BJS home page
About BJS
|
Developing Improved
Identification Systems: Biometric Systems for Investigations and Background
Checks
by Lawrence A. Greenfeld, Director Bureau of Justice Statistics
Presented to the National Conference Sponsored by BJS and SEARCH
Beyond The Technology: The Law and Policy Implications of Biometric Use
New York City
November 5, 2002
I am delighted to
be here to join you in this important national conference that we are
sponsoring together with SEARCH to continue our ongoing commitment to
learn more about the uses of advanced technologies, particularly with
respect to identification needs for investigative purposes as well as
for background checks. Since 1970, BJS and its predecessor agency have
played an important role in partnering with the States and the FBI to
computerize criminal history records and enhance their utility and shareability
across jurisdictions. We have done this for more than 3 decades with substantial
commitments of Federal funds to the States because we know criminals are
mobile and we believe it to be a professional responsibility to keep good
records of their transactions with the justice system. Many of the national
statistics we produce are the direct result of improvements made in criminal
records bookkeeping.
A few years back,
I approached the CODIS folks in the FBI when CODIS was in its infancy
about the possibility of showing, through an analysis of the post-prison
behavior of violent sex offenders, that such offenders were both active
and mobile after prison discharge. I suggested that if we could track
such offenders and their recidivism across States it would be an important
element in demonstrating the need for a national repository of biometric
information on sex offenders. The FBI, for the first time in my memory,
gave us money to carry out such a study.
A few months back,
BJS completed a preliminary followup of the post-prison recidivism of
more than 270,000 offenders discharged from prisons in 15 States. Each
releasee was followed for 3 years but their entire criminal records, both
within the State in which they served time and other States where they
may have been arrested, were merged together. The 270,000 offenders accumulated
4.1 million arrests before the most recent imprisonment and another 744,000
arrests within 3 years of leaving prison. In other words, we had records
of nearly 4.9 million arrests attributable to these offenders, an average
of 18 arrests each. The most active 12% of the released prisoners, those
with 35 or more arrests on their criminal records, accounted for nearly
50% of all the arrests attributed to the entire cohort. We estimate that
during the first year following prison, the released prisoners accounted
for 11% of all murder arrests, 10% of all robbery arrests, and 12% of
all burglary arrests in the States into which they were released.
Just looking at their
records, the 270,000 offenders had arrests for 18,000 murders, 50,000
rapes and sexual assaults, 172,000 robberies, a quarter million assaults,
nearly 1 million arrests for drug offenses, and more than 160,000 arrests
for weapons offenses. Of the 4.9 million arrests they managed to accumulate
on their RAP sheets, about 400,000 were arrests in States other than the
State in which they did time. And these are the things they were caught
for—would the number of out-of-State arrests be even larger if the seamless
sharing of accurate and complete records together with a linking automated
biometric measure were the norm? My hunch is there would be many more
arrests relative to actual offending if a multitude of biometric identifiers
were routinely collected by justice agencies when they have contact with
an offender and if such information were properly archived and made accessible
for law enforcement purposes.
In statistics, it
is always our goal to make a measurement of some phenomenon as close to
the phenomenon as we can. It is better, for example, to estimate and describe
the crime rate by talking to victims rather than the police since much
of what victims experience is never brought to the attention of the police.
Similarly, identifying biological material left behind by an offender
at a crime scene is a more precise descriptor of who the offender was
than can be obtained from an eyewitness or bystander to an incident. This
is especially true for homicide, rape, and sexual assault offenses.
Several years ago,
BJS produced a study of offenders serving time in State prisons throughout
the nation who had committed a violent crime against a child. Surprisingly
and most unexpectedly, two out of three offenders serving time after conviction
for rape or sexual assault said that the victim had been a child under
the age of 18. In a more recent BJS statistical research report using
six years of NIBRS data from 12 States, we found that two out of three
victims of rape or sexual assault had been a child-in other words, law
enforcement data confirm what offenders had self-reported ---that is,
rape and sexual assault are largely crimes against children. In addition,
both NIBRS data and prisoner data confirm that the vast majority, over
90% of the cases from both data sources, involve victims and offenders
with a prior relationship. Even when the victim was an adult, 3 out of
4 incidents involve victims and offenders who know one another.
