U.S. Customs Service: Better Targeting of Airline Passengers for Personal
Searches Could Produce Better Results (Letter Report, 03/17/2000,
GAO/GGD-00-38).

Pursuant to a congressional request, GAO reviewed policies and
procedures for conducting personal searches to determine the controls
Customs Service has in place to ensure that airline passengers are not
inappropriately selected or subjected to personal searches, focusing on
how the Customs' personal search data: (1) identifies the
characteristics--race and gender--of passengers who were more or less
likely to be subjected to more intrusive searches; and (2) the results
of searching those passengers.

GAO noted that: (1) inspectors select passengers for further examination
on the basis of Customs' policies and procedures and their professional
judgment and experience; (2) of those selected for further examination,
about 102,000 passengers were subjected to some form of personal search;
(3) of those 102,000 passengers, 95 percent were searched by inspectors
for contraband or hidden weapons by patting the passenger's clothed
body, 4 percent were strip-searched, and 1 percent were subjected to an
x-ray examination; (4) about 3 percent of frisked passengers had
positive results; 23 percent of the strip-searches were positive; and 31
percent of x-ray searches were positive; (5) searched passengers of
particular races and gender were more likely than other passengers to be
subjected to more intrusive types of personal searches after the initial
patdown; (6) however, in some cases those types of passengers who were
more likely to be subjected to more intrusive personal searches were not
as likely to be found carrying contraband; (7) White men and women and
Black women were more likely than Black men and Hispanic men and women
to be strip-searched rather than patted down or frisked, but they were
less likely to be found carrying contraband; (8) Black women were found
to be 9 times more likely than White women who were U.S. citizens to be
x-rayed after being frisked or patted down in fiscal year 1998; (9) on
the basis of x-ray results, Black women who were U.S. citizens were less
than half as likely to be found carrying contraband as White women who
were U.S. citizens; (10) patterns of selecting passengers for more
intrusive searches indicated that these searches sometimes resulted in
certain types of passengers being selected at rates that were not
consistent with the rates of finding contraband; (11) during GAO's
review, Customs developed new policies and procedures for personal
searches that include new requirements for supervisory review and
approval and procedures intended to ensure that passengers subjected to
personal searches know their rights; and (12) GAO identified four
management controls, such as training provided to inspectors and
supervisors on conducting personal searches and more systematic
evaluation of complaints, which Customs uses to help ensure that
inspectors use their search authority fairly and judiciously.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  GGD-00-38
     TITLE:  U.S. Customs Service: Better Targeting of Airline
	     Passengers for Personal Searches Could Produce Better
	     Results
      DATE:  03/17/2000
   SUBJECT:  Customs administration
	     Smuggling
	     Drug trafficking
	     Search and seizure
	     Contraband
	     International travel
	     Racial discrimination
	     Sex discrimination
	     Internal controls
IDENTIFIER:  John F. Kennedy International Airport (NY)
	     Miami International Airport (FL)
	     Chicago-O'Hare International Airport (Chicago, IL)
	     Los Angeles International Airport (CA)
	     Customs Service/INS Interagency Border Inspection System
	     Customs Service Advanced Passenger Information System
	     Federal Law Enforcement Training Center Criminal
	     Investigator Training Program
	     Customs Service Data Improvement Plan

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO report.  Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved.                 **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
** A printed copy of this report may be obtained from the GAO   **
** Document Distribution Center.  For further details, please   **
** send an e-mail message to:                                   **
**                                                              **
**                                            **
**                                                              **
** with the message 'info' in the body.                         **
******************************************************************

United States General Accounting Office
GAO

Report to the Honorable

Richard J. Durbin

United States Senate

March 2000

GAO/GGD-00-38

U.S. CUSTOMS SERVICE
Better Targeting of Airline Passengers for

Personal Searches Could Produce Better Results

Ordering Copies of GAO Reports
The first copy of each GAO report and testimony
is free. Additional copies are $2 each. Orders
should be sent to the following address,
accompanied by a check or money order made out
to the Superintendent of Documents, when
necessary. VISA and MasterCard credit cards are
accepted, also. Orders for 100 or more copies to
be mailed to a single address are discounted 25
percent.
Order by mail:
U.S. General Accounting Office
P.O. Box 37050
Washington, DC 20013
or visit:
Room 1100
700 4th St. NW (corner of 4th and G Sts. NW)
U.S. General Accounting Office
Washington, DC
Orders may also be placed by calling (202) 512-
6000 or by using fax number (202) 512-6061, or
TDD (202) 512-2537.
Each day, GAO issues a list of newly available
reports and testimony. To receive facsimile
copies of the daily list or any list from the
past 30 days, please call (202) 512-6000 using a
touch-tone phone. A recorded menu will provide
information on how to obtain these lists.
Viewing GAO Reports on the Internet
For information on how to access GAO reports on
the INTERNET, send e-mail message with "info" in
the body to:
[email protected]
or visit GAO's World Wide Web Home Page at:
http://www.gao.gov

Reporting Fraud, Waste, and Abuse in Federal
Programs
To contact GAO FraudNET use:
Web site:
http://www.gao.gov/fraudnet/fraudnet.htm
E-Mail: [email protected]
Telephone: 1-800-424-5454 (automated answering
system)

United States
General Accounting Office
Washington, D.C. 20548-0001
Official Business
Penalty for Private Use $300
Address Correction Requested

                    Bulk Rate
               Postage & Fees Paid
                       GAO
                 Permit No. G100

(264447)

Contents
Page 341GAO/GGD-00-38 Airline Passengers Selected for Personal
Searches
Letter                                                                      1
                                                                             
Appendix I                                                                 38
Objectives, Scope,
Methodology, and
Additional Results of
Analysis
                           Analysis of Customs Personal Search             38
                           Data
                           Customs' Policies and Procedures for            57
                           Conducting Personal Searches
                           Customs' Management Controls                    59
                                                                             
Appendix II                                                                61
Inspector Survey
                                                                             
Appendix III                                                               72
Supervisor Survey
                                                                             
Appendix IV                                                                84
Comments From the U.S.
Customs Service
                                                                             
Appendix V                                                                 87
GAO Contacts and Staff
Acknowledgments
                                                                             
Tables                     Table 1: Customs' Policies for                   5
                           Personal Search by Type, Suspicion
                           Level, and Approving Official
                           Table 2: Number and Percent of                  10
                           Passengers With Positive or Negative
                           Search Results by Type of Search,
                           Fiscal Years 1997 and 1998
                           Table 3: Differences Between U.S.               12
                           Citizen White and Minority Men and
                           Women in Being Strip-Searched, in
                           Fiscal Year 1998
                           Table 4: Differences Between U.S.               13
                           Citizen White and Minority Men and
                           Women in the Likelihood of Being X-
                           rayed, in Fiscal Year 1998
                           Table 5: Differences Between U.S.               14
                           Citizen White and Minority Men and
                           Women in the Likelihood of Strip-
                           searches Yielding Positive Results in
                           Fiscal Year 1998
                           Table 6: Differences Between U.S.               15
                           Citizen White and Minority Men and
                           Women in the Likelihood of X-ray
                           Searches Yielding Positive in Fiscal
                           Year 1998
                           Table 7: COMPEX Targeting Efficiency            28
                           Goals and Targeting Efficiency for
                           Fiscal Years 1997 and 1998
                           Table I.1: Number and Percentage of             40
                           Passengers Subjected to Personal
                           Searches in 1997 and 1998 by Gender,
                           Race, Fiscal Year, Citizenship
                           Status, Type of Search, and Search
                           Results
                           Table I.2: Number and Percent of                41
                           Passengers Subject to Types of
                           Searches by Fiscal Year, Gender,
                           Race, and Citizenship
                           Table I.3: Number and Percent of                42
                           Passengers With Positive or Negative
                           Search Results by Fiscal Year,
                           Gender, Race, Citizenship Status, and
                           Type of Search
                           Table I.4: Number and Percentage of             44
                           Frisks/Patdowns, Strip-Searches, and
                           X-ray Searches by Fiscal Year,
                           Citizenship, Race, and Gender
                           Table I.5: Number and Percentage of             45
                           Positive and Negative Searches for
                           Strip-Searches by Gender, Race,
                           Citizenship Status, and Fiscal Year
                           Table I.6: Number and Percentage of             46
                           Positive and Negative Searches for X-
                           ray by Gender, Race, Citizenship
                           Status, and Fiscal Year
                           Table I.7: Models Fitted Tables in              47
                           Which (1) Being Strip-Searched Versus
                           Nonintrusively Searched and (2) X-
                           rayed Versus Nonintrusively Searched
                           are Cross-Classified by Fiscal Year,
                           Citizenship, Gender, and Race
                           Table I.8: Models Fitted to Tables in           48
                           Which Search Outcomes are Cross-
                           Classified by Fiscal Year,
                           Citizenship, Gender, and Race for
                           Passengers (1) Strip-Searched and (2)
                           X-rayed.
                           Table I.9: Expected Frequencies Under           50
                           the Preferred Model in Which Type of
                           Search Is Cross-Classified by Fiscal
                           Year, Citizenship, Race, and Gender,
                           for Strip-searches versus
                           Nonintrusive Searches
                           Table I.10: Expected Frequencies Under          52
                           the Preferred Model for the Table in
                           Which Type of Search Is Cross-
                           Classified by Fiscal Year,
                           Citizenship, Race, and Gender, for X-
                           ray Searches versus Nonintrusive
                           Searches
                           Table I.11: Expected Frequencies Under          54
                           the Preferred Model for the Table in
                           Which Result of Search Is Cross-
                           Classified by Fiscal Year,
                           Citizenship, Race, and Gender, for
                           Strip-Searches Only
                           Table I.12: Expected Frequencies Under          56
                           the Preferred Model for the Table in
                           Which Result of Search Is Cross-
                           Classified by Fiscal Year,
                           Citizenship, Race, and Gender, for X-
                           ray Searches Only
                                                                             

Abbreviations

APIS      Advanced Passenger Information System
COMPEX    Compliance Examination Measurement
System
FLETC     Federal Law Enforcement Training Center
GPRA      Government Performance and Results Act
of 1993
MBM       monitored bowel movement
OPR       Office of Professional Responsibility
PAU       Passenger Analytical Unit
TECS      Treasury Enforcement Communications
System
APAS      Automated Passenger Analysis System

B-281000

Page 28GAO/GGD-00-38 Airline Passengers Selected f
or Personal Searches
B-281000

     March 17, 2000

The Honorable Richard J. Durbin
United States Senate

Dear Senator Durbin:

The Customs Service faces a major challenge in
effectively carrying out its drug interdiction and
trade enforcement missions while facilitating the
flow of cargo and persons into the United States.
To carry out its mission, Customs inspectors are
authorized to detain and search airline passengers
they suspect may be bringing contraband, such as
illegal drugs, into the country.  Concerns have
been raised about Customs' policies and procedures
for selecting or "targeting" passengers for
examinations and conducting personal searches,
including strip-searches and x-rays.

Recently some airline passengers have filed
lawsuits against the Customs Service claiming that
they had been singled out by inspectors for
examination on the basis of discriminatory
practices and were inappropriately subjected to
personal searches, particularly strip-searches.

Approximately 140 million passengers entered the
United States on international flights during the
two fiscal years 1997 and 1998.  From these
arriving international passengers, Customs
inspectors selected about 102,000 passengers for
some form of personal search.

You asked us to review Customs' policies and
procedures for conducting personal searches and to
determine the controls Customs has in place to
ensure that airline passengers are not
inappropriately selected or subjected to personal
searches.  Specifically, we agreed to provide
information and analysis on Customs' personal
search data to identify the characteristics-race
(White, Hispanic, Asian, or Black) and gender-of
passengers who were more or less likely to be
subjected to more intrusive searches (i.e., strip-
searched or x-rayed) and the results of searching
those passengers.  We also agreed to describe
Customs' current policies and procedures for
conducting personal searches and Customs'
management controls over the personal search
process.

     Our analysis of personal search data for
fiscal years 1997 and 1998 was limited to data
contained in Customs databases and therefore
focused on the passenger characteristics available
for 102,000 arriving international passengers whom
Customs subjected to some form of personal search.
As a result, we could not include in our analysis
any information about the remainder of the
approximately 140 million arriving international
passengers, including passengers who had only
their baggage searched.

Results in Brief
Inspectors select passengers for further
examination on the basis of Customs' policies and
procedures and their professional judgment and
experience. Of those selected for further
examination, about 102,000 passengers were
subjected to some form of personal search.  Of
those 102,000 passengers, 95 percent were searched
by inspectors for contraband (e.g., illegal drugs)
or hidden weapons by patting the passenger's
clothed body (commonly referred to as a frisk or
patdown);    4 percent were strip-searched; and 1
percent were subjected to an x-ray exam.  About 3
percent of the passengers only frisked or patted
down had positive results (i.e., contraband was
found); 23 percent of the strip-searches were
positive; and 31 percent of the x-ray searches
were positive.1

     Generally, searched passengers of particular
races and gender were more likely than other
passengers to be subjected to more intrusive types
of personal searches (being strip-searched or x-
rayed) after being subjected to frisks or
patdowns.  However, in some cases those types of
passengers who were more likely to be subjected to
more intrusive personal searches were not as
likely to be found carrying contraband.
Specifically, White men and women and Black women
were more likely than Black men and Hispanic men
and women to be strip-searched rather than patted
down or frisked, but they were less likely to be
found carrying contraband.  The most pronounced
difference occurred with Black women who were U.S.
citizens.  They were 9 times more likely than
White women who were U.S. citizens to be x-rayed
after being frisked or patted down in fiscal year
1998.  But on the basis of x-ray results, Black
women who were U.S. citizens were less than half
as likely to be found carrying contraband as White
women who were U.S. citizens.  Some patterns of
selecting passengers for more intrusive searches
indicated that these more intrusive searches
sometimes resulted in certain types of passengers
being selected for such searches at rates that
were not consistent with the rates of finding
contraband.

     We recommend that Customs compare the
characteristics of those passengers subjected to
personal searches with the results of those
searches to better target passengers carrying
contraband.

     During the course of our review, Customs
developed new policies and procedures for personal
searches that include new requirements for
supervisory review and approval and procedures
intended to ensure that passengers subjected to
personal searches know their rights. We identified
four management controls, such as training
provided to inspectors and supervisors on
conducting personal searches and more systematic
evaluation of complaints, that Customs uses to
help ensure that inspectors use their search
authority fairly and judiciously.  In conjunction
with improved data on the characteristics of those
passengers subjected to personal searches, as
recommended above, these policies and controls
could better safeguard the rights of U.S. citizens
and the traveling public.

Background
In addition to collecting revenue from
international trade, Customs' mission is to ensure
that goods and persons entering and exiting the
United States do so in compliance with U.S. laws
and regulations.  The Customs Service is
responsible for carrying out its trade enforcement
and drug enforcement missions and facilitating the
flow of cargo and persons into the United States.

Congress, in exercising its power to protect the
nation's borders, has given Customs broad
authority to conduct searches of persons and their
luggage.   Customs regulations provide that
persons, such as passengers coming into the United
States, are liable to inspection and a search by a
Customs officer for carrying contraband, such as
illegal drugs.

Customs' Search Authority
Routine border searches of persons entering the
United States that are conducted without a warrant
or probable cause in general have been held to be
reasonable by the Supreme Court simply because the
searches occur at the border, e.g., ports of
entry.  The Supreme Court has not defined a
"routine" search at ports of entry, but various
federal courts have made determinations on what
constitutes a routine border search.   Various
federal courts, in general, have found that
routine border searches require no suspicion or a
low level of suspicion.

The Supreme Court has indicated that strip, x-ray,
and body cavity searches would be considered
nonroutine but has specifically not expressed a
view on the level of suspicion required for such
searches.  Various federal courts, with some
exceptions, have held that nonroutine searches
require "reasonable suspicion" in order to be
justified.

