NEW UPDATE: Completed trials may now upload and register supplementary documents (e.g. null results reports, populated pre-analysis plans, or post-trial results reports) in the Post Trial section under Reports, Papers, & Other Materials.
CSI in the tropics: Experimental evidence of improved public service delivery through coordination
Initial registration date
June 29, 2020
June 30, 2020 12:12 PM EDT
Universidad de los Andes
Other Primary Investigator(s)
Additional Trial Information
We randomly assign the investigations of 66% of the 1,683 homicides occurring in Bogota, Colombia, during 2016 to a new procedure in which teams of responsible functionaries are in charge of a given case throughout the investigation process. We find a statistically significant 30% increase in the conviction rate in the treatment group relative to the control group. Indicators of the quality of the investigative process also improve, as well as the rate at which a formal accusation is presented before a court. Complementary findings suggest that the team work induced by the treatment improves communication, coordination, feedback, and motivation.
Collazos, Daniela et al. 2020. "CSI in the tropics: Experimental evidence of improved public service delivery through coordination." AEA RCT Registry. June 30.
Seeking to increase judicial efficiency, the District Attorney’s Office of Colombia developed and implemented a new policy that restructured its murder investigation procedure to avoid the loss of information between public servants and increase coordination and leadership. Specifically, fixed investigative teams of lab experts, detectives and a district attorney were created to handle homicide cases from the initial investigation until the bill of indictment.
Intervention Start Date
Intervention End Date
Primary Outcomes (end points)
Our main outcomes are actions and decisions taken by the CSIs, detectives, and the prosecution. We use the Prosecution Office outcomes to measure performance: indictment, bill of charges, sentence and conviction rates.
Primary Outcomes (explanation)
Indictment: The suspect(s) was (were) formally charged with the commission of the crime.
Bill of charges: A formal written document accusing the suspect(s) of having committed a crime is filed.
Sentence: Cases that had a sentence, although the defendant was not necessarily found guilty.
Conviction: Cases in which the defendant was found guilty.
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
In order to randomly allocate homicide cases between treatment and control groups, in cooperation with the Prosecution Office we placed strict rotation rules on the investigative teams on call for each shift. The first step was to form six fixed investigative units (nine lab experts, five detectives, and a prosecutor) that would cover the city of Bogota. Four of the groups were assigned to a treatment status via a simple raffle.
Experimental Design Details
For teams of functionaries, simple raffle in front of the public servants. For homicide cases, rotation rules between investigative teams shifts guarantee random allocation of homicides to the treatment and control groups.
Was the treatment clustered?
Sample size: planned number of clusters
Sample size: planned number of observations
Close to 1500 cases
Sample size (or number of clusters) by treatment arms
Control group: close to 500; Treatment group: close to 1000
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
From our power calculations we obtained an average Minimum Detectable Effect (MDE) of 0.13 Standard Deviation (SD) for the indictment and an average MDE of 0.14 SD for the bill of charges, the two main outcomes of the program.
INSTITUTIONAL REVIEW BOARDS (IRBs)
Universidad de los Andes
IRB Approval Date
IRB Approval Number
Proceedings No: 689 of 2017
Analysis Plan Documents
June 26, 2020
Post Trial Information
Is the intervention completed?
Intervention Completion Date
December 04, 2016, 12:00 AM +00:00
Is data collection complete?
Data Collection Completion Date
Final Sample Size: Number of Clusters (Unit of Randomization)
Was attrition correlated with treatment status?
Final Sample Size: Total Number of Observations
1,683 homicide cases
Final Sample Size (or Number of Clusters) by Treatment Arms
573 in the control group and 1,110 in the treatment group
Reports, Papers & Other Materials
This paper evaluates the impacts of increased coordination, accountability, and leadership among teams of responsible public officials, with evidence from homicide investigations in Colombia. We randomly assigned the investigations of 66% of the 1,683 homicides occurring in Bogotá, Colombia, during 2016 to a new investigation procedure emphasizing these features. We find a statistically significant 30% increase in the conviction rate in the treatment group relative to the control group. Indicators of the quality of the investigative process also improve, as well as the rate at which a formal accusation is presented before a court. Complementary findings suggest that the treatment produces well-coordinated teams that can communicate more fluently. Also, a survey of investigative team members reveals that work motivation, the extent to which they receive feedback on their performance, the pertinence and effectiveness of their roles, and the perceived quality and coordination of the team all improve under the new scheme.
Daniela Collazos & Leopoldo Fergusson & Miguel La Rota & Daniel Mejía & Daniel Ortega, 2020. "CSI in the tropics Experimental evidence of improved public service delivery through coordination," Documentos CEDE 2020-20, Universidad de los Andes - CEDE.
REPORTS & OTHER MATERIALS