Title: Comparing Drug Use Rates of Detained Arrestees in the United
States and England. 
Series: Research Report
Author(s): Bruce Taylor and Trevor Bennett 
Published: National Institute of Justice, April 1999
Subject(s): Criminal justice system, drugs and crime, drug abuse, drug
testing, and drug treatment
53 pages
125,000 bytes

------------------------------

Figures, charts, forms, and tables are not included in this ASCII plain-text
file. To view this document in its entirety, download the Adobe Acrobat
graphic file available from this Web site or order a print copy from NCJRS
at 800-851-3420 (877-712-9279 For TTY users).

------------------------------

U.S. Department of Justice
Office of Justice Programs
National Institute of Justice

Comparing Drug Use Rates of Detained Arrestees in the United States and
England 

Research Report

U.S. Department of Justice
Office of Justice Programs
810 Seventh Street N.W.
Washington, DC 20531

Janet Reno
Attorney General

Raymond C. Fisher
Associate Attorney General

Laurie Robinson
Assistant Attorney General

Noel Brennan
Deputy Assistant Attorney General

Jeremy Travis
Director, National Institute of Justice

K. Jack Riley
Director, Arrestee Drug Abuse Monitoring Program

Office of Justice Programs
World Wide Web Site
http://www.ojp.usdoj.gov

National Institute of Justice
World Wide Web Site 
http://www.ojp.usdoj.gov/nij

------------------------------

Comparing Drug Use Rates of Detained Arrestees in the United States and
England

By Bruce Taylor and Trevor Bennett

April 1999
NCJ 175052

Jeremy Travis
Director

Findings and conclusions are those of the authors, Dr. Bruce Taylor and
Dr. Trevor Bennett, and do not necessarily reflect the official position or
policies of the U.S. Department of Justice, the Home Office, or the
University of Cambridge. Comments and questions should be addressed to
them. Dr. Bruce Taylor, Social Science Analyst, can be reached at the
National Institute of Justice, Office of Research and Evaluation, 810
Seventh St. N.W., Room 7308, Washington, DC 20531; e-mail
taylorb@ojp.usdoj.gov. Dr. Trevor Bennett, Acting Head of Department,
Institute of Criminology, University of Cambridge, can be reached at the
University at 7 West Road, Cambridge, CB3 9DT, England; e-mail
thbl@cam.ac.uk.

------------------------------

The National Institute of Justice is a component of the Office of Justice
Programs, which also includes the Bureau of Justice Assistance, the Bureau
of Justice Statistics, the Office of Juvenile Justice and Delinquency
Prevention, and the Office for Victims of Crime.

------------------------------

ACKNOWLEDGMENTS

The authors would like to acknowledge the contributions of individuals
whose efforts made this project possible: National Institute of Justice (NIJ)
Director Jeremy Travis, for encouraging us to write this report and for the
direction he has provided in guiding the development of the International
Arrestee Drug Abuse Monitoring (I-ADAM) program; ADAM Director K.
Jack Riley, for his efforts in shaping the broad content of this report and for
key technical recommendations; the support and encouragement of the
Research Development and Statistics Directorate of the Home Office, in
particular, Malcolm Ramsay, Principal Research Officer, Home Office
Research Development and Statistics Directorate, who provided insightful
comments and suggestions on earlier drafts of this report; and Richard Gill
(and others) at the Forensic Science Service in Chepstow, England, who
gave advice on methods of matching the English screening test cutoff levels
to those used in the United States.

Numerous others within NIJ, the Office of Justice Programs, and other
organizations have contributed to this report. In particular, the authors
extend special thanks to Thom Feucht, James Finckenauer, Nora
Fitzgerald, Sally Hillsman, Dana Hunt, Robert Kaminski, Patrick A.
Langan, John MacDonald, Marvene OĠRourke, Bill Rhodes, and Christy
Visher for their helpful comments or suggestions with earlier drafts. In
addition, two anonymous reviewers provided valuable critiques that have
enhanced the technical aspects of this report. Of course, the responsibility
for errors and omissions lies solely with the authors.

