Consequences
of Profiling for
Crime/Terrorism
: Ethical dilemnas
By David Patterson, Ph.D., Professor of Criminal Justice
Abstract
Over
the past few decades, law enforcement has relied more and more frequently on
the practice of profiling to help identify criminal perpetrators. More recently, it has been used in attempts
to identify possible terrorists. Upon
analysis of multiple crimes in which the unknown perpetrators have similar
characteristics we can develop a predictive description of the characteristics
of the probable perpetrator. These
characteristics typically include variables such as age, sex, race, and
intelligence, among others. A profile is
a predictive device. Fortunately,
sometimes predictions are right.
Unfortunately, sometimes predictions are wrong.
In
addition to being correct, erroneous predictions can also result in two types
of errors. Alpha errors are when we do not correctly predict the perpetrator.
Beta errors are when we predict one to be the perpetrator and the person is in
fact innocent. Either type of error
cause consequences for both the individual and the profiling agency. These consequences can be economic,
political, and ethical.
This
paper will concentrate on errors in prediction and the consequences of those
errors from two ethical perspectives, the deontological approach and the
utilitarian approach.
Inability
to adequately measure the magnitude of these two types of errors makes it
difficult to know the true success/failure rate of profiling techniques. Better prediction leads to smaller ethical
dilemmas. An attempt is made to relate
the severity and frequency of offenses with the ethical approach most likely to
be used to justify law enforcement intervention.
Consequences of Profiling for Crime/Terrorism : Ethical dilemnas
In
its most general sense criminal profiling can be defined as the scientific
establishment of characteristics of a criminal based on behavioral and other
characteristics that are revealed by the way in which criminal activities have
been committed. Most criminal profiles
today seem to be the result of “behavioral” profiling as developed in the U. S.
by the Behavioral Science Unit of the Federal Bureau of Investigation (in the
1950’s) and other similar organizations around the world. Historically we can find less sophisticated
attempts at profiling, such as in the attempt to identify Jack the Ripper in
In
addition to profiling based on behavioral characteristics, it is also possible
to create profiles based on other considerations. Cultural factors (such as religion and social
class) are prominent variables identifying certain terrorist
individuals/groups. Physical (especially race, age, gender) factors may also
create a basis for a profile. Finally,
geographical factors (nationality, ethnicity) contribute to an eventual
profile.
Profiling
may be achieved through either inductive or deductive reasoning. Inductive profiling involves the analysis of
characteristics of other criminals (e.g. serial murders) and assumes that all
serial murderers have similar characteristics.
Primary users of the inductive method include the Federal Bureau of
Investigation (FBI), psychologists and criminologists. The FBI profiling strategy consists of six
stages: (1) profiling inputs, (2) decision process models, (3) crime
assessment, (4) criminal profile, (5) investigation, and (6) apprehension
(Douglas, Ressler, Burgess and Hartman, “Criminal Profiling from Crime Scene
Analysis” in Behavioral Sciences and the Law. Vol. 4 (1986), pp. 401 -
426.)
Deductive
profiling, on the other hand, involves the construction of a criminal’s
characteristics based on the analysis of the forensic evidence acquired at the
crime scene(s) of the offender’s criminal activity. Primary users of the
deductive method are those police officers who actually investigate the crime
scene. Turvey suggests that the
inductive method is unreliable and that the deductive method will be more
effective (for additional discussion see: www.corpus-delecti.com/Profiling_law.html).
In
making a distinction between criminal and terrorism profiling, it might be
suggested that the inductive technique might be used more frequently for
terrorist profiling because we are often not looking for an individual
perpetrator but a whole class of potential perpetrators. In criminal profiling, since we are more
likely to be seeking a single individual, the deductive approach might be more
frequently chosen as the technique used.
It should be reinforced that a profile does not identify a suspect. A profile can only give some kind of “general
biographical description of the most likely type of unknown suspect”
(www.criminalprofiling.ch/introduction.html).
This
paper is an attempt neither to defend nor to criticize profiling. When profiling succeeds, good publicity is
received and the use of profiling is reinforced. Few questions are raised about its use. However, when profiling fails there is little
publicity, its use is not necessarily reduced and few questions are asked. The failure of the use of the profile however
does create consequences and these consequences should be addressed. This is the focus of this paper.
Behavioral
profiling assumes that there is a relationship between personality and
criminality. Upon sufficient repetition
of crime, a series of behavioral characteristics may evolve from the crime
scene investigation data (deductive reasoning).
Usually implicit in this profile is that assumption of some type of
behavioral abnormality which can be classified (paranoia, psychopathology,
etc.). We can cite any number of cases
(in the United States) where criminal profiles (based on behavioral profiling)
have been quite successful including the Unabomber (although it took us several
decades), the “Son of Sam”, and George Metesky (the mad bomber) to name a few.