Passenger Inspections and Personal Searches
In fiscal year 1998, approximately 71.5 million
passengers arrived at airports in the United
States on international flights.  When passenger
data are available, some passengers entering the
United States are subject to record checks
conducted by Customs inspectors prior to flight
arrival.  When passengers enter the United States,
at the control point or "primary" inspection, INS
inspectors are to perform record checks, ask
passengers questions about their travel, and let
them pass through the airport or refer them to
Customs for further examination.

Passengers referred for further examination are
sent to "secondary," where Customs inspectors can
subject passengers to additional questions and a
search of their luggage and personal belongings.
If an inspector suspects that a passenger may be
carrying or smuggling contraband, the inspector
may search a passenger's clothing and ultimately
subject the passenger to a personal search.
Personal searches range from a frisk or a patdown
to more intrusive searches that include strip-
searches, body cavity searches, x-rays, and
monitored bowel movements (MBMs).  Customs
inspectors carry out strip-searches.2  Customs
requires licensed medical personnel are required
by Customs to perform the latter three types of
searches at medical facilities.

According to Customs, the nature of the suspicion
determines the degree of search or detention
warranted.  In other words, the greater the
intrusion, the greater must be the reason for
conducting the search.  Less intrusive
searches-frisks or patdowns-require "some" or
"mere suspicion"3 by the inspector.  Customs
policy is that a patdown will be conducted only if
an officer has some or mere suspicion that
contraband is being concealed on the passenger.
In addition, more intrusive searches require that
an inspector have "reasonable suspicion"4 that a
passenger may be smuggling contraband.   Customs
instructs inspectors never to use a person's
gender, race, color, religion, or ethnic
background as a factor in determining any level of
suspicion.  Further, inspectors are required to
articulate their facts or reasons for conducting a
search and for each successive level of search and
to obtain approval from a supervisor or the port
director in advance of the search.  Supervisory
approval is not required for an immediate patdown
for weapons (frisk).  According to Customs'
Personal Search Handbook (Handbook),5 if the
officer cannot articulate such facts, then a
patdown, strip-search, x-ray, body cavity search,
or MBM detention should not be performed.  Table 1
lists the current search type; the reason (or
suspicion) for the search; and the official who
can authorize the search, if required.

Table 1: Customs' Policies for Personal Search by
Type, Suspicion Level, and Approving Official
Search type      Search type           Suspicion level    Approval required
                 definition            required
Immediate        A search necessary    Suspicion that a   No approval
patdown          to ensure that        weapon may         required.
(commonly        a person is not       be present.
referred to as a carrying a weapon.
frisk)
Patdown for      A search for          One articulable    On-duty supervisor.
merchandise      merchandise           fact.
                 (including
                 contraband) hidden
                 on a person's body.
Partial body     The removal of some   Reasonable         On-duty supervisor.
search           of the clothing by a  suspicion based
                 person to recover     on specific,
                 merchandise           articulable facts.
                 reasonably suspected
                 to be concealed on
                 the body.
X-ray            The use of a medical  Reasonable         Port director and
                 x-ray by              suspicion based    court order needed
                 medical personnel to  on specific,       without the consent
                 determine the         articulable facts. of suspect.  Only
                 presence of                             port director
                 merchandise within                      needed with the
                 the body.                               suspect's consent.
Body cavity      Any visual or         Reasonable         Port director and
                 physical intrusion    suspicion based    court order needed
                 into the rectal or    on specific,       without the consent
                 vaginal cavity.       articulable facts. of suspect.  Only
                                                        port director
                                                        needed with the
                                                        suspect's consent.
MBM              The detention of a    Reasonable         Port director.
                 person for the        suspicion based
                 purpose of            on specific,
                 determining whether   articulable facts.
                 contraband or other
                 merchandise is
                 concealed in the
                 alimentary canal.
Source: U.S. Customs Service.
     The Handbook provides direction and guidance
to its inspectors when they conduct each type of
personal search. The Handbook contains factors for
inspectors to consider when making decisions to
conduct a personal search.  The factors have been
grouped into six primary categories:

ï¿½    Behavioral analysis is the recognition of
physiological signs of nervousness.  Examples
include cold sweats, flushed face, and avoiding
eye contact.
ï¿½    Observational techniques involve recognizing
physical discrepancies in appearance.  Examples
include unexplained bulges in clothing and an
unnatural gait.
ï¿½    Inconsistencies are conflicts identified in
the interview/and or documentation.  Examples
include catching the person making a false
statement and an unreasonable explanation for
travel.
ï¿½    Intelligence is information developed by
another officer that may include information from
automated intelligence systems.
ï¿½    K-9 involves searches conducted in response
to an alert by a Customs canine.
ï¿½    Incident to a seizure or arrest involves
searches conducted in continuation of a seizure
action related to previously discovered illegal or
hidden merchandise.  For example, finding
contraband in a false-sided suitcase would provide
a basis for conducting a patdown to determine if
additional merchandise is hidden on the person.

Changes to Personal Search Requirements
In 1999, Customs directed reviews of personal
search policies and procedures to ensure that its
targeting practices did not discriminate on the
basis of race, color, ethnicity, religion or
gender.  As a result, Customs has implemented a
number of changes to its personal search policies
and procedures.  For example, since May 1999,
Customs has required supervisory approval for
patdown searches for merchandise, including
contraband.  Also, a Customs officer is to notify
the responsible U.S. Attorney in cases where
passengers are detained for more than 8 hours and
probable cause for arrest has not been developed.

Customs has also been testing new nonintrusive
technology to give airline passengers an option
for less intrusive searches.  At selected
airports, Customs is using body scan imaging
instead of patdowns.  This procedure, used only
with the passenger's consent, allows inspectors to
see if contraband is concealed under clothing.

Impetus for Changes to Customs Personal Search
Policy
The recent changes that Customs has made to its
personal search procedures were a result of
internal decisions and other studies.  According
to the Assistant Commissioner for Field
Operations, the basis for recent changes in
Customs' personal search policies and procedures
was to establish uniformity and standardization
throughout the ports for conducting personal
searches and to establish internal controls over
the process. He added that Customs needed
uniformity because various ports had different
procedures for conducting personal searches.

     In 1999, two groups reported on the need for
Customs to reassess its methods for targeting
passengers for personal searches.

ï¿½    In June 1999, Treasury's Office of
Professional Responsibility (OPR) issued a report
on Customs search policies and procedures that
included an assessment of Customs' methods for
"targeting" or selecting passengers for interview
and examination.  The objective of the OPR study
was to ensure that Customs' targeting policies
were not based on discriminatory factors.  The
study focused on targeting practices at the Port
of Miami (FL) and training provided at the Customs
Academy in Glynco, GA.  The report noted that
Customs' advanced passenger targeting is based
more on high-risk flights-from drug source or
transit countries-rather than information on
passengers. The report also found that "Customs
did not conduct routine reviews to ensure that the
criteria and targeting policy, including trend
analysis, are accurate and do not become
discriminatory."   The report cited New York's JFK
Airport and Miami Airport as large airports that
had used trend analysis as an important and
successful tool in the fight against drug
smuggling.  In fiscal years 1997 and 1998, Miami
and JFK were first or second in the country for
drug seizures.  The report cited that the drug
seizure rate was approximately 70 percent in
instances where a suspected internal drug carrier
was sent to the hospital for an x-ray.  The OPR
review team believed that trend analysis of drug
seizures is a useful tool and is nondiscriminatory
as long as articulated facts support the searches
and are not based on factors such as race or
gender.

OPR recommended that Customs "should continue to
conduct regular, periodic reviews to validate the
targeting criteria and process to assure that the
factors used in targeting continue to be
reasonable predictors of illegal activity, and to
ensure that targeting is not based on race, color,
ethnic origin, or gender."  OPR also recommended
that "Customs should also conduct periodic program
reviews to ensure that the techniques, criteria,
efficiency, and cost effectiveness of passenger
targeting are accomplishing the interdiction
mission."   In response to the OPR report, Customs
agreed with both recommendations and has stated
that it "will continue to conduct periodic reviews
of the targeting criteria and process, as well as
the program review to ensure that our mission is
accomplished."

ï¿½    In its July 1999 report, Customs' Passenger
Targeting Committee reported  that Customs'
passenger selectivity and targeting methodologies
require regular reassessment and updating.  The
Committee reported that ".the use of a standard
list of targeting criteria has little value in a
constantly changing travel environment."  The
Committee recommended that Customs eliminate the
use of the articulable factor list and emphasize
to inspectors that certain factors once thought to
be suspicious may no longer be relevant.  In
addition, the Committee recommended that Customs
pursue new data collection technology that will
provide a method to routinely assess whether
inspectors properly target individuals.
The Assistant Commissioner for Field Operations
said that Customs had developed the 43 factors
contained in Appendix C of the Handbook on the
basis of professional judgment, experience in the
field, and information from past seizures, and the
factors provided a compendium of successful drug
seizures from a historical perspective.   He said
that although Customs still believes that many
factors on the original list are viable reasons to
select (target) passengers for personal searches,
Customs has decided to eliminate the factors as a
result of the Passenger Processing Committee's
findings.  In place of the list, Customs has
included in the revised Handbook six primary
categories, which we previously discussed, for
inspectors to use to determine whether to conduct
a personal search.

Scope and Methodology
We agreed to provide information and analysis on
Customs' personal search data to identify race and
gender of passengers who were more or less likely
to be subjected to more intrusive searches (i.e.,
strip-searched or x-rayed) and the results of
searching those passengers.  To do this, we
analyzed Customs' personal search reports for air
passengers subjected to personal searches during
fiscal years 1997 and 1998.  Such searches were
done only after passengers were sent to secondary.

We also agreed to determine Customs' policies and
procedures for conducting personal searches and
how they are implemented at airports.   To do
this, we interviewed Customs headquarters
officials in the Office of Field Operations and
port officials at four airports-John Fitzgerald
Kennedy International Airport (JFK) in New York,
O'Hare International Airport in Chicago, Los
Angeles International Airport (LAX), and Miami
International Airport.  We also reviewed Customs'
program documents, including its Handbook.

Further, we agreed to describe Customs' management
controls over the personal search process.  We
interviewed key officials at Customs headquarters
and at the four airports.  Our review focused on
four internal controls intended to help ensure
that inspectors follow Customs' policies and
procedures:

ï¿½    training on targeting and personal search
procedures,
ï¿½    supervisory review and approval required for
most personal searches,
ï¿½    Customs' use of its program to measure the
efficiency of its targeting, and
ï¿½    a complaint process used to identify and
resolve passenger complaints.

To obtain information on Customs' (1) personal
search policies and procedures and how they are
implemented at airports and (2) management
controls over the personal search process, we also
conducted a national mail survey of 1,176
inspectors and 177 supervisors at 15 airports6
with the largest volume of international
passengers.  Specifically, we asked inspectors and
supervisors for their views about personal search
policies, procedures, and practices; Customs
training; and supervisory approval for personal
searches. The response rates for inspectors and
supervisors were 66 percent (775) and 74 percent
(131), respectively.  In reporting the results of
the survey, we report the number or percentage of
inspectors and supervisors who provided their
responses to a particular survey question.

See appendix I for a more detailed discussion of
our scope and methodology and appendices II and
III for mail survey results for inspectors and
supervisors.

     We conducted our work from August 1998
through January 2000 in accordance with generally
accepted government auditing standards.  We
requested comments on a draft of this report from
the Secretary of the Treasury.  The U.S. Customs
Service provided comments that are summarized at
the end of this letter; included in the report,
where appropriate; and reproduced in appendix IV.
Customs also provided technical comments that we
incorporated where appropriate.

Customs Subjected Certain Types of Passengers to
More Intrusive Searches Than Other Passengers, But
Those Searches May Have Had Fewer Positive Results
Approximately 140 million international passengers
entered the United States during fiscal years 1997
and 1998. From the entering international
passengers, Customs inspectors selected some
passengers for further examination.  Of those
selected for further examination, about 102,000
passengers were subjected to some form of personal
search. Our analysis was limited to data contained
in Customs databases and therefore focused on only
certain passenger characteristics of the 102,000
international arriving passengers who Customs
subjected to some form of personal search.  As a
result, we could not include in our analysis any
information about the remainder of the
approximately 140 million arriving international
passengers, including passengers who had only
their baggage searched.   In commenting on our
report,  Customs said that it did not dispute our
statistical analysis.  However, it also pointed
out that factors outside of the information
gathered during the personal search, such as
country of departure, may contribute to the
decision to carry out a more intrusive search.

Table 2 provides search results for 102,000
passengers who were subjected to some form of
personal search during fiscal years 1997 and 1998.

Table 2: Number and Percent of Passengers With
Positive or Negative Search Results by Type of
Search, Fiscal Years 1997 and 1998
                                        Negative      Positive      Total
                                                                   number
Search type                           Number Percen Number Percen         
                                                 t             t
Frisk/patdown                         93,764     97  3,005      3   96,769
Strip                                 2,989     77    883     23    3,872
X-ray                                   985     69    434     31    1,419
Body cavity                              14     44     18     56       32
Note: Fourteen incidents of monitored bowel
movements were excluded.
Source: GAO analysis of Customs' data.

     Generally, passengers of particular races and
gender were more likely than other passengers who
had been selected for further searches to be
subjected to more intrusive types of personal
searches (being strip-searched or x-rayed).
However, in some cases those types of passengers
who were more likely to be subjected to more
intrusive personal searches were not  more likely,
or even as likely, to be found carrying
contraband. The most pronounced difference
occurred with Black women who were U.S. citizens.
They were 9 times more likely than White women who
were U.S. citizens to be x-rayed after being
frisked or patted down in fiscal year 1998.  But
on the basis of x-ray results, Black women who
were U.S. citizens were less than half as likely
to be found carrying contraband as White women who
were U.S. citizens were.  Some patterns of
selecting passengers for more intrusive searches
indicated that these more intrusive searches
sometimes resulted in certain types of passengers
being selected for such searches at rates that
were not consistent with the rates of finding
contraband.

Gender and Race Were Strongly Related to the
Likelihood of Passengers Being More Intrusively
Searched and the Likelihood of Them Carrying
Contraband
To determine the types of characteristics that
influenced the likelihood of passengers who were
selected for further searches being subjected to
more intrusive personal searches and their
likelihood of being found carrying contraband, we
examined the effects of the four variables within
Customs' data.  These variables were gender; race
(White, Black, Hispanic, and Asian); citizenship
status (U.S. citizens or noncitizens); and fiscal
year (1997 and 1998). We measured the effect of
each of these variables on the likelihood of
passengers being intrusively searched and carrying
contraband, and we controlled for the effects of
the other variables.7

We found that gender and race were related to the
likelihood of passengers who were selected for
further searches being subjected to more intrusive
searches but not necessarily in the same way that
they were related to the likelihood that
passengers were found carrying contraband.
Generally, citizenship and fiscal year were not as
strongly related.  The following results focus on
the relationships between passengers' race and
gender for U.S. citizens in fiscal year 1998 and
the likelihood of being subjected to strip-
searches or x-ray examinations and the likelihood
of finding contraband wherever searched.  The
tables show the (1) gender ratios, which compare
the likelihood for women relative to men, within
race categories; and (2) race ratios, which
compare the likelihood for Blacks, Asians, and
Hispanics relative to Whites within gender
categories.

We note, however, where these effects differ for
noncitizens and in fiscal year 1997. We did not
include information on the origin of flights,
which inspectors told us is a consideration in
selecting passengers for searches, because this
factor was not collected and therefore not
included in Customs' database.

Appendix I contains a detailed discussion of the
effects of all four variables on passengers being
subjected to more intrusive searches and their
likelihood of being found carrying contraband.

