I am David R.
Loesch, Assistant Director in Charge of the Criminal Justice
Information Services Division (or CJIS Division) of the
Federal Bureau of Investigation. The FBI has been asked
to comment on proposed legislation which would authorize
nationwide name checks -- rather than fingerprint-based
checks -- of volunteers and we welcome this opportunity
to provide the Congress with the FBI's views on this matter.
Our position is predicated upon the extensive experience
and lengthy discussions by legal, policy, and technical
specialists within the FBI, representatives of state criminal
identification systems, and other federal agency information
specialists who both contribute and use Criminal History
Record Information (or CHRI). In light of this collective
wisdom, the FBI remains firmly opposed to name-based criminal
history checks for noncriminal justice purposes, including
that particular application envisioned by HR 3410.
Name checks are
often argued as a substitute for fingerprints because of
historical delays in processing fingerprint submissions.
In 1997 the CJIS Division made a concerted effort to reduce
its fingerprint backlog, which in turn, resulted in a corresponding
drop in the number of days necessary to process fingerprint
cards. The CJIS Division has eliminated its backlog, and
the time to process civil fingerprint cards at the CJIS
Division is presently eleven days.
Let me assure
you that the FBI is doing everything within its power to
continue to reduce its processing times. The Integrated
Automated Fingerprint Identification System (or IAFIS) became
operational in July 1999. This next generation system culminated
a decade-long effort costing approximately $640 million
to wed the speed of name checks with the certainty of fingerprints.
The IAFIS concept of operations detailed two hour criminal
and twenty-four hour noncriminal turnaround on electronically
submitted fingerprints, and it is operating within these
parameters.
The implementation
of the IAFIS created the reality of fingerprint images taken
at local law enforcement agencies by live-scan or card-scan
equipment and then electronically transmitted to the FBI
for processing by the FBI IAFIS. The IAFIS allows the FBI
to provide no slower than 24-hour turnaround for civil fingerprint
electronic submissions, providing the benefit of a prompt
turnaround akin to that associated with name checks but
with the certitude of a fingerprint check.
During the past
two months, representatives of the FBI have met with representatives
from the National Center for Missing and Exploited Children,
Boys and Girls Clubs, the Department of Housing and Urban
Development, Department of Justice (DOJ) Office of Policy
Development, as well as private employment associations
and state governments. There is consistent recognition of
the preferability of a fingerprint-based system, as well
as an appreciation that, although no single application
may warrant incurring the costs attending live scan technology,
aggregation of uses (and users) more than justify its application.
In short, we
are taking important steps to safeguard children and the
elderly by improving the speed of background checks, but
we are doing so without running the substantial risks posed
by name-based checks.
HR 3410 anticipates
that the proposed name checks would be performed using the
National Crime Information Center (or NCIC) Interstate Identification
Index (or III), a predominantly name-based, descriptor system.
A system designed to promptly provide CHRI to law enforcement,
it has operated since 1981 consistent with that intention.
The III is an automated system of criminal history records
accessed via the NCIC. It holds automated criminal history
records on approximately 36 million individuals, but does
not contain the manual criminal history records of approximately
5 million additional individuals. Name checks on these individuals
cannot be accomplished by performing an NCIC III check.
For almost twenty
years, NCIC III has been relied upon because law enforcement
"on the scene" had no fingerprint or positive
identification alternative. This "quick and dirty"
mechanism was relied upon with the understanding that a
law enforcement officer could determine whether the suspect
and the subject of the criminal record were the same by
relying upon his training and experience. This same logic
also supports the use of name-based checks in one unique,
broad-based system -- the FBI's national instant check system
(NICS) for proposed gun purchasers. The FBI's NICS examiners
-- like law enforcement officers -- have the training and
experience to evaluate name-based background checks correctly,
but the same is generally not true of others who would seek
to use name-based checks for non-criminal justice purposes.
Since some 94
percent of the records in III are the property of state
and local law enforcement, the CJIS Advisory Policy Board
(APB) was established pursuant to the Federal Advisory Committee
Act, 5 U.S.C. Appendix 2. Comprised of representatives of
federal, state, and local law enforcement agencies, the
CJIS APB provides advice to the FBI Director and the Attorney
General regarding proper use of CHRI. Regarding the instant
matter, CJIS APB and FBI policy require positive fingerprint
identification by the FBI or a State Identification Bureau
(SIB) before use of the III to obtain criminal history records
for noncriminal justice purposes.