In
the specific case of the “Unabomber”, the Unabomber in 1995 wrote an extensive
35,000 word Manifesto of his evaluation of the problems with the
technological world (NOTE: the entire text of his Manifesto can be found
at http://www.thecourier.com/manifest.htm). This manifesto was published after
having sent numerous mail bombs, which resulted in several fatalities between
1978 and 1995. Profilers, psychologists
and linguists thoroughly analyzed this document and established his
“profile”. The Unabomber was
characterized as being very intelligent, very educated, a loner, and a
sociopath. Unfortunately, these
characteristics did not provide very many specific clues allowing law
enforcement to identify a suspect.
Fortunately, with the publication of the Manifesto, David
Kaczynski recognized the writing style as being very close to that of his
brother Ted and contacted the authorities leading to his arrest in April,
1996. Theodore John Kaczynski had earned
a bachelor’s degree in mathematics from
In
spite of the success in the Unabomber case however, we have also had some
spectacular failures. Profiles of many
perpetrators have been created with no arrests and in the case of the District
of Columbia Sniper case in 2002 (Boyd Lee Malvo and John Allen Muhammad), the
profile was simply wrong – neither assailant fit the profile. In addition, most
criminal profiles are established on a single assailant theory. Attempts to profile in cases of terrorism is
made even more difficult given the tendency of terrorists acts to be a
conspiracy, that is, committed by more than one individual To the extent that this is true a profile may
be ineffective or even counterproductive.
In
the basic jargon of hypotheses testing, when we predict an outcome incorrectly
we can distinguish between two types of errors in prediction. An Alpha error (or type one) is when we do
not predict one to be the perpetrator and yet they are. A Beta error (or type
two or “false positive”) is when we predict one to be the perpetrator and the
person is in fact innocent. Each of
these two errors carries consequences—some of which are ethical in nature. If we commit an Alpha error, we have not only
failed to identify the real perpetrator but (especially in the case of
terrorism) we may have failed to identify many potential terrorists because we
have a faulty model (we are looking for the wrong kind of person). If we commit a Beta error, we subject
innocent individuals to probable loss of liberty, public humiliation, and
economic costs to “prove” their innocence, and a host of other negative consequences
(such as allegations of racial or religious discrimination). To what extent is it acceptable to subject
large numbers of suspects to arrest/detention in order to make sure we “get the
right one”?
After
the bombing of the twin
Another
element of security has long been computerized passenger profiling (used by the
airlines). Based on a variety of “risk”
criteria, certain passengers would be selected for “secondary” screening. This technique (CAPPS-2) has been scrapped
(it was too cumbersome and ineffective) but another system will be implemented
to replace it. This technique did not catch
the 9/11
Other
forms of profiling have been used at points of entry into the
Because
of the American intervention in
In
all of the above scenarios, we can ask the question: how many of these
individuals are terrorists and how many of them are citizens wrongly
accused? Further, if our profile not
correct, how many real “terrorists” might be slipping through our dragnet
(Alpha error)? A superficial observation
would be that the number of “terrorists” identified (and prosecuted/ convicted)
seems to be quite small (successful prediction), yet the number of individuals
who have be labeled as “terrorist” (but not charged/convicted) is quite large
(Beta error). Further, the number of
“average” citizens who are inconvenienced by the consequences of profiling is
quite massive (Beta error). The
Transactional Records Access Clearinghouse (TRAC) reported that between 2001
and 2003, more than 6,400 people have been recommended for prosecution (2,681
cases completed) in the
The
ethical issue being raised in this paper is this: How many and what kinds of
errors in prediction (and in what magnitude) can we tolerate? Unfortunately, as with most ethical issues
there is no definitive answer. In
looking for ethical justifications for accepting errors, we might use either
the utilitarian or the deontological point of view.
From
the utilitarian perspective, the risk of terrorism might be so great that a
larger number of Beta errors are acceptable (as long as the perpetrators and
potential future perpetrators are apprehended).
After 9/11, as long as terrorist profiling reduced terrorism American
citizens were willing to suffer the inconveniences of long lines at airport
checkins, security screenings and higher costs of flying in exchange for the
warm cozy feeling of being safe.
However,
from the deontological perspective we might conclude that government has the
absolute duty to protect all citizens regardless of the number of innocent
casualties that might occur. Cultural
definitions here might be critical. Some
countries (such as the United States) puts the greatest importance on the
liberty rights of the citizen while others countries (perhaps France, Japan,
China) sees the right of the state to be of greater importance than the rights
of its citizens. The immediate impact of the 9/11 Bombing was that American
citizens were immediately willing to relinquish some of their liberty in
exchange for their perceived safety.