Passengers With Certain Characteristics Are More
Likely to Be Subjected to More Intrusive Personal
Searches After Being Patted Down or Frisked
To describe how the likelihood of being strip-
searched or x-rayed was affected by passengers'
gender or race, we computed the likelihood of more
intrusive personal searches being conducted for U.
S. citizens by race and gender.  We then compared
the likelihoods across types of passengers to
determine how much more or less likely certain
types of passengers were to receive more intrusive
searches than other passengers. Table 3 shows the
likelihood of being strip-searched by race among
men and women who were U.S. citizens in fiscal
year 1998.

Table 3: Differences Between U.S. Citizen White
and Minority Men and Women in Being Strip-
Searched, in Fiscal Year 1998
Race                                Gender     Likelihood    Gender Race ratio
                                               of being     ratio
                                                 strip-
                                               searched
White                               Male          0.0821                    
White                               Female        0.0821      1.00          
Black                               Male          0.0506                0.62
Black                               Female        0.1421      2.81      1.73
Asian                               Male          0.0093                0.11
Asian                               Female        0.0260      2.81      0.32
Hispanic                            Male          0.0217                0.26
Hispanic                            Female        0.0611      2.81      0.74
Note:  The gender ratios compare the expected
likelihoods of being strip-searched for women
relative to men, within race categories.  The race
ratios compare the expected likelihoods of being
strip-searched for Blacks, Asians, and Hispanics
relative to Whites within gender categories.
Source: GAO analysis of Customs' data.

As shown in table 3, of those passengers who were
searched, Whites were more likely to be strip-
searched than all minority groups except Black
women; and among Whites, men and women were
equally likely to be strip-searched.  Among
minorities, however, there were large differences
between men and women in the likelihood of being
strip-searched as opposed to patted down or
frisked.  Black, Asian, and Hispanic women were
nearly 3 times as likely to be strip-searched,
after being patted down or frisked, as men of
those races; but White women and men were equally
likely to be strip-searched.  For example, the
likelihood of being strip-searched for Black men
was .0506; and the likelihood was .1421 for Black
women (.1421/.0506 equals 2.81, or nearly 3
times).8 Further, most minorities, especially
Asians (by factors of .11 for men and .32 for
women), tended to be strip-searched less often
than Whites, except for Black women, who were more
likely to be strip-searched than all other types
of passengers.

Moreover, in fiscal year 1997 Black women were
also more likely to be strip-searched than all
other women and men passengers.  This was not the
situation for noncitizens in 1998; among them,
White men and women were most likely to be
searched.

For U.S. citizens, table 4 shows the differences
between men and women of all races in the
likelihood of being x-rayed in fiscal year 1998.

Table 4: Differences Between U.S. Citizen White
and Minority Men and Women in the Likelihood of
Being X-rayed, in Fiscal Year 1998
                                    Gender     Likelihood    Gender Race ratio
Race                                           of being     ratio
                                                x-rayed
White                               Male          0.0053                    
White                               Female        0.0073      1.39          
Black                               Male          0.0457                8.65
Black                               Female        0.0635      1.39      8.65
Asian                               Male          0.0008                0.16
Asian                               Female        0.0012      1.39      0.16
Hispanic                            Male          0.0196                3.71
Hispanic                            Female        0.0272      1.39      3.71
Note:  The gender ratios compare the expected
likelihoods of being x-rayed for women relative to
men, within race categories.  The race ratios
compare the expected likelihoods of being x-rayed
for Blacks, Asians, and Hispanics relative to
Whites within gender categories.
Source:  GAO analysis of Customs' data.

Our analysis shows substantial differences across
races in the likelihood of being subjected to x-
ray searches but smaller differences between men
and women.  Among searched U.S. citizens of all
races in fiscal year 1998, women were 1.39 times
more likely than men were to be x-rayed.  Black
and Hispanic men and women were about 4 to 9 times
more likely than White men and women to be x-rayed
after being patted down or frisked.  The
likelihood of Asian men and women being x-rayed
was low relative to any other type of passengers.

Gender differences shown in table 4 were similar
for U.S. citizens in fiscal year 1997; however,
among searched noncitizens in both fiscal years,
women were less likely than men were to be x-
rayed.  The race differences noted in table 4 were
similar for U.S. citizens in fiscal year 1997.
Such differences were not as great for noncitizens
as for citizens.

Some Types of Passengers More Likely to be Subject
to More Intrusive Searches Are Not As Likely to be
Found Carrying Contraband
Although searched passengers with certain
characteristics were subject to more intrusive
searches, they were not always as likely to be
found carrying contraband.  Table 5 shows the
differences in the likelihood of strip-searches
yielding positive results for White and minority
men and women who were U.S. citizens in fiscal
year 1998. White and Asian men and women were less
likely than Black and Hispanic men and women to be
found carrying contraband when they were strip-
searched.  Among all races, especially Blacks,
women were less likely to be found carrying
contraband than men were.

Table 5: Differences Between U.S. Citizen White
and Minority Men and Women in the Likelihood of
Strip-searches Yielding Positive Results in Fiscal
Year 1998
Race                              Gender     Likelihood of    Gender    Race
                                           strip-searches     ratio   ratio
                                           being positive
White                             Male              0.2506                  
White                             Female            0.1947      0.78        
Black                             Male              0.6155              2.46
Black                             Female            0.2756      0.45    1.42
Asian                             Male              0.2506              1.00
Asian                             Female            0.1947      0.78    1.00
Hispanic                          Male              0.5880              2.35
Hispanic                          Female            0.4568      0.78    2.35
Note:  The gender ratios compare the expected
likelihoods of strip-searches being positive for
women relative to men, within race categories.
The race ratios compare the expected likelihoods
of strip-searches being positive for Blacks,
Asians, and Hispanics relative to Whites within
gender categories.
Source: GAO analysis of Customs' data.

Black women were about half (or .45 times) as
likely as Black men to be found carrying
contraband during strip searches; White, Hispanic,
and Asian women were about three-quarters (or .78
times) as likely to be found carrying contraband
as their male counterparts.  Blacks and Hispanics
were more likely than Whites to be found carrying
contraband; Asians were less likely.   The
differences between men and women and between
minorities and Whites shown in table 5 were
similar for U.S. citizens and noncitizens in both
years.

As noted earlier, the types of searched passengers
who had higher likelihoods of being strip-searched
were not always as likely to be found carrying
contraband. Comparing the likelihoods and
likelihood ratios in tables 3 and 5 shows that
Black women, who had the highest likelihood of
being strip-searched among U.S. citizens (.1421),
were half as likely to be found carrying
contraband during a strip-search as Black men (.45
times) and less likely than Hispanic men
(.2756/.5880 equals .47 times) and Hispanic women
(.2756/.4568 equals .60).   However, Black women
were 1.4 times as likely to be found carrying
contraband than White women.

Table 6 shows the differences in the likelihood of
x-ray searches yielding positive results for White
and minority men and women who were U.S. citizens
in fiscal year 1998.

Table 6: Differences Between U.S. Citizen White
and Minority Men and Women in the Likelihood of X-
ray Searches Yielding Positive in Fiscal Year 1998
Race                                Gender      Likelihood  Gender Race ratio
                                                of x-rays   ratio
                                                    being
                                                 positive
White                               Male            0.5869                  
White                               Female          0.5869    1.00          
Black                               Male            0.5869              1.00
Black                               Female          0.2816    0.48      0.48
Asian                               Male            0.3407              0.58
Asian                               Female          0.3407    1.00      0.58
Hispanic                            Male            0.3407              0.58
Hispanic                            Female          0.3407    1.00      0.58
Note:  The gender ratios compare the expected
likelihoods of x-ray searches being positive for
women relative to men, within race categories.
The race ratios compare the expected likelihoods
of x-ray searches being positive for Blacks,
Asians, and Hispanics relative to Whites within
gender categories.
Source:  GAO analysis of Customs data.

For all types of passengers who were subject to x-
rays, except Blacks, women were somewhat more
likely than men to be found carrying contraband.
Among Blacks who were x-rayed, women were about
half as likely as men to be found carrying
contraband. Apart from Black men, who were equally
as likely to be found carrying drugs as their
White counterparts, minorities were generally less
likely to be found carrying contraband than Whites
who were x-rayed.  Further, for those subjected to
x-ray searches, Black women were least likely to
be found with contraband.   The differences shown
for U.S. citizens in fiscal year 1998 were similar
for U.S. citizens and noncitizens in both years.
Again, some of the types of searched passengers
with a relatively high likelihood of being x-rayed
had a relatively low likelihood of being found
with contraband.  Hispanic women, for example,
were 4 times as likely to be x-rayed as White
women were, but they were about two thirds as
likely to have contraband found during an x-ray.

Customs Processes, Policies, and Procedures for
Targeting Passengers and Conducting Personal
Searches
     Customs inspectors rely on their professional
judgment and their experience, as well as guidance
and intelligence information, when deciding to
select or target passengers for interviews or
examinations that may include personal searches.9

     To understand the rationale for inspectors'
decisions to conduct personal searches, we
analyzed 70 randomly selected search reports. The
reports contained the inspectors' reasons for
subjecting passengers to patdowns.  They usually
provided more than one reason for conducting
patdowns.

     In our survey, few inspectors and supervisors
indicated that the Customs' award program
influenced their peers in referring or approving
requests for personal searches or in pursuing or
approving requests for more intrusive searches.

     As a result of evaluations of its personal
search procedures and on its own initiative,
Customs has implemented a number of changes.

Customs' Personal Search Guidance and Recent
Changes
     Inspectors use the Handbook, which provides
guidance and sets forth procedures for conducting
personal searches.  On the basis of internal
decisions and studies, Customs has revised its
personal search procedures.

Handbook
     At the time of our fieldwork (from March 1999
to June 1999), inspectors were following the March
1997 Handbook, which included in Appendix C a list
of 43 objective articulable facts that could lead
to a personal search.  This list comprised actions
(facts) that might lead an inspector to decide
that a personal search was warranted.10  Customs
had developed the 43 factors contained in Appendix
C on the basis of professional judgment,
experience in the field, and information from past
seizures.  According to a Customs official, the
factors provided a compendium of information about
drug seizures from a historical perspective.  The
September 1999 Handbook eliminated Appendix C and
now lists six primary categories for inspectors to
consider when making decisions about subjecting
passengers to personal searches. Customs officials
from the Passenger Programs Division told us that
the original 43 factors contained in the Handbook
were developed to assist inspectors on the basis
of previous successful drug seizures.

Changes to Customs Policy
During the course of our review, Customs developed
new policies and procedures for personal searches.
These policies include new requirements for
supervisory approval and review related to
personal searches, legal considerations, and
procedures intended to ensure that passengers
subjected to personal searches know their rights
and understand Customs' search policy.  Policy
changes include the following:

ï¿½    A supervisor must approve all patdown
searches, except immediate searches for weapons.
ï¿½    A port director must approve all searches
that require moving a person to a medical facility
for a medical examination.
ï¿½    Any person detained for 2 or more hours will
be able to have a Customs officer notify a person
of the passenger's choice about the delay.11

Studies
Information Sources for Inspectors to Use to
Select Passengers for Searches
     Customs inspectors can receive information on
potential drug smugglers from their Passenger
Analytical Units (PAUs), which are located at
international airports. PAUs are generally
composed of senior inspectors with access to law
enforcement databases that can assist them in
identifying the potential risk from incoming
flights and passengers to determine the level of
examination needed.12

PAUs identify or target flights and individual
passengers on the basis of their research of
information.  In addition, PAUs may develop their
own targets on the basis of information from past
seizures.

Inspectors also have access to information from
the Advanced Passenger Information System (APIS).13
APIS provides information on passengers destined
for the United States in advance of their arrival.
Information on passengers is processed through the
Interagency Border Inspection System, which
includes the combined databases of Customs, the
Immigration and Naturalization Service, the
Department of State, and about 21 other federal
agencies, along with access to the Federal Bureau
of Investigation's National Crime Information
Center. Using this system, inspectors have the
ability to check records to try to identify
terrorists, drug traffickers, and other violators.

Eighty-five percent of inspectors and 92 percent
of supervisors responding to our surveys said that
they relied on PAU data to assist them in
targeting passengers to be sent to secondary
examination from a moderate to very great extent.
This includes information developed by PAUs and
from automated systems.  Inspectors and
supervisors said that they also relied heavily on
information from seizure reports or information
from past seizures.

Inspectors' Reasons for Conducting Patdowns Varied
To determine what factors inspectors reported for
conducting personal searches, we randomly selected
70 search reports for fiscal years 1997 and 1998. 14
Our analysis showed that inspectors usually
provided more than one reason for conducting
patdowns.  In responding to our mail survey to
inspectors at the 15 largest airports, inspectors
said that they generally relied on a number of
factors in making their decisions to refer
passengers at secondary for personal searches.

In the 70 reports we reviewed, the inspectors
provided a total of 208 different reasons for
conducting patdowns.  The majority of the search
reports (45 of 70) provided 3 or more reasons for
performing a patdown.  The most common of the 208
reasons listed by inspectors for conducting
patdowns were:

ï¿½    behavior, such as exhibiting signs of
nervousness (16 percent);
ï¿½    arriving from a country known to be a source
of drugs (15 percent);
ï¿½    appearance (12 percent);
ï¿½    having a trip of short duration (7 percent);
ï¿½    wearing thick-soled or bulky shoes (6
percent); and
ï¿½    giving evasive answers to the inspector's
questions (6 percent).

When asked in our survey about the extent to which
inspectors used the five following factors in
their decisions to send passengers from primary to
secondary, inspectors generally agreed about the
following five factors in their decisions to send
passengers for personal searches.15  Most of the
inspectors responded that they consider to a
moderate, great, or very great extent:

ï¿½    general information e.g., a high-risk flight
or a flight from a high-risk country (85 percent);
ï¿½    specific information and intelligence about
an individual (93 percent);
ï¿½    observational and behavioral techniques (98
percent);
ï¿½    professional judgment based on past
experience (96 percent); and
ï¿½    interviews with individual passengers (97
percent).

Most Inspectors and Supervisors Do Not Believe
Awards Influenced the Numbers of Passengers
Selected for Searches
Customs has an awards program that provides
recognition to its staff for their work.  Of the
17 Customs' awards for its employees, 3 encourage
their efforts for searches that result in
subsequent seizures from passengers carrying
contraband.

ï¿½    The Commissioner may grant the Blue Eagle
Award for significant narcotics seizures.16
ï¿½    Local management officials, including Customs
Management Center directors, may confer the
Special Enforcement Actions Ribbon for a notable
seizure.
ï¿½    Local joint committees may confer Superior
Achievement Awards, which may include letters of
commendation for uncovering the existence of new,
unknown, or unique methods of smuggling.

 We surveyed inspectors and supervisors in order to
determine their opinions about the extent to which
Customs' awards system may have influenced their
personal search decisions.  Eighty-eight percent
of inspectors who responded to our survey
indicated that Customs' awards program influenced
inspectors to "some or little" or "no extent" in
referring passengers for personal searches.  In
addition, 89 percent of the responding inspectors
said that Customs' awards program influenced
inspectors to some, little, or no extent in
pursuing more intrusive searches.  Furthermore, 96
percent of the supervisors who responded to our
survey said that to no extent are supervisors more
likely to approve inspectors' requests to refer
passengers for personal searches because it would
improve the inspectors' chances of receiving an
award.   In addition, 97 percent of the responding
supervisors said that to no extent are supervisors
more likely to approve inspectors' requests to
pursue more intrusive personal searches because it
would improve the inspectors' chances of receiving
an award.

Customs' Management Controls for Personal Searches
We identified four management controls that
Customs uses to help ensure that inspectors use
their search authority fairly and judiciously and
to safeguard the rights of U.S. citizens and the
traveling public.

ï¿½    Training is provided to inspectors and
supervisors on how to select or target individuals
for interview and examination and how to conduct a
personal search.