Obviously the nature
of victims and offenders in rape/sexual assault cases poses unique problems
and opportunities for the collection of biometric evidence such as DNA
for investigative purposes. Child victims and potential offenders who
are friends and family members make such investigative evidence gathering
emotionally difficult. Because of the victim’s youth and the closeness
of the victim-offender relationship, the majority of rapes and sexual
assaults are never reported to police-BJS data from victims of rape and
sexual assault indicate that only about 32% of incidents are reported
to police, thus to a very large degree precluding the collection of biometric
information about the offender. There is no alternative to the reporting
of crime victimization to the police for insuring that both case-related
and offender-related biometric information is properly collected, analyzed,
and retained.
Biometric data on
offenders is critical, particularly given the high mobility of offenders
and the incredibly prolific criminal careers of ex-prisoners as shown
in our recidivism work. In addition, those matters which are brought to
the attention of police are the most serious in terms of injury and victim
consequences. Livescans today have reduced from days and weeks to minutes
the time required to do a record check-what is so central however is how
good that check is. What fraction of all records of some biometric characteristic
can actually be checked? For fingerprint-based systems, we estimate that
roughly 42 million of the 65 million prints of individuals can be accessed
in one way or another for a background check. But we do not know about
the coverage or accessibility of other types of biometrically-based records.
In 2000, BJS conducted
its second DNA Forensic Laboratory Survey. We identified 110 publicly
funded law enforcement laboratories in 47 States that perform DNA typing
and analysis. Since the first survey in 1997, the amount of work received
by these labs has grown by more than half-casework related samples increased
51% and convicted offender samples rose 53%. What is most heartening,
however, is that the amount of work completed for case samples increased
73% and work completed on offender samples grew a remarkable 231% over
the 3-year period of the study. In 2000, these DNA labs reported 893 employees,
a 33% increase since 1997. But even with the improvements in staffing
and productivity, nearly 16,000 casework samples were backlogged and more
than 265,000 convicted offender samples were still backlogged.
The findings from
labs meshes with the data from prosecutors offices nationwide. A 2001
BJS survey of local prosecutors revealed that about two-thirds of the
more than 2,100 local prosecutors offices nationwide had used DNA evidence
in a case during the preceding year. The most frequently cited problem
with the use of such evidence was not the improper collection or even
the inconclusiveness of the results—far and away the biggest problem in
prosecuting such cases was the "excessive " delay in receiving the results
from the DNA lab. This coincides with the data from labs about backlogs
of case and offender samples.
Aside from fingerprints
and DNA, BJS’s recurring Law Enforcement Management and Administrative
Statistics Program or LEMAS provides some data from police departments
nationwide which are relevant to assessing the extent to which facial
recognition, digital imaging and video cameras are in use by local law
enforcement.
A great deal of attention
has been focused on the ability to identify individuals using digital
imaging and video technology. Video cameras can make a record to identify
individuals and place them in a particular time-space context. Of particular
interest is the use of video cameras in concert with facial recognition
software that takes measurements from digital images and matches them
against a database of known or wanted offenders, either "real-time" or
during later review. An important question is the extent to which local
agencies are using these technologies and could potentially contribute
to national data resources. According to the LEMAS data, nearly half of
all local police departments, employing 61% of all local police officers,
regularly used video cameras in 2000. Overall, local police departments
had about 35,000 video cameras in operation.
About 3 in 10 local
police departments, employing 65% of all officers, regularly used digital
imaging for mug-shots. This included more than half of all agencies serving
10,000 or more residents, and about three-quarters of those serving 250,000
or more residents. Provided some level of standardization, these digital
images could contribute substantially to national data resources for facial
recognition. In addition, sixteen percent of local departments, including
roughly half of agencies serving 50,000 or more residents, used digital
imaging technology for suspect composites. Use of digital imaging in this
context increases the ability to quickly disseminate suspect composites
to other agencies, and may also contribute to facial recognition databases.
In summary, what
I am suggesting is that 3 things must occur for a credible biometric
identification system to exist whether based on fingerprints, DNA, facial
recognition, or retinal or iris scans:
- Victims have to
report their victimization so that relevant biometric evidence can be
gathered from the victim and crime scene and such evidence should also
be directly gathered from those persons who share a prior relationship
to the victim. A major effort to improve victim reporting to police will
stimulate demand for the collection of more biometric evidence and of
course the associated the burden of processing that evidence.