The great weight
of the evidence supports the FBI and the CJIS APB's conclusion
that a name check of criminal history record systems is
a "rough" process which produces many "false
negatives" (in which a criminal is not identified)
and "false positives" (in which an individual
without a criminal record is identified as having a record).
It is only through the processing of fingerprints that one
can actually verify whether a criminal history record is
maintained on a particular individual.
Several examples
warrant specific mention.
- To fulfill the FBI's
obligations under the Government Performance and Results
Act of 1993, 31 U.S.C. §1116, in 1998 the CJIS
Division analyzed a statistically valid sample of the
6.9 million fingerprint cards submitted for employment
and licensing purposes during FY 1997. When compared
with the criminal prints on file at the FBI, some 8.7
percent, or approximately 600,000 of the fingerprints,
resulted in "hits." Of greatest importance,
we determined that some 11.7 percent of the hits, or
70,200 civil fingerprint cards, reflected entirely different
names than those listed in the applicants' criminal
history records and were only identified because of
the fingerprint submissions. Hence, the criminal history
records of these persons -- whom we deemed intentionally
provided false names to evade detection of their records
-- would have been missed entirely during the background
examination had the record check been name-based. Using
these projections, of the 52,000 applicants for school
system employment, some 5,900 or 11 percent who had
criminal histories would use names entirely different
from that listed on their criminal record. These intentionally
misleading applicants had prior convictions ranging
from assault to drug sales and were only detected because
of positive, fingerprint-based identification.
- Two independent investigations
by the Small Business Administration of individuals
participating in its loan program also support the conclusion
that the criminal element perceives names checks as
an opportunity to perpetrate fraud. The SBA studies
were performed because of the high rate of SBA loan
default and the SBA's suspicion that a significant percentage
was attributable to active and intentional borrower
fraud. In reviewing some 3,300 nonperforming loans,
it was determined that 11.6 percent of these borrowers
had concealed the existence of their criminal past and
had fraudulently certified that they did not have criminal
records. Spurred on by these figures, the SBA conducted
a second investigation of 500 defaulting loans and found
that some 8 percent of nonperforming loans were to borrowers
with intentionally undisclosed criminal records. Fingerprint
checks of these individuals would have thwarted most
of these schemes.
- In 1996, the state of
Ohio went to an "all fingerprint" system for
noncriminal justice purposes. In 1999, that state alone
performed more than 490,000 background checks, and determined
that it "caught" more than 9,000 individuals
with criminal records who would not have been identified
as criminals had name checks (and other descriptors)
been used in lieu of fingerprints. The significance
of this figure cannot be overstated, inasmuch as these
9,000 applicants knew that they were being fingerprinted
and that such fingerprints were going to be analyzed.
It is our opinion that the number would have been much
higher had these applicants been required to undergo
backgrounding predicated instead upon name and descriptor
check.
In addition to
the statistical evidence, various specific incidents warrant
your attention:
- The Department of Defense
(DOD), consistent with a report from the General Accounting
Office (GAO), decided to abandon its traditional practice
of limiting its inquiry to performing "simple"
name checks on those applicants for enlistment who fail
to reveal potentially disqualifying criminal backgrounds.
Its prior practice had contributed, in part, to an inordinate
rate of court martials involving enlistees with prior,
undisclosed, criminal records. Instead of fingerprinting
only those who reveal a potentially disqualifying criminal
history, the DOD is developing a system to fingerprint
all applicants. The GAO Report recommended:
The Secretary
of Defense [should] require all national agency checks
for enlistment into the military service to be based
on a full fingerprint search to (1) reduce the risks
associated with enlisting individuals who have been
convicted of the more serious misdemeanors and felonies
and (2) identify individuals who have used aliases.
- In 1998, New York City
Patrolman Anthony Mosomillo died because a fugitive
parolee in police custody on an unrelated charge was
released on bond by using fake identification to evade
detection during a name-based check. Once released,
he was able to assault and kill Officer Mosomillo. As
a result of that incident, New York City abandoned the
name check policy it had in effect since the mid-1970s
and now uses only fingerprint-based identification prior
to releasing arrestees on bond, explicitly recognizing
that name-based identification is not only untrustworthy,
but dangerous.
- Some states even go so
far as prohibiting placement of personnel until after
the results of a fingerprint-based criminal background
investigation are available. California, which experienced
firsthand the dangers of placing unidentified criminals
in positions wherein they could prey upon those most
at risk, expressly forbids the employment of applicants
during the pendency of their fingerprint-based check.