ï¿½    Supervisory or port director approval is
required for all personal searches (except frisks)
to try to ensure that objective reasons for the
searches are articulated and approved in advance
and that the outcomes of searches are properly
documented.

ï¿½    A complaint process has been developed that
is intended to provide a system for responding to
all passenger complaints.

ï¿½    A program was established to measure the
effectiveness of passenger targeting for secondary
searches and passenger compliance with the laws
that Customs enforces, including drug laws.

Customs Provides Training on Personal Searches
That Most Inspectors and Supervisors Considered
Adequate
Customs is to provide basic training on personal
searches to all new inspectors and offers
additional courses to supplement this training.
The four ports that we visited also provided a
variety of training to inspectors on personal
search policy and procedures that included on-the-
job training.  These courses differed from port to
port.

Most inspectors (89 percent) and supervisors (87
percent) who responded to our survey questions
believed that classroom training they received on
personal searches was more than or generally
adequate. The majority of these inspectors (84
percent) and supervisors (88 percent) also
believed that nonclassroom training provided by
the ports, including on-the-job training, was more
than or generally adequate.

Recent internal and external studies of Customs'
personal search policies recommended changes to
improve Customs' personal search training.  Some
of these recommendations have already been
implemented.   In addition, in June 1999, a new
Assistant Commissioner was appointed to direct the
new Office of Training and Development to
centralize all of Customs' training, including its
personal search training.

Required and Optional Personal Search Training
     Customs requires all entry-level inspectors
to attend an 11-week course, "Customs Inspector
Basic Training," at the Federal Law Enforcement
Training Center (FLETC) in Glynco, Georgia.  Among
their courses, new inspectors receive training on
targeting passengers for interview and examination
and on conducting personal searches.  Training on
personal searches includes a Passenger Selectivity
course that covers interviewing techniques and
behavioral and observational techniques, as well
as courses on professionalism and Customs' search
authority.  According to Customs' Assistant
Commissioner for Training and Development,
mandatory training was not provided for air
passenger supervisors because they have received
personal search training as inspectors.

Most inspectors (93 percent) and supervisors (95
percent) who responded to our survey said that
they had received basic training.  Most of these
inspectors (89 percent) and supervisors (87
percent) believed the required basic training they
received on personal searches was more than or
generally adequate.

In addition to basic training, Customs provided
additional personal search training at FLETC and
at the ports.  Advanced training courses are
offered at FLETC, but these courses are optional.
Training provided by the four ports we visited
included personal search procedures, briefings on
recent drug seizures at airports, on-the-job
training, and updates on personal search policies.
This training may vary from port to port and is
not mandatory.   Fifty-three percent of the
inspectors believed that the required, additional,
and port training they received improved their
ability to identify passengers carrying contraband
at least to a moderate extent.

     In addition to the basic, additional, and
port training, Customs also offered specialized
training for inspectors who had been assigned to
Rover teams.17 Rover training was taught by Rover
supervisors and inspectors but offered only at JFK
and Miami airports.  Sixty-seven percent of the
inspectors who responded to our survey had taken
passenger enforcement Rover training, and 39
percent of the supervisors who responded to our
survey had received Rover enforcement training.

Recent Studies Recommended Changes to Customs'
Personal Search Training and Tracking System
     Treasury, Customs, and a private contractor
had recently completed reviews of Customs'
personal search policies and procedures.  Among
other things, they recommended ways for Customs to
improve personal search training that included (1)
developing new training courses, (2) reevaluating
existing training, and (3) establishing a system
for tracking all training provided to inspectors
and supervisors. Customs had agreed with these
recommendations and has begun to implement some of
them.

ï¿½    Treasury's June 1999 OPR study recommended
changes in personal search training and recording
training received by inspectors and supervisors.
OPR reported that Customs' Basic Training Course
and many of the advanced training courses did not
allocate sufficient time for training inspectors
and supervisors on issues related to
professionalism (e.g., diversity, sensitivity,
ethics, or interpersonal skills).  OPR recommended
that Customs work with FLETC to provide entry-
level instruction on professionalism and promote
professionalism in all job series.

Customs agreed with this recommendation. According
to Customs, it is currently conducting nationwide
professionalism training at all ports by requiring
all inspectors and supervisors to take five new
courses to supplement their basic training. 18  In
addition, the Commissioner has announced that
professionalism training is to be incorporated
into all aspects of training at  FLETC.

OPR also reported that local training was not
always captured in Customs' training database.
Consequently, there was no way to accurately
determine all of the training that inspectors
received at the ports.  OPR recommended that
Customs ensure that (1) all training is entered
into the training database, (2) training records
are reviewed annually to certify they are
accurately maintained, and (3) the Office of
Internal Affairs verifies and certifies training
records during inspections.  Customs agreed with
these recommendations and now requires that all
training be recorded in the training database.
Each port is responsible for the accurate data
entry and maintenance of the records.

Although the four ports we visited had recorded
basic required FLETC training in the training
database, additional courses provided at the ports
were not consistently reported in the training
database.

ï¿½    The March 1999 Passenger Processing Targeting
Committee reviewed how inspectors apply their
knowledge and training to target or select
passengers for inspection.  The Committee made
several recommendations.  For example, the
Committee recommended that Customs reassess the
validity of the Rover training conducted at the
JFK and Miami airports.   Specifically, it
recommended that skilled Rovers and trainers
review the current training to determine if it is
appropriate for all airports or should be adapted
for each port.  According to the Acting Director
of Passenger Operations, Customs has suspended
Rover Training taught at the JFK and Miami
airports.  Customs has engaged a contractor to
review its entire Rover training so that its
policies and procedures are more consistent among
the ports.

ï¿½    The Customs Service hired Boozï¿½Allen &
Hamilton to review air passenger interpersonal
communications that included two areas related to
training.  The contractor's report included three
recommendations related to training.
Specifically, it said that:

(1) Customs should clarify and standardize
policies and procedures to ensure that inspectors
are not confused about or unsure of personal
search policies and procedures.  The contractor
found that search procedures in the field
sometimes vary from what the inspectors were
taught at FLETC.  It recommended that Customs
revise the Handbook, deliver it to all inspectors,
and provide training on the new policy.  Customs
issued a revised Handbook in September 1999;
according to Customs officials, training was
provided to all staff involved with personal
search inspections.

(2) Customs should standardize communication to
passengers subjected to a personal search.  This
should include information about Customs' policy
and the personal search process and providing
courses on interpersonal communications for
supervisors and inspectors at the top 15 airports.
Customs had developed a new mandatory course in
Interpersonal Communications.

(3) Customs should make Rover training available
to all inspectors. It was recommended that Customs
develop an abbreviated course to be made available
to more inspectors or offer the course on video
for wider distribution.  Customs' Office of
Training and Development is reassessing its Rover
training; the Rover training taught at the JFK and
Miami airports has been suspended while the course
is being revised.

New Office of Training and Development
In June 1999, the Commissioner announced the
appointment of a new Assistant Commissioner to
direct a new Office of Training and Development.
The Office is expected to, among other things,
centralize all Customs training, ensure that
mandatory training courses are provided to
inspectors and supervisors, and ensure that all
training is recorded accurately in Customs'
training database.  The new Assistant Commissioner
is also to direct implementation of the
recommendations on training mentioned above.

Supervisory Approval Is Required for Personal
Searches
     According to Customs' Handbook, supervisory
approval is required for patdowns (except for
weapons) and strip-searches.  X-rays (both
consensual and involuntary), body cavity searches,
and MBM detentions prior to the search require
port director approval.  Supervisors and port
directors are expected to make an independent
assessment of the facts and decide whether the
inspector's reasons for the search warrant the
level of search requested.  Each progressive level
of search (e.g., after a strip-search, approving a
subsequent request for an x-ray) has to be
approved separately so that it is clear that the
reasons articulated for the additional searches
are justified.

     Supervisors and port directors do not have to
be present to authorize a personal search, nor do
they have to be present during the search.19  They
can authorize personal searches from a remote
location, including over the telephone or by
radio.  According to officials at the four
airports we visited, remote approval was generally
used only when there was a shortage of supervisors
available.

     Generally, supervisors, who have authorized
personal searches, whether results were positive
or negative, are to review and approve or
disapprove each search report20 prepared by the
inspecting officer.  Their approval is intended to
indicate that the search was approved in advance
and they were satisfied that the reasons provided
for the search were appropriate for the type of
search conducted.  When details are missing or
unclear, the supervisor will work with the officer
to correct the report.

     At the time of our initial fieldwork,
supervisory approval was not required to authorize
a patdown of a passenger.  However, officials at
two of the four airports we visited told us they
required supervisory approval for all patdowns.21

New Policy of Approval for Personal Searches
     During fiscal year 1999, Customs changed its
procedures.  For example, all patdowns are to be
approved in advance by a supervisor, unless there
is concern that a passenger may have a weapon.  In
addition, the port director must now approve all
searches that require moving a person to a medical
facility for a medical examination (i.e., body
cavity, x-ray, and MBM); and this authority cannot
be delegated unless the port director is away
(e.g., on leave) from the port.  Previously, a
supervisor's approval was required for these
requests.  In the event of a prolonged detention
(8 hours or longer) where probable cause that a
passenger has been carrying contraband has not
been developed, any further detention must be
reapproved by the port director.  The previous
policy was that supervisory approval was required
for a medical examination regardless of the length
of detention.

     Customs has also developed a Supervisor's
Check Sheet that is to be used to record all
negative personal searches.  Among other things,
the check sheet requires the supervisor to certify
that the search criteria had been reviewed and
found appropriate and that the supervisor had
reviewed the search criteria and results,
examination, and personal search process with the
inspector for lessons learned.  Positive searches
are to result in enforcement action and are
documented in seizure and arrest reports.

Supervisory Approval Is Required and, According to
Inspectors and Supervisors, Almost Always Obtained
for Personal Searches
     Most inspectors responding to our survey said
that at their respective ports, they are required
to obtain approval from their supervisors (or
seniors) prior to conducting personal searches.
Specifically, 97 percent responded that they
needed approval for patdown searches; and 99
percent responded that they needed approval for
all types of more intrusive searches, such as
strip-searches and x-rays.  Most inspectors
responding to our survey said they had not
conducted personal searches without supervisory
(or senior) approval within the last 6 months.
For example, over 99 percent of the inspectors
responded that they never, or rarely, conducted
strip-searches or x-rays without supervisory
approval.

     Most inspectors responded to our survey that
their requests for personal searches were usually
approved.  For example, inspectors reported that
93 percent of strip-searches requested were
approved, and 75 percent of x-rays requested were
approved.  Supervisors' responses regarding
approvals also indicated that they approved the
majority of search requests.  However, inspectors
and supervisors reported that fewer requests for x-
rays were approved than any other requests for
personal searches.

Customs Responds to Passenger Complaints, but Some
Discrepancies Exist Between the Passengers'
Complaints and Customs' Files
Customs recently revised and centralized its
system for processing and responding to all types
of passenger complaints, including personal search-
related complaints.  During fiscal year 1997,
Customs reported that the 22 largest international
airports combined received approximately 1
complaint for every 58,647 passengers processed.
At the 4 airports we visited, we reviewed all 42
passenger complaints about personal searches filed
during fiscal years 1997 and 1998.  Our review
showed that Customs responded to passenger
complaints, and it also  revealed problems with
the process.  Our analysis of the complaint files
showed that (1) the passenger's complaint and
Customs' record of the search sometimes differed,
and (2) Customs' files did not always contain
documentation on the resolution of the complaint.
Officials at the airports we visited said they did
not receive documentation on how complaints were
resolved, and Customs headquarters did not require
them to keep documentation at the port or at the
Customs Management Center on how cases were
resolved.

According to Customs officials, Customs has
assigned passenger service representatives at 17
of the largest international airports.  The
representative is responsible for dealing with the
passengers' complaints.  The representative is to
interact with the traveling public and try to
improve passenger relations.  In addition, the
representative is to assist travelers in
addressing any Customs-related questions or
concerns.  Until recently, the representative was
to respond in writing to passengers about their
complaints sent to the port.

In March 1999, Customs established a Customer
Satisfaction Unit in headquarters to centralize
and standardize the passenger complaint process.
This Unit is now responsible for reviewing and
responding to all passenger complaints.

Customs has also developed a complaint tracking
system, which is to record all complaints (verbal,
telephonic, and written) and their resolution at
airports around the country.  Customs will be able
to use the system to review and analyze all
complaints nationally and by airport.
Headquarters officials also said that a Complaint
Review Committee is to be established at the major
airports.  The Committee is to periodically
monitor and review the complaint system to
determine whether customer service standards are
being met and to determine whether training or
disciplinary measures are necessary in instances
when Customs' personnel do not meet standards.

Few Personal Search Complaints  Were Filed But
Differences Exist Between Complaints and Customs'
Files
Passenger complaints about personal searches at
the 4 selected airports, including those related
to strip-searches, totaled 42 in fiscal years 1997
and 1998.  Our analysis of these complaints
identified discrepancies in 10 of the 42 complaint
files reviewed.  We found differences between what
was reported by passengers and what was contained
in Customs' records. For example, in fiscal year
1997, one passenger complained that a woman
inspector told her to strip and she was searched
thoroughly.  Customs' record shows that the search
was a patdown, and the written response to the
passenger stated that the personal search she
experienced is referred to as a patdown. In
addition, Customs' files did not contain
documentation on whether discrepancies had been
resolved for any of the 10 cases.22

In fiscal years 1997 and 1998, 4 of the 42
complaints at the 4 ports alleging inspector
misconduct during a personal search were referred
to Customs' Office of Internal Affairs.  According
to airport officials, investigations were
completed, and airport management took the
appropriate action.  In each case, no action was
taken against the inspector because Customs found
no evidence of wrongdoing.

Customs Plans to Use COMPEX Data to Monitor Ports'
Targeting of Passengers
     Customs has developed a program intended to
measure passenger compliance with all the laws it
is responsible for enforcing, including drug laws.
The program, Compliance Examination (COMPEX)
measurement system, requires a port to randomly
examine a sample of passengers drawn from those
who were not targeted in routine passenger
processing.  These data are to be used to estimate
the number of violations that pass through the
port undetected.  COMPEX data are then combined
with the number of violations detected by routine,
targeted examinations to create an estimate of the
rate of violations that occurred among the overall
passenger population.

     COMPEX is designed to allow Customs to
measure the effectiveness of targeting by
comparing the results of targeted exams with the
estimated rate of violations in the overall
population of passengers.  The program was
established to comply with the Government
Performance and Results Act (GPRA),23 which
requires all federal agencies to measure how well
they achieve their missions.  COMPEX data are to
be used nationally to measure the effectiveness of
its targeting efforts in compliance with GPRA.  In
addition, ports are to use COMPEX data as a tool
for improving passenger processing and targeting
efficiency.24  Customs officials said that three of
the four airports we visited have been utilizing
COMPEX since fiscal year 1996.

     Customs' Strategic Plan for Fiscal Years 1997-
2002 states that under COMPEX, Customs had
established a national goal of identifying
passengers carrying contraband from targeted
inspections at a rate of 10 times more than
randomly selected passengers who are carrying
contraband.  This goal is to be achieved by fiscal
year 2002.  However, individual ports have
established interim goals.  Table 9 shows the
interim goals for the four ports we visited and
the ports' targeting efficiency.

Table 7: COMPEX Targeting Efficiency Goals and
Targeting Efficiency for Fiscal Years 1997 and
1998
Port                                Fiscal Port's targeting        Targeting
                                      year efficiency goal,   efficiency, as
                                            as established      measured by
                                           by headquarters     headquarters
LAX                                   1997     3x (better)            4.35x
                                      1998              5x           14.80x
JFK                                   1997              2x            1.46x
                                      1998              5x            3.30x
O'Hare                                1997             10x           14.21x
                                      1998             10x           13.12x
Miami                                 1997             10x            5.30x
                                      1998             10x            6.30x
Source: Customs, COMPEX program data.