Bruce Taylor, Ph.D.
Social Science Analyst 
Arrestee Drug Abuse Monitoring Program
National Institute of Justice
U.S. Department of Justice

Trevor Bennett, Ph.D.
Acting Head of Department
Institute of Criminology
University of Cambridge

------------------------------

Contents

ACKNOWLEDGMENTS


EXECUTIVE SUMMARY

INTRODUCTION

The I-ADAM Program
NEW ADAM
Resolving Differences in Methods

STUDY METHOD--MATCHING THE SAMPLES

Ensuring Eligibility of Study Subjects
Selecting Matched Sites
Selecting Matched Study Subjects
Does the Sampling Method Affect Comparability?
Do Sample Size and Related Factors Affect Comparability?

FINDINGS--SIMILARITIES AND DIFFERENCES

Results of Urinalysis
Does the Country Make a Difference?
Comparing Urinalysis and Self-Reports
Results of Self-Reported Drug Use

DISCUSSION--TOWARD UNDERSTANDING THE WORLDWIDE
DRUG PROBLEM

Differences Depend on Type of Drug
Possible Explanations
Next Steps in Crossnational Research

APPENDIX A

APPENDIX B (See in PDF file)

------------------------------

TABLES (See PDF file)

Table 1. Cutoff Levels for Drugs Detected by Urinalysis

Table 2. Matched Sites Selected for the Study

Table 3. Preweighted Offenses of Detained Arrestees, United States and
England

Table 4. Preweighted Age of Detained Arrestees, United States and
England

Table 5. Effects of Various Exclusions on Sample Size

Table 6. Final Sample Size in Each of the 10 Sites

Table 7. Drug Use by Demographic Characteristics and Type of Crime--
Detained Arrestees, United States and England (Results of Urinalysis)

Table 8. Summary Profiling Subgroups of Detained Arrestees With 
Highest Drug Use Prevalence--England and the United States

Table 9. Drug Use of Detained Arrestees in 5 U.S. Sites and 5 English
Sites--Logistic Regression (Results of Urinalysis)

Table 10. Comparison of Urinalysis Results and Self-Reported Drug Use
by Detained Arrestees--England and the United States

Table 11. Self-Reported Drug Use of Detained Arrestees--5 Matched U.S.
Sites and 5 English Sites (Weighted and Excluded Data)

Table 12. Injection Drug Use by Detained Arrestees--5 English Sites and 5
Matched U.S. Sites

Table 13. Drug-Using Careers of Detained Arrestees in England and the
United States

Table 14. Legal and Illegal Income and Amount Spent on Drugs (Logged 
Values)--Detained Arrestees in England and the United States, Analysis of 
Variance (ANOVA) Between Drug Users and Nonusers (Results of
Urinalysis)

Table A-1. Available Census Demographic Characteristics and Crime Rate
Data--5 English Sites and 23 U.S. Sites

Table A-2. Rates of Drug Use by Detained Arrestees: Original 5 Matched 
U.S. Cities, Alternative 5 Matched U.S. Cities, and All 23 U.S. ADAM 
Cities--Main Effects, Using Logistic Regression--Results of Urinalysis

Table B-1. Drug Use Prevalence of Detained Arrestees: 5 English Sites and
23 U.S. Sites, by Site and Drug Type (Weighted and Excluded Data)--
Results of Urinalysis

Table B-2. Drug Use Prevalence of Male and Female Detained Arrestees,
England and the United States, by Drug Type and Demographic
Characteristics--Results of Urinalysis

Table B-3. Drug Use Prevalence of Younger and Older Detained
Arrestees, England and the United States, by Drug Type and Demographic
Characteristics--Results of Urinalysis

Table B-4. Drug Use Prevalence of White and Nonwhite Detained
Arrestees, England and the United States, by Drug Type and Demographic
Characteristics--Results of Urinalysis

Table B-5. Drug Use Prevalence of Unemployed and Employed Detained
Arrestees, England and the United States, by Drug Type and Demographic
Characteristics--Results of Urinalysis

CHARTS

Chart 1. Drug Use Prevalence of Detained Arrestees: 23 U.S. Sites and 5
English Sites (Raw Data)--Results of Urinalysis