Section 45125 of the California Education Code mandates
that during the background investigation, "[i]f
a school district is notified by the (California) Department
of Justice that it cannot ascertain the required information
about a person, the school district may not employ that
person until the (California) Department of Justice
ascertains that information." This is not a unique
statutory provision or policy initiative.
Of course, the
preceding statistical and anecdotal information only tell
a part of the story, that of the "false negatives."
They fail to relate the consequences of the "false
positives:" those persons who are inaccurately and
unjustly identified as having a criminal record because
of similar descriptors.
Statistics maintained
by the FBI reflect that approximately 8.7 percent of the
6.9 million civil fingerprint submissions in FY 1997 were
matched with criminal records. Yet, when name check inquiries
of all types were made on NCIC's III for calendar year 1997,
an astonishing 39.34 percent of the 33,753,306 inquiries
were "hits." Because of FBI concern that the majority
of these inquiries were in response to what was perceived
as possible criminal activity (wherein recidivism might
upwardly skew the percentage), we removed those inquiries
from our analysis. The remaining 6,049,649 inquiries were
conducted on individuals for purposes such as national security
checks under the Security Clearance Information Act, 5 U.S.C.
§9101, and for criminal justice employment. Logically,
these populations should have no greater incidence of prior
criminality than the general population and, most likely,
reflect a smaller incidence of prior criminality than the
8.7 percent reflected by the general population screened
using fingerprint submissions. The results were nonetheless
astonishing.
Of these 6 million
noncriminal justice inquiries, 977,426, or some 16.2 percent
of the names checked, resulted in "hits". In light
of the 8.7 percent of civil applicants which are historically
identified by fingerprints with criminal records, we can
conclude that at least 7.5 percent (the 16.2 percent less
the 8.7 percent), are "false positives." To translate
such percentage into real terms, had name checks been made
of the 6.9 million civil applicants, some 517,500 "false
positives" would have resulted, and these individuals
would have been, at least temporarily, falsely identified
with criminal records. Stigmatization is the least of the
consequences of "false positives:" because of
the confidentiality typically associated with personnel
decisions and criminal history record information, an applicant
incorrectly identified may well not have an opportunity
to challenge the incorrect determination and denial of employment
or volunteer opportunities.
The validity
of this calculus was again confirmed by another CJIS Division
study. A statistical review of the first seven months of
FY 1998 disclosed that of the 3,038,497 civil fingerprint
cards processed, 382,063 individuals (or some 12.6 percent)
were tentatively identified as having criminal histories
by name checks. However, when fingerprint comparisons were
subsequently made, only 260,159 (or 68.1 percent) of those
tentatively identified were positively identified. The remaining
121,904, or some 31.9 percent of those tentatively identified,
were "false positives."
Recognizing the
forces which call for the instant legislation, in June 1998,
the Attorney General commissioned a task force consisting
of the Bureau of Justice Statistics, the Florida Department
of Law Enforcement, the Department of Housing and Urban
Development, and the FBI, to study the issue of name checks
for noncriminal justice purposes. In July 1999, the Interstate
Identification Index Name Check Efficacy: Report of the
National Task Force to the U.S. Attorney General (Name Check
Efficacy Study) was concluded. The Name Check Efficacy Study
rendered the following conclusions which support the FBI's
position:
- FBI fingerprint searches
are highly preferable to III name checks as a means
of criminal history screening. Individual fingerprint
patterns are known to be unique. For this reason, fingerprint
comparison is, and has for many decades been, the accepted
standard for establishing positive identification of
criminal history record subjects in the United States.
Modern automated fingerprint identification systems
are believed to produce identification error rates of
less than one percent. Compared to FBI fingerprint searches,
III name checks result in appreciable numbers of both
false positives and false negatives.
- In this regard, the study
suggests that if the 6.9 million civil applicant background
checks processed by the FBI in 1997 had been processed
by III name checks alone, approximately 346,500 false
positives and 70,200 false negatives would have resulted.
In addition, it is clear from the number of name checks
being conducted at the state level against state databases
that the absolute number of false positives and false
negatives would be significantly higher in a national
system that permitted III name checks in lieu of fingerprint
searches. The convenience of name checks would encourage
many more requests for national criminal record searches
than the 6.9 million submissions received by the FBI
in 1997 under current procedures requiring the obtaining
and submission of fingerprints.
- Since the prospect of
fingerprint-based criminal history record screening
deters persons with criminal histories from applying
for positions for which they are not fit, it can be
inferred that the incidence of false negatives documented
by the study of Florida civil applicants would be higher
in an environment in which name search-only background
checks were permitted.