     Officials at the four ports said that they
focus on achieving their port-specific goals.
Although headquarters has directed the ports to
use COMPEX data to improve passenger processing
and targeting efficiency, port officials said
headquarters has not provided specific guidance or
direction on how the ports should use COMPEX data.
Headquarters program officials said that they
informally reviewed COMPEX data on a monthly basis
to determine different targeting efficiency rates.
A headquarters program official said headquarters
expects the port directors to take the necessary
management steps to improve their targeting
efficiency rates without further direction from
headquarters.  However, the port officials from
the four ports we visited said that headquarters
has not taken any action on the basis of COMPEX
results.  In commenting on our draft report,
Customs disagreed that it had not provided
guidance to the ports on the use of COMPEX.
Customs pointed out that it had issued a
memorandum on August 7, 1997, to its ports and
management centers concerning the use of COMPEX.
At the time of our review, this memorandum was
neither mentioned nor provided to us by port or
management center officials.

Customs Has Begun to Address COMPEX Issues
Customs officials said they have taken steps to
address issues related to COMPEX and to assess how
the ports are using COMPEX data.

ï¿½    Customs officials created a Passenger Data
Analysis Team in June 1999.  This unit is to
review the targeting practices at airports and to
analyze personal search data.  A Customs official
has stated that Customs expected this unit to
begin analyzing COMPEX data by December 1999. As
of November 1999, this unit had three full-time
employees.

ï¿½    Customs developed a Data Improvement Plan to
address problems, such as missing data and data-
entry errors.  The Plan requires that the port
personnel who are responsible for entering COMPEX
data are to assess the accuracy of data input.
After assessing the data, the port personnel who
entered the data will have 60 days to draft an
improvement plan to correct any data inaccuracies.
According to a headquarters program official, the
port directors have sufficient management skills
to resolve specific problems related to inaccurate
data entry.

ï¿½    Customs has designed a COMPEX survey for use
at most major airports. The survey includes a
variety of questions related to COMPEX, including
how the data are being collected and entered into
the system.  As of November 1999, Customs
officials said they had administered the survey at
Newark, Atlanta (Hartsfield), Miami, and Chicago's
O'Hare airports.   Customs program officials told
us they plan to continue to administer the survey
on an ad hoc basis, but they have not set a
deadline for the completion of the survey or a
time frame for the final analysis of the survey
results.

Conclusions
     Recent litigation has called into question
Customs' personal search procedures that initially
target airline passengers for examination.  As a
result of internal decisions and studies, Customs
revised its personal search practices and
procedures.  In response to OPR's recommendation,
Customs agreed to conduct periodic reviews to
validate the targeting criteria and process to
ensure that (1) the factors used in targeting are
reasonable predictors of illegal activity; and (2)
targeting is not based on race, color, ethnic
origin, or gender.  Customs also agreed to conduct
periodic program evaluations, reviewing the
techniques, criteria, efficiency, and cost
effectiveness of passenger targeting.

     We analyzed personal search data that Customs
had in its databases.  These databases included
only certain passenger data on the 102,000
international arriving passengers whom Customs
subjected to some form of personal search.
Therefore, we could not include in our analysis
information about the characteristics of the
millions of other arriving international
passengers not subject to personal searches.

Our analysis showed that generally, searched
passengers of particular races and gender were
more likely than other passengers to be subjected
to more intrusive types of personal searches
(being strip-searched or x-rayed).  However, some
types of passengers who were more likely to be
subjected to more intrusive personal searches were
not always as likely to be found carrying drugs or
other contraband. For example, White men and women
and Black women were more likely than Black men
and Hispanic men and women to be strip-searched
rather than patted down or frisked, but they were
less likely to be found carrying contraband.  In
fiscal year 1998, Black women who were U.S.
citizens were 9 times more likely to be x-rayed
than White women who were U.S. citizens.  But x-
ray results indicated that Black women who were
U.S. citizens were less than half as likely to be
found carrying contraband as White women who were
U.S. citizens were.

     Customs has not analyzed passenger data to
determine the relationships between the
characteristics of those passengers selected for
personal searches and the results of those
personal searches.  By doing such an analysis,
Customs would be able to determine if the rate at
which inspectors target passengers with certain
characteristics is similar to rates at which those
passengers are found to be carrying contraband.
This information could help Customs ensure
compliance with its policies and procedures to
better target passengers and improve its targeting
efficiency.

Recommendations
     We recommend that Customs analyze the
characteristics of passengers selected for
intrusive searches and the results of those
searches as part of the periodic evaluation it has
agreed to do on the basis of OPR's recommendation.
It should use these data to help to develop
criteria for determining which passengers to
search.

Agency Comments and Our Evaluation
     On February 29, 2000, Customs provided
comments on the draft report.  Customs said that
they were pleased that GAO did not find a pattern
indicating that it had engaged in discriminatory
selection of passengers for personal searches on
the basis of race, gender, or ethnic origin.  It
would be more accurate to say that we did not
address the causes of any patterns in the data
related to selection of passengers for more
intrusive searches and so did not assess the issue
of discrimination in relation to passenger
selection and the search process.

     Customs said that it agreed with our
recommendation and has already taken several
actions to accomplish this objective.  Some of
these actions related to data collection and data
analysis are:

ï¿½    With respect to data collection, Customs said
that it has taken steps to collect more complete
and accurate data on persons subjected to personal
searches.  For example, race, gender, and
citizenship used to be optional fields in its
automated records. Data entry of these elements is
now mandatory.  In addition, other new mandatory
data have been added, such as the port where the
flight arrived from and where the traveler
initiated his or her trip.

ï¿½    With respect to data analysis, Customs said
that it is closely monitoring data on personal
searches.  In May 1999, it established a Passenger
Data Analysis Team that began publishing weekly
reports with data by race and gender. The report
shows the results of Customs personal searches.

 Customs added that due to its increased management
oversight, improved guidance in the form of the
new personal search handbook, and recent training
to Customs officers authorized to conduct personal
searches, it is now achieving improved results.
For example, Customs cited that its positive
search results have increased from 3.5 percent in
fiscal year 1998 to 5.73 percent in fiscal year
1999.

     As agreed with your office, unless you
publicly announce its contents earlier, we plan no
additional distribution of this report until 30
days from its issue date.  At that time, we will
send copies to Senator William V. Roth, Senator
Daniel Patrick Moynihan, Senator Ted Stevens,
Senator Robert C. Byrd, Senator Fred Thompson,
Senator Joseph Lieberman, Senator Ben Nighthorse
Campbell, Senator Byron Dorgan, Senator Susan
Collins, Senator Carl Levin, Senator Phil Gramm,
Senator Paul S. Sarbanes, Senator Michael B. Enzi,
Senator Tim Johnson, Representative C.W. Bill
Young, Representative David R. Obey,
Representative Bill Archer, Representative Charles
B. Rangel, Representative Philip M. Crane,
Representative Sander M. Levin, Representative Amo
Houghton, Representative William J. Coyne,
Representative Jim Kolbe, and Representative Steny
H. Hoyer in their capacities as Chairman or
Ranking Minority Member of Senate or House
Committees and Subcommittees.  We are also sending
copies of this report to the Honorable Jacob J.
Lew, Director, OMB; the Honorable Lawrence H.
Summers, Secretary of the Treasury; and the
Honorable Raymond Kelly, Commissioner of Customs.
We will also make copies available to others on
request.

     The major contributors to this report are
acknowledged in appendix V.  If you or your staff
have any questions concerning this report, please
contact me or James M. Blume, Assistant Director,
on (202) 512-8777.

     Sincerely yours,

 Laurie E. Ekstrand
Director
Administration of Justice Issues

_______________________________
1 Generally, Customs performs patdowns before
conducting x-rays and strip-searches.
2 Inspectors of the same gender as the passenger
are to conduct strip-searches.  The revised
Personal Search Handbook now refers to these types
of searches as partial body searches.
3 According to Customs, "some" or "mere suspicion"
is the minimal level of suspicion required to
conduct a patdown search.  By policy, Customs
requires that inspectors are to articulate to a
supervisor at least one fact (e.g., inconsistent
answers to inspector's questions) before
conducting a patdown.
4According to Customs, "reasonable suspicion" is
more than some or mere suspicion and is based on
specific, articulable facts.  These facts, when
taken together with reasonable inferences from
these facts, would lead a reasonable officer
(inspector) to suspect that a person might have
merchandise contrary to law.
5Customs issued the Handbook in March 1997 and
revised it in September 1999.  The revisions were
not in place when we did our analysis.
6 The 15 airports serve the cities of Atlanta;
Boston; Chicago; Dallas-Ft. Worth; Detroit;
Washington, D.C., metropolitan area; Honolulu;
Houston; Los Angeles; Miami; New York; Newark;
Orlando; San Francisco; and San Juan.
7  We used multivariate logistic regression and
loglinear models to eliminate differences between
groups that were due to chance or random
variation.  Tables 3 through 6 provide likelihoods
(or odds ratios) which are derived from these
models and which provide a simple description of
the differences across the groups of passengers we
compare.  See Appendix I for a fuller discussion
of these models and the likelihoods derived from
them.  Our multivariate results are on the basis
of the 67,553 cases (66%) with no missing data.
We did, however, repeat the analysis including
cases that were coded as missing and found that
the overall results with respect to the effects of
sex, citizenship, and year were similar.
8 The likelihood (or odds) represents the number
of passengers strip-searched compared to the
number of passengers just frisked or patted down.
For example, the likelihood of being strip-
searched for Black men was .0506, which indicates
that there were 5 Black men strip-searched for
every 100 who were frisked or patted down.  In
tables 3 through 6, the gender ratios given
reflect the likelihood for women relative to men,
within race categories.  The race ratios reflect
the likelihood for Blacks, Asians, and Hispanics
relative to Whites, within categories of gender.
Appendix I contains the frequencies and calculated
likelihoods and ratios for each type of search and
search result by year, citizenship status, race,
and gender.
9 Customs provides training to its inspectors on
conducting personal searches.  Training is
discussed in the next section.
10 The list had been in existence for over 10
years.
11 Passengers carrying drugs would not be permitted
to have a Customs officer notify a person about
the delay.
12 Information is also provided to Rovers, which
are teams of specially trained inspectors who
target passengers on high-risk flights on the
basis of observing passengers' behavior, data from
PAUs, and previous drug smuggling trends.
13 Airlines and foreign governments participating
in APIS collect passengers' biographical data,
including name and date of birth, at the time of
the flight's departure and transmit it to Customs
Data Center which performs checks against federal
law enforcement databases while the flight is en
route to the United States.
14 The sample was selected to identify the reasons
inspectors provided for conducting patdowns, but
it should not be projected to the universe of
85,008 patdowns conducted in fiscal years 1997 and
1998.
15 We identified these five factors on our own and
validated them through pretesting our survey.
16 A "significant" seizure is defined as $1 million
in cash, 1,000 pounds of cocaine, 100 pounds of
opium, 5,000 pounds or marijuana or hashish, or 50
pounds of heroin.
17 The Rover teams carry out proactive inspections
by observing and interviewing passengers before
the passengers present their declaration of goods
being brought into the United States to an
inspector at the control point.  Their training
includes training similar to that provided to
other inspectors, such as passenger targeting,
post-seizure analysis, and observational and
interviewing techniques.  The purpose of the Rover
training is specifically to enhance the teams'
ability to target and identify passengers who may
be carrying contraband and examine passengers more
effectively.
18 The five courses are Interpersonal
Communications, Cultural Interaction, Passenger
Enforcement Selectivity Training, Confrontation
Management/ Verbal Judo, and Customs' Personal
Search Policy.
19 The Handbook provides that when possible,
approvals should be granted in person.
20 A search report documents that a personal search
has been performed.  It includes the type of
search, reason(s) for conducting the search, and
the results of the search.  Positive searches
(resulting in enforcement action) are reported on
the Search, Arrest & Seizure Report.  Negative
searches are reported on the Inspectional
Operations Incident Log.
21Beginning in May 1999, Customs required that all
patdown searches must have supervisory approval in
advance of the search.
22 The files did not provide any additional
information regarding the complaint.  Therefore,
we could not determine whether the passenger or
Customs was correct.
23 P.L. 103-62 (1993).
24 Targeting efficiency is a comparison of the
results of targeting passengers for selection to
the results of randomly selecting passengers for
searches.

Appendix I
Objectives, Scope, Methodology, and Additional
Results of Analysis
Page 39GAO/GGD-00-38 Airline Passengers Selected f
or Personal Searches
We were asked to review Customs' policies and
procedures for conducting personal searches and to
determine the controls in place to ensure that
airline passengers are not inappropriately
selected or subjected to intrusive personal
searches. Specifically, we agreed to provide
information and analysis on (1) Customs' personal
search data to identify the characteristics-race
(White, Hispanic, Asian, or Black) and gender-of
passengers who were more likely to be subjected to
intrusive searches (i.e., strip-searched or x-
rayed) and the results of searching those
passengers; (2) Customs' policies and procedures
for conducting personal searches and how they are
implemented at airports; and (3) Customs'
management controls over the personal search
process.

Analysis of Customs Personal Search Data
Customs inspectors are to document all incidents
and personal searches, recording basic information
about the passenger and a brief narrative
describing the search and/or seizure. Data from
these documents are to be entered into one of two
centralized databases. The first is the Treasury
Enforcement Communications System (TECS) Search,
Arrest, and Seizure report, which documents
positive incidences resulting from either a
personal search or baggage search or both (i.e.,
contraband was found in a passenger's luggage
resulting in a seizure and/or arrest, but no
merchandise was found on the passenger's body).
The second is the TECS Incident Log, which records
all negative personal searches. We used data from
both sources for fiscal years 1997 and 1998 for
our analysis.

We received data on the types of searches
passengers were subjected to; the results of the
searches; and the passengers' races, genders, and
citizenship status from both fiscal years. We did
not see the need to independently verify the
accuracy or completeness of the Customs data
because Customs has taken steps to ensure complete
passenger information for each incident (e.g.,
requiring information on origin of flight).

We received information on the passengers' genders
and races-White, Black, Hispanic, and Asian.1 We
also received the citizenship status for each
passenger when it was available. For purposes of
our analysis, we collapsed the information into
two categories to contrast U.S. citizens and
noncitizens. About 20 percent of the cases were
missing data on race, and 20 percent were missing
data on citizenship status.

One of our objectives was to examine information
on personal searches of passengers conducted by
Customs in fiscal years 1997 and 1998 and
determine whether the fiscal year and gender,
race, or citizenship status of the passengers
searched affected the likelihood of whether they
were searched intrusively (x-rayed or strip-
searched) rather than nonintrusively (frisked or
patted down).2 Our analysis focused on only those
passengers subjected to personal searches. Our
analysis did not include all passengers who
entered the country or those who only had their
baggage searched. We also considered whether the
more intrusive searches conducted were positive or
negative to determine whether or not the more
intrusive searches of certain types of passengers
resulted in their having similar likelihoods of
yielding positive results.

In conducting our analysis, we first looked at the
percentages of searched passengers in both fiscal
years 1997 and 1998 for each gender, race, and
citizenship group that were patted down or
frisked, strip-searched, and x-rayed.  We also
looked at the percentages in each group that were
found to be concealing contraband. We then
considered these four factors simultaneously and
used both loglinear and logistic regression models
to determine the effects of these factors on the
likelihood of being strip-searched as opposed to
patted down or frisked and of being x-rayed as
opposed to patted down or frisked. The models,
which are described below, allowed us to determine
the effects of each factor independently of one
another and to determine whether they interacted
with one another (i.e., whether the effect of
gender on the likelihood of being strip-searched
as opposed to patted down was different for
minorities from the likelihood for Whites, or for
citizens versus noncitizens).

