Page 104 - Violence Prevention Through the Engagement of Violence Producers

COMMUNITY SELECTION AND COMPARATIVE CASE STUDY DESIGN

Ten communities were selected by way of systematic matching across levels of violence
79
to create a comparative case study design. Case studies are extremely detailed studies,

and are relatively costly and time-consuming. However, they are necessary for the
evaluation of complex programmes. Four categories of violence were conveniently

created for the purpose of analysis. Low violence is used in the comparative analysis to
mean communities below the civil war benchmark of 30 per 100,000 (<30 per 100,000);

medium violence between the benchmark and Jamaica’s 2017 average homicide rate
(30-60 per 100,000); high describes rates that exceed the country’s average, but not that

of the Iraq war-related deaths (61-205 per 100,000); and extremely high, those that exceed
the Iraq war-related deaths (206 or more per 100,000).

Table 3 sets out the systematic matching employed in the comparative case study design.

Notice that there are two sets of homicide figures: those from the Jamaica Constabulary
Force (JCF) and actual body counts. The communities were placed into violence levels

according to their average homicide rate for the years 2013 to 2016, using official figures
from the JCF. While these data were used for the selection of the communities, for two

reasons they were not expected to be enough to do the contextual profile needed to
assess the contribution of the VIs. First, police data are not complete body counts, as

they do not include civilians killed by the police, nor persons declared ‘missing’ who local
people know are dead. Second, homicide rates alone (while a good proxy for violence)

do not determine the complexity of a community or the degree of difficulty experienced
by VIs. For this reason a complete baseline was included in the evaluation project.

The actual body counts were needed for the years that the VIs worked in the communities.

The team could have used the official figures, especially since body count exercises are
difficult and painful to those assisting. Nonetheless, we wanted the truest possible figures to

allow us to see how the homicides changed over the period the VIs worked (2016 to 2018).
Body counts were meticulously done from multi-sources. First we sat with elders to do the

count. This was followed by counts from 700 youth in the integrated survey. Finally the VIs
were asked to produce the numbers and list of all the persons who died while they worked.

________________________________________________________________________________________________________________
79 Originally 12 communities were selected. Nonetheless, one of the communities (Greendale) was placed in the
wrong category in the original document sent to the team from CSJP. It was dropped for the study, along with its
matching community (Rose Heights) in St. James. This helped to reduce the time spent on fieldwork, especially given
the lateness of the study.
10