- The Task Force acknowledges
that a number of state criminal history record repositories
have for many years used name checks as the sole method
of searching their state criminal history files for
a variety of noncriminal justice purposes. A primary
rationale for these uses of name checks is that fingerprint-based
searches commonly entail long mailing and processing
delays which are inconsistent with the needs of record
users with time-critical requirements. In addition,
the process of having one's fingerprints taken for background
screening purposes may be inconvenient or even impracticable
and usually involves the payment of a fee.
- The development and implementation
of automated fingerprint identification systems and
related technologies providing for the electronic capturing
and transmission of fingerprint images has made it possible
to dramatically reduce fingerprint transmission and
processing delays at both the state and federal levels.
Hence, the Name
Check Efficacy Study confirmed the FBI's position and the
position of law enforcement in general that fingerprints
should be required for noncriminal justice access to CHRI.
As many of you
are aware, the CJIS APB has repeatedly weighed in against
name checks for other than criminal justice purposes. Moreover,
at the June 1997 CJIS APB meeting in St. Petersburg, Attorney
General Reno, in speaking in support of Title I of the National
Criminal History Access and Privacy Act, explicitly affirmed
"the belief reflected in the Interstate Compact that
fingerprint-based [checks] (sic) are better than name-based
ones...." The Attorney General applauded the Interstate
Compact as "a sensible and an efficient approach to
the interstate exchange of criminal history information
for noncriminal justice purposes. The Interstate Compact
will require the certainty which comes from fingerprint-based
identification prior to the release of criminal history
information." As the Attorney General recognized, although
there are some time and cost savings associated with the
name-check system, its imperfection stands in stark contrast
"to the absolute accuracy and reliability associated
with fingerprint-based background checks.
Reliance on III
name checks alone as a method for criminal history record
screening can mean that, on a national basis, large numbers
of persons who do not have disqualifying criminal records
may be unfairly excluded. More importantly, large numbers
of persons may be permitted to volunteer in positions for
which they are unfit and in which they pose societal risks
because their criminal records are not discovered.
Recent federal
legislation also serves to reaffirm the commitment to fingerprints.
The National Crime Prevention and Privacy Compact, Title
II of Pub. L. 105-251 (the Compact), requires that "Subject
fingerprints or other approved forms of positive identification
shall be submitted with all requests for criminal history
record checks for noncriminal justice purposes" (Article
V[a]). Six states (Colorado, Florida, Georgia, Iowa, Montana
and Nevada) have adopted the Compact and its governing body,
the Compact Council, has met and is implementing the Compact.
In addition to
the Compact, recently enacted criminal backgrounding legislation
specifically requires fingerprints. Pub. L. 105-277 (concerning
nursing homes and home health care agencies) and the Volunteers
for Children Act (Section 221 and 222 of Pub. L. 105-251),
which amended the National Child Protection Act, impose
a fingerprint requirement for noncriminal justice access.
Numerous federal
statutes also authorize fingerprint-based access to FBI
CHRI for noncriminal justice purposes. See, e.g., 15 U.S.C.
§78q, concerning the securities industry; 7 U.S.C.
§21(b)(4)(e), concerning the Commodity Futures Trading
Commission; 10 U.S.C. §520a, concerning military recruiting;
42 U.S.C. §2169, concerning Nuclear Regulatory Commission
regulatees; 5 U.S.C. §9101, concerning the five Security
Clearance Information Act (SCIA) agencies; Pub. L. 100-413,
the Parimutuel Licensing Simplification Act (appearing at
note to 28 U.S.C. §534); 49 U.S.C. §44936, the
Aviation Security Improvement Act of 1990; 42 U.S.C. §13041,
concerning day care providers at federal facilities; 25
U.S.C. §3205-07, concerning child care workers on Indian
reservations; 25 U.S.C. §2701 et seq., the Indian Gaming
Regulatory Act; Pub. L. 105-277, concerning workers at nursing
homes and home health care agencies; and 42 U.S.C. §5119a,
the National Child Protection Act, as amended by the Volunteers
for Children Act.
In light of the significant impact upon the states which
contribute 94 percent of the CHRI in the FBI database, a
survey was conducted commencing January 27, 2000, which
indicated that only 10 of 50 states responding viewed favorably
a proposal which would expand name checks to the extent
envisioned by the instant legislation.
These are serious
and complicated areas for discussion, and the FBI appreciates
the opportunity to provide the Subcommittee with our views.
Be assured that the FBI is continuously concerned with improving
its service to both the criminal justice community and other
agencies authorized access to our system of records.