Table I.1 shows the numbers and percentages of
searched individuals overall by fiscal year,
gender, race, citizenship, and the type and
outcome of the search that was conducted. Slightly
more searches were conducted in 1998 than in 1997,
and overall men outnumbered women among the
persons who were searched by roughly 3 to 1
(76,342 men compared to 24,431 women). The fact
that information on citizenship and race is
missing for about 20 percent of the individuals
who were searched makes it somewhat more difficult
for us to reliably estimate the relative numbers
of citizens and noncitizens and the relative
numbers of persons of each race who were searched.
Among those individuals for whom we have data,
noncitizens substantially outnumbered citizens
among persons searched. There were more Hispanics
searched than there were Whites; and Whites
outnumbered Blacks and Asians among individuals
searched by about 2 to 1 (25,634 Whites to 12,777
Blacks) and about 3 to 1 (25,634 Whites to 8,450
Asians), respectively.

Table I.1: Number and Percentage of Passengers
Subjected to Personal Searches in 1997 and 1998 by
Gender, Race, Fiscal Year, Citizenship Status,
Type of Search, and Search Results
                                 Number     Percent
Fiscal year   1997               49,624          49
              1998               52,468          51
                                                   
Gender        Female             24,431          24
              Male               76,342          75
              Missing data        1,319           1
                                                   
Race          White              25,634          25
              Black              12,777          13
              Asian               8,450           8
              Hispanic           31,397          31
              Native                 79           0
             American
              Missing data       23,755          23
                                                   
Citizenship   U.S. citizen       30,004          29
              Noncitizen         50,819          50
              Missing data       21,269          21
                                                   
Type of       Frisk/patdow       96,769          95
search        n
              Strip               3,872           4
              X-ray               1,419           1
              Body cavity            32           0
                                                   
Search        Negative           97,752          96
results
              Positive             4340           4
Note 1:  All percentages are calculated on the
basis of the total number of passengers subjected
to personal searches (102,092).
Note 2:  Some passengers who were x-rayed may have
been strip-searched as well.  All passengers who
were strip-searched and x-rayed were first either
patted down or frisked.
Source: GAO analysis of U.S. Customs data.

Table I.2 shows the numbers and percentages of
individuals in each fiscal year that were
subjected to the various types of searches. The
more intrusive searches occurred with the same
relative frequency in fiscal years 1997 and 1998.

Table I.2: Number and Percent of Passengers
Subject to Types of Searches by Fiscal Year,
Gender, Race, and Citizenship
         Nonintrusiv       Intrusive         Total
              e                             number
            Frisk      Strip-      X-ray   
           /patdown    Search
         Numbe Perce Numbe Perce Numbe Perce 
             r    nt    r    nt    r    nt
Fiscal                                            
year
1997     47,13    95 1,775     4  696     1  49,605
             4
1998     49,63    95 2,097     4  723     1  52,455
             5
                                                  
Gender                                            
Female   22,71    93 1,334     5  364     1  24,414
             6
Male     72,97    96 2,338     3 1015     1  76,329
             6
Missing  1,077    82  200    15   40     3   1,317
data
                                                  
Race                                              
White    23,64    92 1,820     7  154     1  25,618
             4
Black    11,49    90  785     6  490     4  12,769
             4
Asian    8,304    98  138     2    7     0   8,449
Hispanic 30,26    96  569     2  565     2  31,395
             1
Native      72    91    7     9    0     0      79
American
Missing  22,99    97  553     2  203     1  23,750
data         4
                                                  
Citizens                                          
hip
U.S.     28,01    93 1,562     5  421     1  29,994
             1
Noncitiz 48,42    95 1,557     3  824     2  50,802
en           1
Missing  20,33    96  753     4  174     1  21,264
data         7
Note:  All percentages are calculated on the basis
of the total number of passengers subjected to
personal searches (102,060), excluding body cavity
searches.
Source: GAO analysis of U.S. Customs data.

The percentages of passengers being intrusively
searched, as opposed to nonintrusively searched,
appeared to be similar for those persons for whom
information on citizenship and race was missing
and those for whom such information was available.
However, the 1,317 persons for whom information on
gender was missing were considerably more likely
than others to have been strip-searched or x-
rayed. For example, 15 percent of those searched
passengers for whom information on gender was
missing were strip-searched, but 5 percent of
those known to be women and 3 percent of those
known to be men were strip-searched.

Table I.3 shows that positive search results
occurred in about 4 percent of all searches for
both men and women. Searches of Blacks and Whites
were more likely to yield positive findings than
searches of Hispanics and Asians, and searches of
citizens were more often positive than searches of
noncitizens. There were sizable differences in the
percentages of nonintrusive and intrusive searches
that yielded positive results. X-ray results were
positive 31 percent of the time, positive strip-
searches occurred 23 percent of the time, and
positive frisks and patdowns occurred 3 percent of
the time. These latter differences in the
percentages of positive results by type of search
may account for some or all of the differences in
the percentages of positive results across race
and citizenship categories.  As table I.2 showed,
there were differences across race and citizenship
categories in the percentages subjected to the
different types of searches.

Table I.3: Number and Percent of Passengers With
Positive or Negative Search Results by Fiscal
Year, Gender, Race, Citizenship Status, and Type
of Search
               Negative      Positive        Total
                                            Number
            Number Percen Number Percen 
                        t             t
Fiscal year                                       
1997        47,406     96  2,218      4     49,624
1998        50,346     96  2,122      4     52,468
                                                  
Gender                                            
Female      23,445     96    986      4     24,431
Male        73,370     96  2,972      4     76,342
Missing        937     71    382     29      1,319
data
                                                  
Race                                              
White       23,950     93  1,684      7     25,634
Black       11,905     93    872      7     12,777
Asian        8,168     97    282      3      8,450
Hispanic    30,423     97    974      3     31,397
Native          76     96      3      4         79
American
Missing     23,230     98    525      2     23,755
data
                                                  
Citizenship                                       
U.S.        28,259     94  1,745      6     30,004
Noncitizen  49,044     97  1,775      3     50,819
Missing     20,449     96    820      4     21,269
data
                                                  
Type of                                           
search
Frisk/patdo 93,764     97  3,005      3     96,769
wn
Strip        2,989     77    883     23      3,872
X-ray          985     69    434     31      1,419
Body cavity     14     44     18     56         32
Note:  All percentages are calculated on the basis
of the total number of passengers subjected to
personal searches (102,092).
Source: GAO analysis of U.S. Customs data.

Tables I.1, I.2, and I.3 suggest that a number of
characteristics that we examined are related to
passengers being intrusively searched and to
searches being positive. However, the differences
are potentially misleading because they involve
looking at the effect of each characteristic
independently of the other characteristics.
Because we were concerned that the effects of
certain characteristics may account for or
interact with others, we cross-classified the
types and outcomes of searches by each of the four
characteristics (fiscal year, gender, citizenship,
and race) simultaneously and, using multivariate
techniques, estimated the effect of each
characteristic controlling for every other
characteristic.3

Table I.4 shows the numbers and percentages of
searched persons who were patted down or frisked,
strip-searched, and x-rayed for each of the
categories of individuals who were cross-
classified by gender, race, and citizenship in
each fiscal year. Tables I.5 and I.6 show the
actual numbers and percentages of persons searched
for whom positive search outcomes resulted, again
for each of the categories of individuals who were
cross-classified by gender, race, and citizenship
in each fiscal year, for persons strip-searched
(table I.5) and x-rayed (table I.6).

Table I.4: Number and Percentage of
Frisks/Patdowns, Strip-Searches, and X-ray
Searches by Fiscal Year, Citizenship, Race, and
Gender
Fisc Citizens  Race   Gender               Type of search                Total
al   hip                                                              number
year status
                                Frisk /     Strip-search     X-ray     
                             Patdown
                             Number Percen Number Percen Number Percen 
                                      t             t             t
1997 Noncitiz  White  Male    2,567     93    188      7     18      1   2,773
     en        White
    
                      Female    732     93     42      5      9      1     783
               Black  Male    1,717     92     67      4     77      4   1,861
             
                      Female    913     88    100     10     29      3   1,042
               Asian  Male    1,672     98     25      1      2      0   1,699
             
                      Female    625     97     22      3      0      0     647
               Hispan Male    8,089     97     99      1    174      2   8,362
             ic
             
                      Female  2,209     95     76      3     35      2   2,320
     U.S.      White  Male    4,592     94    275      6     24      0   4,891
              
                      Female  1,369     93     98      7     12      1   1,479
               Black  Male    1,176     92     58      5     47      4   1,281
             
                      Female    738     82    106     12     54      6     898
               Asian  Male      876     99      8      1      1      0     885
             
                      Female    338     97      8      2      1      0     347
               Hispan Male    1,989     95     51      2     45      2   2,085
             ic
             
                      Female    548     87     67     11     15      2     630
1998 Noncitiz  White  Male    3,079     89    344     10     36      1   3,459
     en        
    
                      Female    698     88     86     11      7      1     791
               Black  Male    2,106     92     76      3    115      5   2,297
             
                      Female  1,005     89     86      8     38      3   1,129
               Asian  Male    1,584     99     22      1      0      0   1,606
             
                      Female    564     97     18      3      1      0     583
               Hispan Male    9,258     97    101      1    160      2   9,519
             ic
             
                      Female  2,217     95     75      3     41      2   2,333
     U.S.      White  Male    5,440     92    435      7     22      0   5,897
              
                      Female  1,352     92    100      7     15      1   1,467
               Black  Male    1,294     89     94      6     61      4   1,449
             
                      Female    719     85     96     11     33      4     848
               Asian  Male    1,067     99     11      1      0      0   1,078
             
                      Female    340     97     10      3      0      0     350
               Hispan Male    2,178     99     21      1      1      0   2,200
             ic
             
                      Female    484     87     27      5     43      8     554
Note:  These totals are based on the 67,553 cases
with no missing data.  These totals also exclude
body cavity searches and Native Americans.
Source: GAO analysis of U.S. Customs data.

Table I.5: Number and Percentage of Positive and
Negative Searches for Strip-Searches by Gender,
Race, Citizenship Status, and Fiscal Year
Fisca Citizen  Race   Gender             Results for strip-searches
l     ship
year  status
                                   Negative           Positive           Total
                                                                     number
                                Number   Percent   Number   Percent 
1997  Nonciti  White  Male         157        84       31        16        188
      zen
     
                      Female        31        74       11        26         42
               Black  Male          49        73       18        27         67
                      Female        88        88       12        12        100
               Asian  Male          23        92        2         8         25
                      Female        18        82        4        18         22
               Hispan Male          60        61       39        39         99
             ic
                      Female        61        80       15        20         76
      U.S.     White  Male         203        74       72        26        275
     
                      Female        77        79       21        21         98
               Black  Male          31        53       27        47         58
                      Female        77        73       29        27        106
               Asian  Male           8       100        0         0          8
                      Female         8       100        0         0          8
               Hispan Male          26        51       25        49         51
             ic
                      Female        46        69       21        31         67
1998  Nonciti  White  Male         296        86       48        14        344
      zen
     
                      Female        73        85       13        15         86
               Black  Male          47        62       29        38         76
                      Female        74        86       12        14         86
               Asian  Male          17        77        5        23         22
                      Female        16        89        2        11         18
               Hispan Male          72        71       29        29        101
             ic
                      Female        57        76       18        24         75
      U.S.     White  Male         355        82       80        18        435
     
                      Female        87        87       13        13        100
               Black  Male          62        66       32        34         94
                      Female        69        72       27        28         96
               Asian  Male          10        91        1         9         11
                      Female         7        70        3        30         10
               Hispan Male          16        76        5        24         21
             ic
                      Female        17        63       10        37         27
Note:  These totals are based on the 67,553 cases
with no missing data.  These totals exclude body
cavity searches and Native Americans.
Source: GAO analysis of U.S. Customs data.

Table I.6: Number and Percentage of Positive and
Negative Searches for X-ray by Gender, Race,
Citizenship Status, and Fiscal Year
Fisca Citizen  Race                      Results for x-ray searches
l     ship
year  status
                                   Negative           Positive           Total
                                Number   Percent   Number   Percent 
1997  Nonciti  White  Male           9        50        9        50         18
      zen      
     
                      Female         4        44        5        56          9
               Black  Male          38        49       39        51         77
             
                      Female        23        79        6        21         29
               Asian  Male           2       100        0         0          2
             
                      Female         0         0        0         0          0
               Hispan Male         118        68       56        32        174
             ic
             
                      Female        22        63       13        37         35
      U.S.     White  Male          18        75        6        25         24
              
                      Female         4        33        8        67         12
               Black  Male          27        57       20        43         47
             
                      Female        39        72       15        28         54
               Asian  Male           1       100        0         0          1
             
                      Female         1       100        0         0          1
               Hispan Male          32        71       13        29         45
             ic
             
                      Female         9        60        6        40         15
1998  Nonciti  White  Male          24        67       12        33         36
     zen      
     
                      Female         5        71        2        29          7
               Black  Male          71        62       44        38        115
             
                      Female        27        71       11        29         38
               Asian  Male           0         0        0         0          0
             
                      Female         1       100        0         0          1
               Hispan Male         120        75       40        25        160
             ic
             
                      Female        31        76       10        24         41
      U.S.     White  Male          17        77        5        23         22
              
                      Female         9        60        6        40         15
               Black  Male          38        62       23        38         61
             
                      Female        26        79        7        21         33
               Asian  Male           0         0        0         0          0
             
                      Female         1       100        0         0          1
               Hispan Male          32        74       11        26         43
             ic
             
                      Female         6        60        4        40         10
Source: GAO analysis of U.S. Customs data.

To distinguish the statistically significant
effects from insignificant effects in these
tables, we fit a series of models, which varied in
terms of the direct and interactive effects they
included.4 The top panel of table I.7 shows the
models fitted to table I.4 when x-rays are
excluded and the numbers of persons strip-searched
are compared to the numbers patted down or
frisked. The bottom panel of table I.7 shows the
models fitted to table I.4 when strip-searches are
excluded and the numbers of persons x-rayed are
compared to the numbers patted down or frisked.
The two panels in table I.8 show models fitted to
the two tables in which positive and negative
search outcomes are contrasted for strip-searches
and for x-rays.

The models we fit were logistic regression models
that allowed the four characteristics (fiscal
year, gender, citizenship, and race) to be related
to one another but varied in terms of the main and
interaction effects of those characteristics on
whether searches were intrusive and whether
intrusive searches were positive. In each of these
analyses we fit and compared a series of four
models.

ï¿½    Model 1 is a baseline model of independence.
ï¿½    Model 2 is a main effects model, which
includes all direct effects of all independent
variables (e.g., race, gender, fiscal years,
citizenship).
ï¿½    Model 3 includes all two-way interactions
between factors in addition to the main effects
included in Model 2.
ï¿½    Model 4, which eliminates the insignificant
variables (p>.05), is the preferred (final) model
for each table.  Model 4, in all analyses, fits
the data acceptably (p<.10) and/or accounted for
the large bulk (60%) of the associations present
in the data.