- Because of offender
mobility and commitment to the continuation of a criminal career, any
biometric collection program must be national in scope with the sharing
of information from both case samples and offender samples. Investigative
and background checks which can only access part of the potential repository
of biometric data will have limited value. A major contingency to accomplishing
this is not just the development of agreements to share and uniformly
format shared information, rather, the most significant problem is overcoming
the backlog of case and offender samples to be analyzed and typed and
to expedite the transmittal of accurate and complete information to be
used by those trying cases. There must be a timeliness built in to the
acquisition, analysis, and posting of biometric information to a central
database or repository.
- There must be
a predisposition among law enforcement at all levels of government that
multiple identification technologies should become as commonly used and
accessible as fingerprint records to identify individuals, regardless
of whether it is for a background check for employment or a criminal check
in connection with a pretrial release decision or the investigative determination
that the same criminal has committed crimes in multiple locations.
I have intentionally
not mentioned here the applications of these technologies to homeland
security issues. There are many here who are more knowledgeable than I
am about that set of applications and they will be speaking about these
concerns. I do think, however, there is much that we will learn from the
evolving homeland security challenges that will wind up with very practical
applications to the work of criminal justice agencies and vice-versa.
These are two components of public safety which clearly rely upon many
of the same methods and technologies.
I want to thank the
SEARCH staff and the BJS staff who have worked so hard to make this national
conference a reality. Again, let me thank you all for coming and please
feel free to grab BJS staff (please stand) who are here to talk about
our grant programs to assist States and localities to improve criminal
records systems nationwide and our program of criminal justice statistics.
SOURCES:
Langan, Patrick A.
and David J. Levin."Recidivism of
Prisoners Released in 1994". Washington, D.C.: Bureau of
Justice Statistics, June 2002, NCJ 193427.
Greenfeld, Lawrence
A. "Child Victimizers: Violent
Offenders and Their Victims". Washington, D.C.: Bureau of
Justice Statistics, March 1996, NCJ 158625.
Rennison, Callie
Marie. "Rape and Sexual Assault:
Reporting to Police and Medical Attention, 1992-2000". Washington,
D.C.: Bureau of Justice Statistics, August 2002, NCJ 194530.
Greenfeld, Lawrence
A and Peter Brien. "Improving Criminal
History Records for Background Checks". Washington, D.C.:
Bureau of Justice Statistics, February 2002, NCJ 192928.
Snyder, Howard N.
"Sexual Assault of Young Children
as Reported to Law Enforcement: Victim, Incident, and Offender Characteristics".
Washington, D.C.: Bureau of Justice Statistics, July 2000, NCJ 182990.
Steadman, Greg W.
"Survey of DNA Crime Laboratories,
1998". Washington, D.C.: Bureau of Justice Statistics, February
2000, NCJ 179104,.
Steadman, Greg W.
"Survey of DNA Crime Laboratories,
2001". Washington, D.C.: Bureau of Justice Statistics, January
2002, NCJ 191191.
DeFrances, Carol
J. "Prosecutors in State Courts,
2001". Washington, D.C.: Bureau of Justice Statistics, May
2002, NCJ 193441.
Hickman, Matthew
J. and Brian A. Reaves. "Local Police
Departments, 1999". Washington, D.C.: Bureau of Justice Statistics,
May 2001, NCJ 186478.
Reaves, Brian A.
and Matthew J. Hickman. "Sheriffs’
Offices, 1999". Washington, D.C.: Bureau of Justice Statistics,
May 2001, NCJ 186479.
Reaves, Brian A.
and Matthew J. Hickman. "Census
of State and Local Law Enforcement Agencies, 2000." Washington,
D.C.: Bureau of Justice Statistics, NCJ 194066.
OTHER RELEVANT SOURCES:
SEARCH Group, Inc.
"Public Attitudes Toward Uses of
Criminal History Information." Washington, D.C.: Bureau of
Justice Statistics, July 2001, NCJ 187663.
Bureau of Justice
Statistics. "National Criminal
History Improvement Program: Fiscal Year 2002 Program Announcement".
Washington, D.C.: Bureau of Justice Statistics, February 2002, NCJ 192811.
|