Chart 2. Drug Use Prevalence of Detained Arrestees: 5 Matched U.S. Sites 
and 5 English Sites (Weighted and Excluded Data)--Results of Urinalysis

Chart A-1. Drug Use Prevalence of Detained Arrestees: 5 Original
Matched Sites, 5 Alternative Matched Sites, and 23 U.S. Sites 
(Weighted and Excluded Data)--Results of Urinalysis

MAPS (See PDF file)

Map 1. NEW ADAM Sites

Map 2. ADAM Sites

------------------------------

EXECUTIVE SUMMARY

This report is a product of the recent establishment of the International
Arrestee Drug Abuse Monitoring (I-ADAM) program, administered by the
National Institute of Justice, U.S. Department of Justice. Participating
countries collaborate in implementing national programs similar to that of
the Arrestee Drug Abuse Monitoring (ADAM) program (formerly the
Drug Use Forecasting [DUF] program) in the United States. Under the
ADAM program, detained arrestees in urban jurisdictions throughout the
United States are tested periodically to determine the extent of illicit drug
use in this at-risk population.

At the time I-ADAM was launched in 1998, one of the participating
countries, England, had already established a pilot program of drug testing
detained arrestees and had published the first set of results.[1] The
generation of the dataset of drug use among detained arrestees, which was
based on procedures similar to those of the ADAM program, presented an
early opportunity to compare drug use by this group in the United States
with that in another country.

The analysis presented in this report compares the findings from surveys of
arrestees detained in five locations in England with those from similar
surveys conducted in five matched locations in the United States. The data
were adjusted and weighted in various ways to make the two samples for
both countries as similar as possible. After excluding nonmatched cases, 
the final dataset consisted of 4,470 in the United States and 839 in
England.

The report compares several aspects of drug use in the two countries:

--Urinalysis results for use of six types of drugs.

--Self-reported use of 10 types of drugs.

--Extent to which drugs are injected.

--Extent to which arrestees had received drug treatment.

--Extent to which arrestees wanted to receive drug treatment.

--Drug-using "careers" (age of first drug use).

--Levels of legal and illegal income.

Comparison of the two countries reveals that the use of opiates/heroin,
methadone, and amphetamines tends to be higher among detained arrestees
in England than in the United States. For benzodiazepines and marijuana,
comparison reveals no real difference between the two countries. Only for
cocaine/crack was use significantly higher in the United States. The study
also revealed a number of notable correlations between drug use and
various demographic and related characteristics. For several of these
characteristics, the subgroups with the highest drug use rates are the same
in both countries. Injection as a method of administering drugs is
moderately high in both countries, with some distinct differences
between the two countries in preference of administration for specific
drugs. Few differences between the two were found in the extent to which
arrestees received drug treatment or their reported need for it. There was
also little difference in age of initiation of drug use (although there were
some differences when it came to specific drug types). The findings on
legal and illegal income indicate that detained arrestees in England tend to
spend more on drugs and to report higher levels of illegal income than their
counterparts in the United States. 

Rates of Drug Use

--In both countries, a large proportion of detained arrestees tested
positive for one or more drugs (England, 59 percent; United States, 68
percent).

--For opiates, methadone, and amphetamines, the percentage of detained
arrestees who tested positive was higher in England.

--For cocaine, "any drug," and "multiple drugs," the percentage who
tested positive was higher in the United States.

--For benzodiazepines and marijuana, there was no significant difference
between the two countries.

--Female detained arrestees in both countries were more likely than males
to test positive for opiates, methadone, cocaine, and amphetamines.

--Male detained arrestees in both countries were more likely than females
to test positive for marijuana.

--In both countries, older detained arrestees (age 21 or over) were more
likely than younger ones (age 20 or under) to test positive for opiates,
cocaine, amphetamines, benzodiazepines, and methadone.

--In both countries, younger detained arrestees (age 20 or under) were
more likely than older ones (age 21 or over) to test positive for marijuana.

--In both countries, nonwhite detained arrestees were more likely than
white detained arrestees to test positive for marijuana.