Table I.7: Models Fitted Tables in Which (1) Being
Strip-Searched Versus Nonintrusively Searched and
(2) X-rayed Versus Nonintrusively Searched are
Cross-Classified by Fiscal Year, Citizenship,
Gender, and Race
Marginals/effects fitted                      Likelihood    df    P Explained
                                                  ratio            variation
                                             chi-square                   a
Strip searches versus nonintrusive searches
Model 1        {YCSR} {T}                         1686.6    31 <.001       0%
Model 2        {YCSR} {YT}{CT}{RT}{ST}             355.8    25 <.001      79%
Model 3        {YCSR}                               37.2    13 <.001      98%
              {YCT}{YRT}{YST}{CST}{CRT}{SRT}
Model 4        {YCSR}{YT}{CT}{RT}{YRWT}{CRB,H       62.7    22 <.001      96%
              T}{SRWT}

 X-ray searches versus nonintrusive searches
Model 1        {YCSR} {T}                          733.3    31 <.001       0%
Model 2        {YCSR} {YT}{CT}{RT}{ST}              59.6    25 <.001      92%
Model 3        {YCSR}                               17.6    13  .17      98%
              {YCT}{YRT}{YST}{CST}{CRT}{SRT}
Model 4        {YCSR}{RT}{ST}{CST}{CRWT}            32.8    25  .14      96%
a The percentage of explained variation refers to
the amount of variability in the 32 different
observed likelihoods in each table that is
accounted for by the model.
Note: Bracketed letters refer to variables that
are associated with one another under the
differing models: Y = Fiscal year, C =
Citizenship, S = Gender, and R = Race. Subscripts
denote the particular categories of race that are
involved in the interactions under the differing
models: W = White, H = Hispanic, B = Black. T= the
outcome of being subject to either a strip or x-
ray search.
Source: GAO analysis of U.S. Customs data.

Table I.8: Models Fitted to Tables in Which Search
Outcomes are Cross-Classified by Fiscal Year,
Citizenship, Gender, and Race for Passengers (1)
Strip-Searched and (2) X-rayed.
Marginals/effects fitted                      Likelihood    df    P Explained
                                                  ratio            variation
                                             chi-square                   a
Positive versus negative results for strip-searches
Model 1         {YCSR} {T}                         138.0    31 <.001       0%
Model 2         {YCSR} {YT}{CT}{RT}{ST}             46.8    25 .005      66%
Model 3         {YCSR}                              22.6    13 .047      84%
               {YCT}{YRT}{YST}{CST}{CRT}{SRT}
Model 4         {YCSR}{YT}{CT}{RT}{ST}{SRbT}        40.6    25 .026      71%

Positive versus negative results for x-ray searches
Model 1         {YCSR} {T}                          43.5    31 .067       0%
Model 2         {YCSR} {YT}{CT}{RT}{ST}             25.4    25 .440      42%
Model 3         {YCSR}                               5.7    13 .956      87%
               (YCT}{YRT}{YST}{CST}{CRT}{SRT}
Model 4         {YCSR}{YT}{RT}{SRbT}                17.7    28 .934      60%
a The percentage of explained variation refers to
the amount of variability in the 32 different
observed likelihoods in each table that is
accounted for by the model.
Note: Bracketed letters refer to variables that
are associated with one another under the
differing models: Y = Fiscal year, C =
Citizenship, S = Gender, and R = Race. Subscripts
denote the particular categories of race that are
involved in the interactions under the differing
models: W = White, H = Hispanic, B = Black. T= the
outcome of the search being positive for finding
contraband.
Source: GAO analysis of U.S. Customs data.

The Likelihood of Being Strip-Searched
The model chosen as the preferred model to
describe the effects of fiscal year, race, gender,
and citizenship on the likelihood of being strip-
searched as opposed to simply frisked or patted
down is one that involves the effects of all
factors, some of which are interacting. The
effects of the four factors can be seen in the
ratios in table I.9 and are described as follows:

ï¿½    The effect of fiscal year (or the change from
1997 to 1998) is different for minorities and
Whites. Among searched minorities, the likelihood
of being strip-searched declined slightly by a
factor of .87. For Whites, however, the likelihood
of being strip-searched increased between 1997 and
1998 by a factor of 1.40. This was true for both
men and women and for citizens as well as
noncitizens.

ï¿½    The effect of citizenship is different for
Blacks and Hispanics from the effects for Whites
and Asians. Among Whites and Asians, citizens are
less likely than noncitizens to be strip-searched,
by a factor of 0.76; among Blacks and Hispanics,
citizens are more likely than noncitizens to be
strip-searched, by factors of 1.51 and 1.97,
respectively.
ï¿½    Race interacts with fiscal year and
citizenship. In general, though, minorities are
less likely to be strip-searched than Whites.
Among citizens, however, Black women were nearly 3
times as likely as White women (2.78) to have been
strip-searched as opposed to nonintrusively
searched in 1997, and they were nearly twice as
likely (1.73) to have been strip-searched in 1998.
ï¿½    Gender has an effect for minorities, but no
effect for Whites. Among all three minority
groups, women were 2.81 times as likely as men to
be strip-searched. Among Whites, however, men and
women were equally likely to be strip-searched.
This gender effect was the same for both citizens
and noncitizens in both fiscal years.

Table I.9: Expected Frequencies Under the
Preferred Model in Which Type of Search Is Cross-
Classified by Fiscal Year, Citizenship, Race, and
Gender, for Strip-searches versus Nonintrusive
Searches
Fisc Citizen  Race  Gender      Strip-searches      Gende   Race Citizen       
al   ship                 (expected frequencies)      r  ratio   ship       
year status                                       ratio         ratio Fiscal
                                                                       year
                                                                      ratio
                           Nonintrus Strip- Likelihood                        
                              ive Searc s of strip-
                                     h    search
1997 Nonciti  White Male     2558.23 196.7    0.0769                           
    zen                              7
    
                    Female    718.72 55.28    0.0769  1.00                     
              Black Male     1718.10 65.90    0.0384         0.50              
                    Female    914.45 98.55    0.1078  2.81   1.40              
              Asian Male     1673.75 23.25    0.0139         0.18              
                    Female    622.70 24.30    0.0390  2.81   0.51              
              Hispa Male     8086.00 102.0    0.0126         0.16              
            nic                      0
                    Female   2206.78 78.22    0.0354  2.81   0.46              
     U.S.     White Male     4597.33 269.6    0.0587                0.76       
                                     7
                    Female   1385.72 81.28    0.0587  1.00          0.76       
              Black Male     1166.42 67.58    0.0579         0.99   1.51       
            
                    Female    725.84 118.1    0.1628  2.81   2.78   1.51       
                                     6
              Asian Male      874.73 9.27    0.0106         0.18   0.76       
            
                    Female    336.00 10.00    0.0298  2.81   0.51   0.76       
              Hispa Male     1990.44 49.56    0.0249         0.42   1.97       
            nic
            
                    Female    574.78 40.22    0.0700  2.81   1.19   1.97       
1998 Nonciti  White Male     3090.24 332.7    0.1077                       1.40
    zen                               6
    
                    Female    707.79 76.21    0.1077  1.00                 1.40
              Black Male     2111.29 70.71    0.0335         0.31          0.87
            
                    Female    997.16 93.84    0.0941  2.81   0.87          0.87
              Asian Male     1586.75 19.25    0.0121         0.11          0.87
            
                    Female    562.82 19.18    0.0341  2.81   0.32          0.87
              Hispa Male     9257.03 101.9    0.0110         0.10          0.87
            nic                      7
            
                    Female   2223.19 68.81    0.0310  2.81   0.29          0.87
     U.S.     White Male     5429.16 445.8    0.0821                0.76   1.40
                                     4
                    Female   1341.81 110.1    0.0821  1.00          0.76   1.40
                                     9
              Black Male     1321.16 66.84    0.0506         0.62   1.51   0.87
            
                    Female    713.57 101.4    0.1421  2.81   1.73   1.51   0.87
                                     3
              Asian Male     1068.12 9.88    0.0093         0.11   0.76   0.87
            
                    Female    341.13 8.87    0.0260  2.81   0.32   0.76   0.87
              Hispa Male     2152.20 46.80    0.0217         0.26   1.97   0.87
            nic
            
                    Female    481.58 29.42    0.0611  2.81   0.74   1.97   0.87
Note:  The gender ratios compare the likelihood of
being subjected to strip-searches for women
relative to men within race, citizenship, and
fiscal year categories.  The race ratios compare
the likelihood of being subjected to strip-
searches for Blacks, Hispanics, and Asians
relative to Whites within gender, citizenship, and
fiscal year categories.  The citizenship ratios
compare the likelihood of being subjected to strip-
searches for U.S. citizens relative to noncitizens
within gender, race, and fiscal year categories.
The fiscal year ratios compare the likelihood of
being subjected to strip-searches in 1998 relative
to 1997 within gender, race, and citizenship
categories.
Source: GAO analysis of U.S. Customs data.

The Likelihood of Being
X-Rayed
The results of the preferred model to describe the
effects of fiscal year, race, gender, and
citizenship on the likelihood of being x-rayed as
opposed to simply frisked or patted down is
depicted in table I.10 and are described as
follows:

ï¿½    There was no change from fiscal years 1997 to
1998.

ï¿½    The effect of citizenship is different for
minorities from the effect for Whites and
different among men and women. Among minority men,
there were no differences between citizens and
noncitizens in the likelihood of being x-rayed.
Among White women, the difference was slight
(i.e., citizens were slightly less likely to be x-
rayed, by a factor of 0.9). Among minority women,
however, citizens were more likely than
noncitizens to be x-rayed, by a factor of 1.7.
This difference was the same in both fiscal years.
ï¿½    The effect of race and the differences
between Blacks and Whites, and between Hispanics
and Whites, are greater among citizens than among
noncitizens. Among noncitizens, Blacks and
Hispanics were more likely than Whites to be x-
rayed, by factors of 4.5 and 1.9, respectively;
Asians were much less likely to be x-rayed than
Whites. Among citizens, however, Blacks and
Hispanics were 8.7 and 3.7 times as likely as
Whites to be x-rayed.
ï¿½    The effect of gender on the likelihood of
being x-rayed is the same for minorities as for
Whites and the same in both fiscal years. It
differed, however, for citizens and noncitizens.
Among noncitizens, women were .8 times as likely
to be x-rayed than men; among citizens, women were
1.4 times as likely as men to be x-rayed.

Table I.10: Expected Frequencies Under the
Preferred Model for the Table in Which Type of
Search Is Cross-Classified by Fiscal Year,
Citizenship, Race, and Gender, for X-ray Searches
versus Nonintrusive Searches
Fisca Citizen  Race    Gender     X-ray searches    Gender Race Citizen Fiscal
l     ship                         (expected       ratio ratio    ship   year
year  status                     frequencies)                   ratio  ratio
                               Nonintru X-ray Likelih                        
                               sive       oods of
                                           x-ray
1997  Nonciti  White   Male    2558.85 26.15 0.0102                           
     zen      
     
                       Female   734.67  6.33 0.0086   0.84                    
               Black   Male    1715.58 78.42 0.0457        4.47               
             
                       Female   907.06 34.94 0.0385   0.84 4.47               
               Asian   Male    1672.60  1.40 0.0008        0.08               
             
                       Female   624.56  0.44 0.0007   0.84 0.08               
               Hispan  Male    8104.11 158.89 0.0196        1.92               
             ic
             
                       Female  2207.52 36.48 0.0165   0.84 1.92               
      U.S.     White   Male    4591.73 24.27 0.0053                0.52       
             
                       Female  1370.94 10.06 0.0073   1.39         0.85       
               Black   Male    1169.54 53.46 0.0457        8.65    1.00       
             
                       Female   744.72 47.28 0.0635   1.39 8.65    1.65       
               Asian   Male     876.26  0.74 0.0008        0.16    1.00       
             
                       Female   338.60  0.40 0.0012   1.39 0.16    1.65       
               Hispan  Male    1994.89 39.11 0.0196        3.71    1.00       
             ic
             
                       Female   548.08 14.92 0.0272   1.39 3.71    1.65       
1998  Nonciti  White   Male    3083.49 31.51 0.0102                       1.00
     zen      
     
                       Female   698.98  6.02 0.0086   0.84                1.00
               Black   Male    2123.91 97.09 0.0457        4.47           1.00
             
                       Female  1004.31 38.69 0.0385   0.84 4.47           1.00
               Asian   Male    1582.67  1.33 0.0008        0.08           1.00
             
                       Female   564.60  0.40 0.0007   0.84 0.08           1.00
               Hispan  Male    9236.90 181.10 0.0196        1.92           1.00
             ic
             
                       Female  2221.30 36.70 0.0165   0.84 1.92           1.00
      U.S.     White   Male    5433.29 28.71 0.0053                0.52   1.00
             
                       Female  1357.04  9.96 0.0073   1.39         0.85   1.00
               Black   Male    1295.77 59.23 0.0457        8.65    1.00   1.00
             
                       Female   707.11 44.89 0.0635   1.39 8.65    1.65   1.00
               Asian   Male    1066.10  0.90 0.0008        0.16    1.00   1.00
             
                       Female   340.60  0.40 0.0012   1.39 0.16    1.65   1.00
               Hispan  Male    2178.29 42.71 0.0196        3.71    1.00   1.00
             ic
             
                       Female   480.91 13.09 0.0272   1.39 3.71    1.65   1.00
Note:  The gender ratios compare the likelihood of
being subjected to strip-searches for women
relative to men within race, citizenship, and
fiscal year categories.  The race ratios compare
the likelihood of being subjected to strip-
searches for Blacks, Hispanics, and Asians
relative to Whites within gender, citizenship, and
fiscal year categories.  The citizenship ratios
compare the likelihood of being subjected to strip-
searches for U.S. citizens relative to noncitizens
within gender, race, and fiscal year categories.
The fiscal year ratios compare the likelihood of
being subjected to strip-searches in 1998 relative
to 1997 within gender, race, and citizenship
categories.

Source: GAO analysis of U.S. Customs data.

The Likelihood of Strip-Searches Being Positive
The results of the preferred model to describe the
effects of fiscal year, race, gender, and
citizenship on the likelihood of strip-searches
having positive results are depicted in table I.11
and are described as follows:

ï¿½    There was a small decline in the likelihood
of strip-searches being positive between 1997 and
1998. All groups were less likely to be found
carrying contraband by a factor of 0.8.

ï¿½    In both fiscal years, among men and women of
all races, citizens were 1.5 times as likely as
noncitizens to be found with contraband when they
were strip-searched.
ï¿½    Race interacted with gender; that is, the
difference between Blacks and Whites was less
pronounced among women than men. For both men and
women, Hispanics were 2.4 times as likely to yield
positive strip-search results as Whites, and
Asians were equally as likely as Whites to yield
positive results. Blacks were, like Hispanics,
about 2.4 times as likely as Whites to yield
positive results among men, but 1.4 times as
likely to yield positive results among women who
were strip-searched.
ï¿½    Although women were, among minorities, more
likely than men to be strip-searched when they
were searched, their strip-searches were not more
likely than those of their male counterparts to
yield positive results. Among Whites, Hispanics,
and Asians, strip-searches of women were less
likely than strip-searches of men to yield
positive results, by a factor of 0.8. Among
Blacks, strip-searches of women were less than
half as likely as strip-searches of men to yield a
positive result.