It
is becoming more evident, terrorism is not simply an American phenomena, it is
an international phenomena. Should other
countries come under terrorist attack similar to the
(1)
Rationale for Acceptability of Governmental
Intervention in
Hypothesis:
the greater the severity of the crime/terrorism risk the greater the likelihood
of the deontological approach being used to justify intervention. The greater
the volume of crime/terrorism the greater the likelihood of the utilitarian
approach being used to justify intervention.
|
PHILOSOPHICAL
RATIONALE FOR GOVERNMENTAL INTERVENTION |
SEVERITY OF CRIME/TERRORISM |
||
|
|
|
HIGH |
LOW |
|
VOLUME OF
CRIME/ TERRORISM |
HIGH |
Both |
Utilitarian |
|
LOW |
Deontological |
Neither |
|
The rationale for this hypothesis is that
when the volume is high, each citizen has an increased risk in individual
victimization and demands action (egoism). When the severity is high citizens
are more likely to “feel the pain” of the few and demand action (altruism).
(2)
Acceptance of Errors in Prediction (Alpha and
Beta errors)
Hypothesis:
The higher the frequency and severity of the crime/terrorism risk the greater
the tolerance for Beta errors in prediction; the lower the frequency and
severity of crime/terrorism the greater the tolerance of Alpha errors in
prediction.
|
ACCEPTANCE
OF ERRORS IN PREDICTION |
SEVERITY OF CRIME/TERRORISM |
||
|
|
|
HIGH |
LOW |
|
FREQUENCY OF
CRIME/ TERRORISM |
HIGH |
Beta |
Both
moderate |
|
LOW |
Both
moderate |
Alpha |
|
The rationale for this hypothesis is this: When both the frequency and the severity of
crime/terrorism occur, citizens are willing to allow innocent people to be
arrested in the hopes that the “dragnet” will also catch the real
perpetrator. However, when the frequency
and severity is low, citizens are much more willing to allow the perpetrator to
go free rather than wrongfully arrest one that is innocent.
What are the costs of alpha and beta
errors and does the cost influence the type of errors that are tolerated? Alpha errors are very cheap in the present tense,
but can be very dangerous to our security in the future. Beta errors become very expensive in the
present but may increase our security in the future.
What is the public confidence in the
criminal justice system? In high profile
offenses, (high frequency or severity) public confidence diminishes when an
arrest is not made (alpha error) but they may be consoled if many are caught up
in the “dragnet” (beta error).
Obviously, the criminal justice
system would like to “get their man”, but in the absence of that, it appears
that criminal justice policy makers seem to have a preference for beta errors –
it gives the pretense that they are doing their job even though it is more
expensive in the present (They do not have to worry about a “profit” motive. Citizens in fear are less likely to complain
about tax increases or budget deficits).
CONCLUSION
It has become
evident that a small number of people can wreak havoc over very large
populations. The ability of law
enforcement to effectively identify and arrest profiled perpetrators is
limited. To do so would be very
expensive with still only limited success.
Although theoretically sound as an investigative technique, profiling
for crime or terrorism has only a very limited set of circumstances in which it
can be done efficiently (taking into account the cost and deprivation of civil
liberties required to achieve success).
The hypotheses presented above, if correct, would raise serious issues
about future use of profiling except in the case of looking for a specific
terrorist, not a class of terrorists.
These hypotheses and conclusions are based primarily on the ethical
issues related to the deprivation of civil liberties required (beta errors) for
a successful arrest to be made. The
massive cost of profiling a class of terrorists should be somewhat
self-evident.
REFERENCES
“Criminal
Terrorism Enforcement Since the
Douglas,
Ressler, Burgess and Hartman, “Criminal Profiling from Crime Scene Analysis” in
Behavioral Sciences and the Law. Vol. 4 (1986), pp. 401 - 426.
Meyer, C.B.
“Introduction to Criminal Profiling”, www.criminalprofiling.ch/introduction.html.
Petherick,
Wayne, “Criminal Profiling”, www.crimelibrary.com/criminology/criminalprofiling2.
Schneier,
Bruce. “Terrorism False Positives”, www.schneier.com/blog/archives/2005/01/terrorists_and.html,
Schneier,
Bruce. “Terrorists and Border ID Systems”, www.schneier.com/blog/archives/2005/01/terrorists_and.html,
Schneier,
Bruce. “Behavioral Assessment Profiling”,
www.schneier.com/blog/archives/2004/11/profile_hinky.html,
“Theodore
Kaczynski”, http://en.wikipedia.org/wiki/Theodore_Kaczynski.
“The
Unabomber’s Manifesto”, http://www.thecourier.com/manifest.htm.
Turvey, Brent,
“Deductive Criminal Profiling: Comparing Applied Methodologies Between
Inductive and Deductive Criminal Profiling Techniques”, www.corpus-delecti.com/Profiling_law.html, January, 1998.
Winerman, Lea.
“Criminal Profiling: the reality behind the myth”, www.apa.org/monitor/julaug04/criminal.html.