--More than 55 percent of the detained arrestees in England and slightly
less than 50 percent of those in the United States said they had used at least
1 of 10 selected drugs in the past 3 days.

--In England, the self-reported use of 8 of 10 selected drugs plus alcohol
(marijuana, opiates, amphetamines, methadone, benzodiazepines, LSD,
inhalants, and alcohol) was higher than in the United States.

--In the United States, the self-reported use of 3 of 10 selected drugs plus
alcohol (crack cocaine, powder cocaine, and barbiturates) was higher than
in England.

Urinalysis Versus Self-Reports in Measuring Drug Use

--Overall, for more than 90 percent of the detained arrestees in the
United States and England, the findings of the self-report survey and the
urinalysis were in agreement.

--Underreporting drug use was higher in the United States than in
England.

--In the United States, 8 percent of the detained arrestees underreported
drug use compared with 2 percent who overreported. Underreporting in
the United States was especially evident for use of marijuana (17 percent)
and cocaine (17 percent).

--In England, 4 percent of the detained arrestees underreported drug use,
while 5 percent overreported. Slightly more than 10 percent of the detained
arrestees in England overreported marijuana use and 7 percent
underreported it.

Injecting Drugs

--Detained arrestees in England were significantly more likely than those
in the United States to say they had injected amphetamines at some time in
their lives (16 percent and 2 percent, respectively).

--Detained arrestees in the United States were significantly more likely
than those in England to say they had injected cocaine (11 percent and 8
percent, respectively).

Extent of Treatment for Drug and Alcohol Abuse

--There was no significant difference between arrestees in the United
States and England in the proportion who reported having ever received
drug treatment (28 percent and 26 percent, respectively).

--There was no significant difference among detained arrestees in the
United States and England in the proportion reporting having ever received
treatment for alcohol problems (12 percent and 11 percent, respectively).

--Detained arrestees in the United States were more likely than those in
England to say they would like to receive drug treatment (33 percent and
22 percent, respectively).

--Detained arrestees in the United States were not notably more likely
than those in England to say they would like to receive treatment for
alcohol problems (14 percent and 13 percent, respectively).

Drug-Using Careers

--Detained arrestees in the United States reported using 5 of 10 drug
types (marijuana, powder cocaine, barbiturates, methadone, and
benzodiazepines) at an earlier age than did detained arrestees in England.

--Detained arrestees in England reported using 5 of 10 drug types (crack
cocaine, opiates, amphetamines, LSD, and inhalants) at an earlier age than
arrestees in the United States.

Legal and Illegal Income

--Detained arrestees in England had higher illegal incomes than those in
the United States.

--Detained arrestees in England spent more on drugs than those in the
United States.

--Detained arrestees in the United States had higher legal incomes than
those in England.

--In both the United States and England with one exception, detained
arrestees who tested positive for any specific drug had higher illegal
incomes and spent more money on drugs than those who tested negative
for that drug. The one exception was U.S. detained arrestees who used
amphetamines.

Note

1. Bennett, T.H., Drugs and Crime: The Results of Research on Drug
Testing and Interviewing Arrestees, Home Office Research Study No. 183,
London: Home Office, 1998.

------------------------------

INTRODUCTION

There is a widely held belief that crime rates and drug use are much higher
in the United States than in England. However, recent research has
revealed that rates for some of the most common crimes (such as robbery,
assault, burglary, and motor vehicle theft) are in fact higher in England.[1]
There has been no similar comparison of drug use in the two countries. A
study conducted under the I-ADAM (International Arrestee Drug Abuse
Monitoring) program was intended to fill that gap, and the findings are
reported here.