Table I.11: Expected Frequencies Under the
Preferred Model for the Table in Which Result of
Search Is Cross-Classified by Fiscal Year,
Citizenship, Race, and Gender, for Strip-Searches
Only
     Citizens  Race     Gender  Results from strip-                           
Fisc hip                            searches                                
al   status                        (expected      Gende   Race Citizen  Fiscal
year                             frequencies)       r  ratio   ship    year
                                                ratio         ratio   ratio
                                Negat Positi Likeliho                       
                              ive     ve  ods of
                                        positive
1997 Noncitiz  White    Male    155.4  32.60  0.2098                           
    en                           0
    
                        Female  36.11   5.89  0.1630 0.78                      
               Black    Male    44.22  22.78  0.5152        2.46               
             
                        Female  81.25  18.75  0.2307 0.45   1.42               
               Asian    Male    20.66   4.34  0.2098        1.00               
             
                        Female  18.92   3.08  0.1630 0.78   1.00               
               Hispanic Male    66.35  32.65  0.4921        2.35               
             
                        Female  54.98  21.02  0.3823 0.78   2.35               
     U.S.      White    Male    209.6  65.34  0.3116               1.49        
                                6
                        Female  78.90  19.10  0.2421 0.78          1.49        
               Black    Male    32.86  25.14  0.7652        2.46   1.49        
             
                        Female  78.95  27.05  0.3427 0.45   1.42   1.49        
               Asian    Male    6.10   1.90  0.3116        1.00   1.49        
             
                        Female  6.44   1.56  0.2421 0.78   1.00   1.49        
               Hispanic Male    29.46  21.54  0.7310        2.35   1.49        
             
                        Female  42.73  24.27  0.5679 0.78   2.35   1.49        
1998 Noncitiz  White    Male    294.3  49.66  0.1687                       0.80
    en                           4
    
                        Female  76.03   9.97  0.1311 0.78                  0.80
               Black    Male    53.74  22.26  0.4143        2.46           0.80
             
                        Female  72.54  13.46  0.1855 0.45   1.42           0.80
               Asian    Male    18.82   3.18  0.1687        1.00           0.80
             
                        Female  15.91   2.09  0.1311 0.78   1.00           0.80
               Hispanic Male    72.36  28.64  0.3958        2.35           0.80
             
                        Female  57.36  17.64  0.3075 0.78   2.35           0.80
     U.S.      White    Male    347.8  87.18  0.2506               1.49    0.80
                                2
                        Female  83.70  16.30  0.1947 0.78          1.49    0.80
               Black    Male    58.19  35.81  0.6155        2.46   1.49    0.80
             
                        Female  75.26  20.74  0.2756 0.45   1.42   1.49    0.80
               Asian    Male    8.80   2.20  0.2506        1.00   1.49    0.80
             
                        Female  8.37   1.63  0.1947 0.78   1.00   1.49    0.80
               Hispanic Male    13.22   7.78  0.5880        2.35   1.49    0.80
             
                        Female  18.53   8.47  0.4568 0.78   2.35   1.49    0.80
Note:  The gender ratios compare the likelihood of
being subjected to strip-searches for women
relative to men within race, citizenship, and
fiscal year categories.  The race ratios compare
the likelihood of being subjected to strip-
searches for Blacks, Hispanics, and Asians
relative to Whites within gender, citizenship, and
fiscal year categories.  The citizenship ratios
compare the likelihood of being subjected to strip-
searches for U.S. citizens relative to noncitizens
within gender, race, and fiscal year categories.
The fiscal year ratios compare the likelihood of
being subjected to strip-searches in 1998 relative
to 1997 within gender, race, and citizenship
categories.
Source: GAO analysis of U.S. Customs data.

The Likelihood of X-Ray Searches Being Positive
The results of the preferred model to describe the
effects of fiscal year, race, gender, and
citizenship on the likelihood of x-ray searches
being positive are depicted in table I.12 and are
described as follows:

ï¿½    There was a somewhat larger decline in the
likelihood of x-ray searches being positive
between 1997 and 1998 than was the case for strip-
searches. All groups were less likely to be found
to be carrying contraband when x-rayed, by a
factor of 0.7.

ï¿½    Citizenship had no effect on the likelihood
of x-ray results being positive.
ï¿½    Race interacted with gender in its effect on
whether x-ray searches were positive. Hispanics
and Asians were less likely to be found carrying
contraband than Whites among both male and women
passengers x-rayed, and Blacks were less likely to
be found carrying contraband than Whites among
women. Among men, Blacks were equally as likely to
be found carrying contraband as Whites.
ï¿½    X-rays were equally as likely to yield
positive results for women and men among Whites,
Hispanics, and Asians, but less than half as
likely to yield positive results for women than
men among Blacks.

Table I.12: Expected Frequencies Under the
Preferred Model for the Table in Which Result of
Search Is Cross-Classified by Fiscal Year,
Citizenship, Race, and Gender, for X-ray Searches
Only
Fis Citize  Race    Gende   Results from x-ray                             
cal nship          r            searches                                  
yea status              (expected frequencies) Gender   Race Citizen Year
r                                              ratio  ratio    ship ratio
                                                              ratio
                          Negativ Positiv Likelih                          
                             e       e oods of
                                      positiv
                                            e
199 Noncit  White   Male     9.93    8.17 0.8232                           
7   izen    
   
                    Femal    4.99    4.11 0.8232   1.00                    
                 e
            Black   Male    42.29   34.81 0.8232          1.00             
          
                    Femal   20.86    8.24 0.3950   0.48   0.48             
                 e
            Asian   Male     1.42    0.68 0.4779          0.58             
          
                    Femal    0.07    0.03 0.4779   1.00   0.58             
                 e
            Hispan  Male   117.80   56.30 0.4779          0.58             
          ic
          
                    Femal   23.75   11.35 0.4779   1.00   0.58             
                 e
    U.S.    White   Male    13.22   10.88 0.8232                  1.00     
          
                    Femal    6.64    5.46 0.8232   1.00           1.00     
                 e
            Black   Male    25.83   21.27 0.8232          1.00    1.00     
          
                    Femal   38.78   15.32 0.3950   0.48   0.48    1.00     
                 e
            Asian   Male     0.74    0.36 0.4779          0.58    1.00     
          
                    Femal    0.74    0.36 0.4779   1.00   0.58    1.00     
                 e
            Hispan  Male    30.52   14.58 0.4779          0.58    1.00     
          ic
          
                    Femal   10.22    4.88 0.4779   1.00   0.58    1.00     
                 e
199 Noncit  White   Male    22.75   13.35 0.5869                       0.71
8   izen    
   
                    Femal    4.47    2.63 0.5869   1.00                0.71
                 e
            Black   Male    72.53   42.57 0.5869          1.00         0.71
          
                    Femal   29.73    8.37 0.2816   0.48   0.48         0.71
                 e
            Asian   Male     0.07    0.03 0.3407          0.58         0.71
          
                    Femal    0.82    0.28 0.3407   1.00   0.58         0.71
                 e
            Hispan  Male   119.41   40.69 0.3407          0.58         0.71
          ic
          
                    Femal   30.66   10.44 0.3407   1.00   0.58         0.71
                 e
    U.S.    White   Male    13.93    8.17 0.5869                  1.00 0.71
          
                    Femal    9.52    5.58 0.5869   1.00           1.00 0.71
                 e
            Black   Male    38.50   22.60 0.5869          1.00    1.00 0.71
          
                    Femal   25.83    7.27 0.2816   0.48   0.48    1.00 0.71
                 e
            Asian   Male     0.07    0.03 0.3407          0.58    1.00 0.71
          
                    Femal    0.82    0.28 0.3407   1.00   0.58    1.00 0.71
                 e
            Hispan  Male    32.15   10.95 0.3407          0.58    1.00 0.71
          ic
          
                    Femal    7.53    2.57 0.3407   1.00   0.58    1.00 0.71
                 e
Note:  The gender ratios compare the likelihood of
being subjected to strip-searches for women
relative to men within race, citizenship, and
fiscal year categories.  The race ratios compare
the likelihood of being subjected to strip-
searches for Blacks, Hispanics, and Asians
relative to Whites within gender, citizenship, and
fiscal year categories.  The citizenship ratios
compare the likelihood of being subjected to strip-
searches for U.S. citizens relative to noncitizens
within gender, race, and fiscal year categories.
The fiscal year ratios compare the likelihood of
being subjected to strip-searches in 1998 relative
to 1997 within gender, race, and citizenship
categories.
Source: GAO analysis of U.S. Customs data.

Customs' Policies and Procedures for Conducting
Personal Searches
     To determine Customs' policies and procedures
for conducting personal searches, we interviewed
Customs headquarters officials in the Office of
Field Operations, Passenger Processing Operations.
We also met with officials from the Office of
Chief Counsel to discuss Customs' border search
authority, several recent lawsuits, and the bases
for changing and drafting new search policies
implemented at airports around the country.

     We also reviewed agency program documents,
including the Handbook (March 1997) and the
revised version (September 1999); Customs' Law
Course; and Passenger Selectivity Training
provided at the Federal Law Enforcement Training
Center (FLETC) at Glynco, GA.

     We reviewed recent directives and proposed
changes to Customs' personal search policies and
procedures.  We also reviewed reports and
recommendations from internal and external
organizations, including the Customs' Passenger
Processing Targeting Committee and the Department
of the Treasury's Office of Professional
Responsibility.

     To determine how policies and procedures were
implemented at airports around the country and
obtain views of inspectors and supervisors, we
conducted an anonymous national mail survey. We
were not able to interview inspectors or
supervisors conducting passenger processing
because Customs headquarters officials told us
that we would be interfering with on-going
investigations or "active" cases.5

     We had planned to obtain the views of
inspectors and supervisors at the four airports we
visited about (1) air passenger inspection
policies and procedures; (2) training related to
passenger processing; and (3) the impact, if any,
of awards and promotions on the number and types
of personal searches conducted. Customs' Chief
Counsel's Office insisted that one of its
attorneys be present at all of our interviews with
the inspectors and supervisors in order to protect
the interests of Customs and possible legal
interests of inspectors who could be subject to
litigation involving personal searches. Also, the
Chief Counsel said that counsel would treat the
interviews as informal depositions and help ensure
that our questions were consistent with the
objectives of the review. Customs also stated that
"inspectors cannot be immune from disciplinary
action resulting from an admission to GAO that
they violated Customs policy or a regulation,
which is made in front of Chief Counsel.
Therefore, responses are not confidential."
Customs added that inspectors and supervisors
would meet with us on a voluntary basis.

     In commenting on our draft report, Customs
said that inspectors and supervisors who we sought
to interview were all potential defendants who
could be personally liable in civil lawsuits on
the basis of actions, policies, and procedures on
which we wanted to question them.  In addition,
Customs added that Treasury employees, including
its attorneys who would attend inspector
interviews, are obligated by Customs to report
evidence of misconduct and could not grant blanket
immunity to inspectors who would be interviewed,
as requested by the National Treasury Employees
Union.

     Subsequently, the National President of the
National Treasury Employees Union recommended that
inspectors not participate in interviews with us
because there was no guarantee that employees
would be immune from discipline based on an
admission to us that they violated Customs policy
or regulation. Prior to Customs' decision to treat
our interviews as depositions and its concern
about the issue of disciplinary action, the Union
was supportive of our interviewing inspectors.

     On January 14, 1999, the Commissioner of
Customs sent a letter to the Comptroller General
suggesting that as an alternative to inspector and
supervisor interviews, we conduct an anonymous
mail survey of airport inspectors and supervisors.
We subsequently developed an anonymous mail survey
to obtain the views of inspectors and supervisors
assigned to air passenger processing. The Union
provided us with a letter to its members that
supported our mail survey and requested its
members to complete it.

     We sent surveys to all 1,176 inspectors and
all 177 supervisors assigned to passenger
processing at 15 airports in the United States
with the largest volume of international
passengers. The surveys included questions on
targeting methods, personal search practices, and
training. The response rate for inspectors was 66
percent, and for supervisors the response rate was
74 percent. See appendix II and appendix III for
copies of our survey instruments.

     The practical difficulties of conducting any
survey may introduce nonsampling errors.
Differences in the wording of questions, in the
sources of information available to respondents,
or in the types of people who do not respond can
lead to somewhat different results. We took steps
to minimize nonsampling errors. For example, we
developed our survey questions with the aid of a
survey specialist and pretested the questions
prior to submitting them to inspectors and
supervisors.

     To obtain data on the reasons passengers were
selected for searches, we reviewed 70 random
patdown search reports for all searches conducted
in fiscal years 1997 and 1998. We focused on
patdowns because they represent the first level of
personal search.  Because patdowns at that time
were to be conducted at the discretion of the
inspector, we could obtain the reasons provided by
the inspector for initiating the search. Our
sample was selected from the database for the
Automated Passenger Analysis System, which
contains all passenger search and seizure
information, both positive and negative. A
separate population was created to limit the scope
of the sampling to fiscal years 1997 and 1998 in-
bound commercial air passengers who had been
subject to patdown searches. The populations
consisted of a total of 85,800 search reports.
Because of the small sample size, we did not
project the results to the population.

Customs' Management Controls
     To determine the management controls in place
that were intended to ensure that passengers are
not inappropriately subjected to personal
searches, we interviewed key officials at Customs
headquarters and at the four ports. We identified
four internal controls that could be used to
ensure that inspectors are conducting personal
searches in accordance with prescribed guidelines.
These are (1) training on personal search
procedures; (2) supervisory oversight and approval
for referrals to pursue personal searches; (3) a
complaint process to identify and resolve
passenger complaints related to personal searches;
and (4) the Compliance Examination Measurement
System (COMPEX), which measures the effectiveness
of passenger targeting for secondary searches and
passenger compliance with the laws that Customs
enforces, including drug laws.

     At each port we interviewed the port
directors, supervisory inspectors designated by
the ports, and passenger service representatives
responsible for handling passenger complaints. To
determine training standards and requirements for
supervisors and inspectors related to personal
search policies and procedures, we interviewed
individuals responsible for training and reviewed
the mandatory training requirements, advanced and
refresher training requirements, and on-the-job
training provided at the ports.

     We reviewed all passenger complaint files for
complaints related to personal searches filed at
the four ports in fiscal years 1997 and 1998. We
reviewed the files to determine the process for
handling and resolving passenger complaints at the
ports.

     As previously discussed, the national mail
surveys sent to inspectors and supervisors at 15
airports also included questions about training,
personal search procedures, and supervisory
oversight of personal searches conducted at the
ports.

     To understand how Customs is using and
implementing COMPEX, we interviewed program
officials at the four ports, Customs Management
Centers, and headquarters. We also obtained data
from headquarters program officials.

     We conducted our work from August 1998
through January 2000 in accordance with generally
accepted government audit standards. We received
comments on a draft of this report from the
Department of the Treasury.  The U.S. Customs
Service provided comments that are summarized at
the end of the letter; included in the report,
where appropriate; and reproduced in appendix IV.
Customs also provided technical comments that we
incorporated where appropriate.

_______________________________
1 The number of Native Americans subjected to
personal searches in 1997 and 1998 (n=79) was too
small to include in our multivariate analyses.
2 Another type of personal search is a body cavity
search.  Body cavity searches are defined as any
visual or physical intrusion into body cavities.
However, only 32 such searches were performed in
both fiscal years 1997 and 1998, so we excluded
these from any analyses.
3 Our multivariate results are on the basis of the
67,553 cases with no missing data.  However, we
did repeat the analysis including cases that were
coded as missing and found that the overall
results with respect to the effects of sex, race,
citizenship, and year were similar.
4 A main effect includes, for example, the
relationship between gender and the likelihood of
being strip-searched.  An interaction effect
includes, for example, the relationship between a
category of race and gender on the likelihood of
being strip-searched.
5 In developing our methodology, we discussed
personal searches with inspectors and supervisors
at  Washington Dulles International Airport,
located in Virginia.

Appendix II
Inspector Survey
Page 71GAO/GGD-00-38 Airline Passengers Selected f
or Personal Searches

Appendix III
Supervisor Survey
Page 83GAO/GGD-00-38 Airline Passengers Selected f
or Personal Searches

Appendix IV
Comments From the U.S. Customs Service
Page 86GAO/GGD-00-38 Airline Passengers Selected f
or Personal Searches

Appendix V
GAO Contacts and Staff Acknowledgments
Page 87GAO/GGD-00-38 Airline Passengers Selected f
or Personal Searches
GAO Contacts
Laurie E. Ekstrand, (202) 512-8777
James M. Blume, (202) 512-8777

Acknowledgments
     In addition to those named above, Kathleen H.
Ebert, Wendy C. Simkalo, Lisa G. Shibata, Bunnie
Lempesis, Norman J. Rabkin, Barry J. Seltser,
Wendy Ahmed, Douglas M. Sloane, Stuart M. Kaufman,
Donna M. Leiss, Geoffrey R. Hamilton, Jan B.
Montgomery, Lauren E. Bloomquist, and Gretchen T.
Leyman made key contributions to this report.

*** End of Document ***