28
International Journal of Security and Terrorism • Volume: 4 (1)
these crime analysis dimensions by making index variables; the authors of the current study
created them as latent variables by performing factor analysis.
The operationalization of crime analysis is based partially on O’Shea and Nicholls’
(2003) study. In their study, O’Shea and Nicholls conceptualized crime analysis with three
main titles, which they referred to as the “dimensions of crime analysis” (p. 238). These three
dimensions are as follows: crime analysis functions, statistical methods, and data utilization.
However, only the first two groups of variables were used as main independent variables in
this study. There were 22 types of crime analysis activities in O’Shea and Nicholls’ (2003)
survey, in which they ask respondents to indicate how frequently they undertook each type
of crime analysis. The answers were coded as never, some, often, and very often. The types
of crime analysis specified are the following: target profile, victim, link, temporal, spatial,
financial, flowcharting, program evaluation, case management, crime scene profiling, crime
forecasting, crime trends, citizen surveys, victim surveys, employee surveys, environmental
surveys, intelligence, productivity, civil litigation, patrol strategy, workload distribution, and
displacement/diffusion analyses. In the second dimension (i.e., statistical methods), for
each statistical method, respondents were asked to indicate how frequently they use the
corresponding method. These methods are the use of the following: frequencies, mean-
median-mode, standard deviation, cross tabulations, correlation, regression, and cluster
analysis. The answers were coded as “never,” “some,” “often,” and “very often.” Using
these two groups of variables, an exploratory factor analysis was performed.
4
Table 1 shows the factor analysis results of the main independent variables. Based
on the results, six factor coefficients were created. Factor component scores less than
0.50 were not taken into account. These six factor components are as follows: statistical
analysis, crime analysis, intelligence analysis, survey analysis, patrol strategy analysis, and
displacement/diffusion analysis. The rationale for using factor analysis is the possibility of a
latent variable, which is not directly observable but can be assessed using indicators such
as the frequency of employing specified crime analysis methods.
When factor analysis was used (see Table 1), variables were clustered as follows:
five under the statistical factor correlated highly with the latent variable and had values
ranging from 0.700 to 0.829; seven under the crime analysis factor correlated with the latent
variable and had values ranging from 0.524 to 0.683; five under the intelligence analysis
factor had component scores ranging from 0.556 to 0.756; four under the survey analysis
factor had factor scores ranging from 0.587 and 0.818; three under the patrol strategy
analysis component ranging from 0.663 to 0.744; and one correlated with the latent variable
displacement/diffusion analysis and had a score of 0.523.
4
SPSS version 16.0 software was used.
29
Intelligence - Led Policing:
How the Use Of Crime Intelligence Analysis Translates in to the Decision-Making
Table 1. Exploratory Factor Analysis
Crime Analysis Type
1
2
3
4
5
6
Standard deviation
0.829
Mean, median, mode
0.735
Regression
0.734
Cross tabulations
0.702
Correlation
0.700
Target
profile analysis
0.683
Crime trends
0.675
Victim analysis
0.619
Crime forecasting
0.586
Frequencies
0.554
Spatial analysis
0.534
Temporal analysis
0.524
Financial analysis
0.756
Flowcharting
0.684
Program
evaluation
0.569
Link analysis
0.564
Intelligence analysis
0.556
Citizen
surveys
0.818
Victim surveys
0.799
Employee surveys
0.703
Environmental surveys
0.587
Productivity analysis
0.744
Workload
distribution
0.731
Patrol strategy analysis
0.663
Displacement/diffusion analysis
0.523
Note. The scores from 1 to 6 represent the number of principal components and latent variables. The values in
these (component) columns indicate the level of correlation depending on the exploratory factor analysis.
3. Statistical Analysis and Findings
3.1. Statistical Analysis
All of the processes regarding data merging and data analysis were done using two
statistical software packages: Stata version SE10 and SPSS version 19.0. Because the
dependent variables (i.e., command-level manager, patrol officer, and detective) have an
ordinal/categorical level of measurement, it cannot be measured simply as an ordinary least
squares (OLS) model (Agresti, 2002; Aldrich & Nelson, 1984; Long, 1997; Long & Freese,
2006; McCullagh & Nelder, 1989; Powers & Xie, 1999), but an ordered logistic regression