The I-ADAM Program

I-ADAM is a component of the ADAM program, established by the
National Institute of Justice (NIJ), the research arm of the U.S. Department
of Justice, to monitor drug abuse among detained arrestees in urban
jurisdictions throughout the United States. The forerunner of ADAM was
the Drug Use Forecasting (DUF) program. Launched in 1987 by NIJ, DUF
demonstrated the feasibility of urinalysis as an effective means of measuring
drug abuse by arrestees. (See "DUF/ADAM Research.") A decade after it
was established, the program was renamed ADAM to reflect a redesign
intended to make it more rigorous methodologically (by using
representative, probability-based sampling, for example), wider ranging
geographically (covering up to 75 cities), and broader based as a
"platform" on which to study policy and research questions. By focusing on
arrestees, NIJ created in the ADAM program an effective method of
studying hardcore drug use. Because they often do not reside in households
stable enough to be included in broad community household surveys,
hardcore drug users are often not counted in those surveys (for example,
the U.S. Department of Health and Human Services' [HHS's] National
Household Survey on Drug Abuse), and they often drop out of school and
thus are not included in HHS's Monitoring the Future study, a periodical
survey of drug use by high school students. Interviewing arrestees in
detention facilities is also more cost-effective than interviewing hardcore
drug users at the street level using ethnographic sampling strategies.

Aims of I-ADAM. I-ADAM aims to integrate the process of monitoring
drug abuse by arrestees at the international level and the research related to
that process. The program began in 1998 at a conference attended by
representatives of nine countries: Australia, Chile, England, the
Netherlands, Panama, Scotland, South Africa, the United States, and
Uruguay. I-ADAM will be the first international drug prevalence program
to generate standardized data on drug abuse among the high-risk
population of detained arrestees. It will serve as a base for coordinating
drug-related research and drug control policies within and among
participating countries.

Method. At each I-ADAM data-collection site, trained interviewers will
conduct one-on-one interviews with adult male and adult female detained
arrestees and take voluntary urine specimens from each of them. The
directors of these sites, in collaboration with NIJ, will determine which
drugs the arrestees need to be tested for and how many drugs to include in
the drug test panel (list of illicit drugs). All I-ADAM sites will test for at
least five common drugs: marijuana, cocaine (including crack), opiates
(including heroin), amphetamines, and benzodiazepines.

In consultation with the other I-ADAM sites, NIJ has developed a core
survey instrument. Once the sites agree on the details of this core survey, it
will be used by all of them. Later, addenda surveys will be developed to
cover special topics related to drug abuse (for example, domestic violence)
and will be available to all the I-ADAM sites. 

The basic requirements of I-ADAM data collection include the ability to
conduct interviews within 48 hours of arrest (because many drugs cannot
be detected beyond 2 to 3 days of consumption), the availability of a pool
of interviewers who are not law enforcement officials or lockup staff, and
the ability to maintain confidentiality of information for the arrestees who
consent to participate in the research.

NIJ's role. NIJ is providing technical assistance to initiate and operate each
I-ADAM site: Visits to most I-ADAM sites to assess their state of
readiness to collect I-ADAM data, assistance to local officials in building a
coalition of local support, and advice on a variety of scientific issues (for
example, sampling and data analysis).

In developing the I-ADAM program, NIJ will focus on four main areas--
instrumentation, drug testing, clearinghouse development, and training--as
follows: 

--Promoting the core survey instrument and conducting comparability
checks among the participating countries. The survey will be translated into
different languages and a common data entry system will be developed.

--Examining the impact of using different drug-testing kits on substantive
results and methods, and reconciling any differences.

--Serving as a clearinghouse for I-ADAM information. This role covers
storage of common data on a secured Web page
(http://www.Adam-NIJ.net/adam/iadam.htm), storage of core and
specialized addenda surveys, and documentation of program
implementation in each country.

--Providing assistance in developing an interviewer training program and
developing training materials for each participating country.

The participating countries. At the time I-ADAM was established, one of
the participating countries, England, had completed pilot surveys of drug
use among detained arrestees at five sites in a program called NEW ADAM
(New England and Wales Arrestee Drug Abuse Monitoring), had published
the research findings,[2] and had expanded to a second stage of data
collection in three new sites. The status of the other participating countries
is as follows:

--Australia began data collection in January 1999 for a 3-year funded
study in three cities under the title of the DUMA program (Drug Use
Monitoring in Australia).

--Chile began data collection in January 1999 in two cities.

--South Africa has obtained funding to conduct a nine-city study in
mid-1999.

--Scotland has obtained funding to conduct a two-region study in early
1999.

