corrupt behaviour. Our conjecture is that if we can increase intrinsic motivation among police officers by instilling a new identity, new mission that they can internalise and identify with, the moral cost of engaging in corruption will increase and thus, they are more likely to refrain from taking/asking for a bribe. Our main treatment is an innovative and intensive identity-building training programme, which uses psychological and team-building exercises to re-shape the officers' social identity and values. We will be working with a training expert from the Brazil's Federal Highway Police to deliver the training to the treatment group of 350 traffic police officers in Ghana.
External Link(s)
Registration Citation
Citation
Borcan, Oana et al. 2019. "Strengthening Professionalism and Accountability within the Ghana Police Service, using Identities, Norms and Narratives." AEA RCT Registry. September 23. https://doi.org/10.1257/rct.3631-1.0.
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
Interventions
Intervention(s)
The intervention will be a training programme aiming to engage the officers' intrinsic motivation for the job, foster a sense of identity and belonging to an effective and honest police force, shift beliefs about what constitutes corruption and thereby change the officers' behaviour in their interaction with citizens. The training will take place over one week. The curriculum will be designed by the research team in collaboration with a police training expert from Brazil's Federal Highway Police and with the advice of a leading professor in psychiatry from the University of Oxford.
Intervention Start Date
2019-04-29
Intervention End Date
2019-05-10
Primary Outcomes
Primary Outcomes (end points)
We are primarily interested in whether the training will change the attitudes of officers regarding: 1) unethical/corrupt behaviour by members of the police force, including their own propensity to engage in such behaviour; 2) their relationship with the citizens. The main outcomes will be collected through a survey (sentiments towards the citizens, perceptions of corruption, in general and in the GPS, tolerance for bending/breaking rules, propensity to notice and report unethical behaviour of colleagues) and two experimental games on honesty.
Primary Outcomes (explanation)
The two experimental games will gauge the propensity to cheat in one individual dice roll game and another similar game played in pairs.
Secondary Outcomes
Secondary Outcomes (end points)
Firstly, we are also interested in survey measures of aspirations and job satisfaction of police officers, which are collected in the main police officers survey.
Another set of secondary outcomes will serve to measure the treatment effect on actual police behaviour reported from other sources. The main source will be a sample of tro-tro drivers who interact with traffic police multiple times every day. We will collect information on the frequency and duration of interactions with the traffic police, police attitudes and any payment exchanges, including who initiated the exchange and the payment amount. To cross-check the incidence of police interactions reported by tro-tro drivers, we will also collect survey data from a sample of tro-tro passengers and route observation notes from our own decoy passengers along the routes of the interviewed tro-tro drivers.
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
The experimental subjects are officers within Ghana's Motor Traffic and Transport Department (MTTD) from the Accra region. There are 14 police divisions, further divided into 42 districts, employing around 700 police officers. This is our target population. Of these, 1 district (Central or Headquarters) is distinctly larger than the rest (around 200 officers). Officers in this district are sometimes deployed to different districts to cover duty posts where the need arises. This district will not be included in the experiment. We will randomly assign two thirds of the police districts to the treatment group. The unit of randomisation will be the police districts in order to reduce contamination across treated and control groups, due to interaction of police officers at the police district headquarters and within-district mixed teams on the ground. All police officers within the treated districts will be offered the training programme in April-May 2019. None of the police officers in the control districts will be offered the training programme at this time, but may be offered the programme in the spring of 2020, after the endline data collection, by local police commanders which we will train to deliver our tailored training programme. We will compare outcomes from police officers' endline surveys and experimental games, as well as drivers, passengers and household surveys, across treated and control districts. Baseline officer survey data was collected in October 2018 and driver and household and drivers/passengers survey data will be collected in April-May 2019. Endline data will be collected in November 2019. This design will allow us to estimate the Intent-to-treat effect (ITT) as the difference in means in outcomes at endline between the treatment and the control groups. In the event of less than perfect compliance we will estimate the local average treatment (LATE) effect by instrumenting compliance with the treatment offer.
Experimental Design Details
Not available
Randomization Method
The randomization will be done in office by a computer. Roughly 20 police districts will be randomly assigned to receive the training programme offer.
Randomization Unit
The randomization unit is the police district.
Was the treatment clustered?
Yes
Experiment Characteristics
Sample size: planned number of clusters
There will be around 30-35 clusters (districts), allowing for exclusion of small districts.
Sample size: planned number of observations
We aim to include in the study in total around 500 officers, which the police effective in all districts, excluding the Central district to avoid spillovers. However, acknowledging that part of the police effective might still be deployed to service the traffic throughout the training period, we aim to train at least 50% of all officers in districts assigned to treatment. With 2/3 of districts in the treatment arm and 1/3 of districts in the control arm, we aim to train a sample of at least 175 officers and compare their primary outcomes to those of: 1. around 150 officers in control districts and 2. around non-treated 175 officers in treated districts.
Sample size (or number of clusters) by treatment arms
We will assign 2/3 of the study districts in the treatment group (20-23 districts) and (10 -12) in the control group.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
In terms of the experimental games on truth-telling, considering a conservative 10 clusters in the control group (sample 1) and 20 clusters (2) in the treatment group with a mean of 15 subjects per cluster in sample 1 and 8 subjects per cluster in sample 2, given a proportion of 58% of subjects reporting a false outcome (and given the 0 intra-cluster correlation detected at baseline), for a power of 80% we can detect a minimum effect of a 30% drop in false reporting (a 30% decrease in lying-consistent behaviour). In terms of the driver survey reports of the payment amounts exchanged with police, considering a conservative of 10 clusters in the control group (sample 1) and 20 clusters (2) in the treatment group with a mean of 33 driver-route observations per cluster in both samples (N=1000 driver-route observations), given a an average payment of 28 cedi (SD 37 cedi), as per our preliminary driver surveys (and assuming 0 intra-cluster correlation), for a power of 80% we can detect a minimum effect of a 7.5 cedi (0.2 of a SD) or 27% drop in the payment amount.
Supporting Documents and Materials

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

Request Information
IRB
INSTITUTIONAL REVIEW BOARDS (IRBs)
IRB Name
Department of Economics, University of Oxford
IRB Approval Date
2018-05-29
IRB Approval Number
ECONCIA18-19-06
Analysis Plan

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

Request Information