EL REICH VENIDERO
I. MITO Y TIPO 1.
The current research has been made possible by a data sharing agreement between the
International Research Centre for Investigative Psychology (IRCIP) and the Greater Manchester
Police (GMP). As part of this agreement, law enforcement data pertaining to crime incidents was
extracted and shared for research purposes. The data itself represents a static cross-section view
of the crime condition from a sub-section of the Greater Manchester area. As far as can be
determined after-the-fact, data manipulations on behalf of GMP were minimal: sensitive data
such as specific individual’s names were removed but very little else was done prior to data being handed off. Finally it should be noted that data was accessed in accordance with IRCIP
and university guidelines and that the research itself was approved by the ethics panel (SREP).
The data itself covered a wide range of criminal activity for the period: it was not limited
to only solved crimes for example; nor was it limited to only specific crime sub-types. This both
opened many interesting lines of inquiry while also introducing several important challenges that
needed to be addressed. First, as the design philosophy has a distinct focus on predictive
outcomes, then it was paramount to identify the subset of records for which the responsible party
of structured outcomes including: charged, undetected, identified by non-charged among others.
For the current research, only charged crimes were considered as these were thought to hold the
highest burden of proof for attribution within the available data set. It is very important to note,
however, that a police charge does not itself constitute a guilty verdict. From the data that was
made available it was not always possible to determine with certainty that a given crime event
could be attributed to the individual indicated within the data. This represents a fundamental
limitation within the data.
Limitations such as these are common when working with second-hand data sources as is
the case here. Contrary to data from classical laboratory experiments, ‘real world’ data are often saddled with a number of issues. First it must be understood that often secondary data are being
used to investigate questions it was not initially collected for which in turn can cause a host of
problems. Law enforcement records, for example, are often simply a log of events: what
happened, where it happened, what did witnesses say, etc. The information that is collected as
part of on-going casework is inherently biased towards achieving a given agency’s goals which
may or may not align with subsequent research questions down the road.
At a more practical level, law-enforcement officers typically do not sign-up because of
their love of forms and paper work. Record keeping is often a secondary concern behind
resolving active incidents and ensuring public safety in general. Thus records may be recorded
well after-the-fact and data contained therein may not always be as specific or accurate as
possible. These challenges are not themselves new or unique to the current research presented
here, nearly all researchers working with secondary data – and crime data specifically –
Youngs, 2009 for a general discussion and Woodhams & Bennell, 2015 for a more targeted
discussion within crime linkage analysis).
Fundamentally, all of these concerns stem from the fact that researchers consuming data
second-hand were not present for data collection and may not be aware of the various ‘quirks’ –
that is business rules – that have impacted how said data was collected in the first place. A
primary example of this is how crime events are recorded which often results in duplicated or
inflated incidence numbers. For example, consider a burglar breaks into student housing, and in
so doing gains access to a number of unlocked rooms. This one criminal event could lead to
many individual crime records. The business rules for collecting such complaint data may dictate
one record be generated for every victim. From the law-enforcement perspective this makes
sense: one offender committed a string of acts that impacted multiple people. However, from the
researcher’s perspective, the fact that there were multiple victims may be the result of
circumstance and that that one evening’s set of events should not be more heavily weighted than another where the same perp only victimized one individual.
The key question for the current research and researchers in general is how to rectify law-
enforcement’s representation of criminal events and what they may or may not represent for their perpetrator. For this particular example, such incidents were generally aggregated into a singular
event in the current research. The rational for doing so is that the number of rooms that were
actually unlocked is a matter of circumstance and was not believed to be the result of any
specific action taken on behalf of the offender. Further methodological reasons for dropping such
‘duplicate’ events are covered in more detail in later chapters. A general discussion of the data made available to the current research follows which is intended to provide context for all