--The Netherlands and Panama each have made progress in planning a
pilot program and are actively seeking funds.

All the active I-ADAM participating countries have obtained funding
through local or national government sources.

NEW ADAM

England was chosen as the comparison site for this study because it was
the only participating country outside the United States at the time of this
study to have generated data in a process similar to that used in ADAM.
Before this study, there were no surveys comparable to those of the
ADAM program to measure arrestee drug use in England, although some
studies of drug and alcohol abuse among prisoners have been conducted
there. (See "Research in England on Arrestee Drug Abuse.") In this
respect, this study is a first for England. It is also the first attempt to
quantitatively compare drug use of detained arrestees in two countries that
use similar research methods.

The methods of interviewing and drug testing in NEW ADAM are based
on the procedures used in ADAM. Information supplied by NIJ for the
design of the NEW ADAM program was used to aid in the development of
the English research instruments. In many respects, the data-collection
methods of the two countries are very similar. Because the ADAM
procedures are fairly well known, only those of NEW ADAM are
presented in detail here, with an emphasis on similarities and differences
between the two.

Data collection in England. The NEW ADAM research was conducted in
five sites: Cambridge and London in the South, Manchester and
Nottingham in the Midlands, and Sunderland in the North. Convenience
sampling was used in the first three surveys (those conducted at
Cambridge, London, and Manchester), and probability sampling was used
in the latter two (those conducted at Nottingham and Sunderland). In the
latter method, the interviews took place 7 days a week, 24 hours a day, and
covered all eligible arrestees brought to the facilities.

The English samples of study subjects were drawn from male and female
adult arrestees; juveniles were deemed ineligible. In most sites, 28
consecutive days were needed to reach the target number of study subjects,
compared with an average of 14 consecutive days in the U.S. sites. The
amount of time was longer because fewer arrestees are processed through
English custody blocks[3] (on average, about 500 a month).

The questionnaire used in the self-report part of the research was based on
those used in ADAM and covered:

--Self-reported drug use (ever, in the past 12 months, in the past month,
and in the past 3 days).

--Injecting drugs and sharing needles.

--Dependency on drugs and alcohol.

--Links between drugs and crime.

--Legal and illegal sources of income.

--Amount of money spent on alcohol and other drugs.

--Treatment needs.

The procedure for collecting urine specimens also was based on ADAM,
using a similar "chain of custody" approach. The specimens were tested for
eight types of drugs (marijuana, opiates, methadone, cocaine,
amphetamines, benzodiazepines, LSD, and alcohol),[4] with a screening
test similar to that used in ADAM.[5]


In all, 839 arrestees were interviewed and 622 provided urine specimens.
Of those asked to volunteer for the interviews, between 84 and 87 percent
agreed to do so; of those who were asked to volunteer a urine specimen,
between 63 and 82 percent did so. It should be noted, however, that there
were minor differences between those who participated in the study and
those who were eligible but did not participate and between those who
agreed to provide a urine specimen and those who did not.[6]

Resolving Differences in Methods 

This report consists of a quantitative comparison of drug use by arrestees
in the first two I-ADAM countries. It presents similarities and differences
between England and the United States in drug use, examines drug use in
terms of various demographic and related characteristics of the users, and
explores issues related to drugs and criminal behavior.

Although, as noted above, the procedures used by the two countries are in
many respects similar, there are differences. The authors acknowledge the
challenge of conducting effective retrospective comparisons between
countries when the nature of both the research methods used in data
collection and the survey sites are different in each. For this study,
differences in research methods presented less of a problem than
differences in the survey sites, because the research in England was based
closely on the procedures developed in the U.S. ADAM program.

The differences in survey sites were potentially more problematic.
However, an attempt was made to ensure that the sites were more
comparable by matching procedures and by basing all comparisons on the
best-matched sites. Thus, the effects of some differences between the two
countries were addressed and, we hope, some useful comparative analyses
were made.

Notes

1. Langan, P.A., and D.P. Farrington, Crime and Justice in the United
States and in England and Wales, 1981-96, Washington, D.C.: U.S.
Department of Justice, Bureau of Justice Statistics, 1998, NCJ 173402.

2. Bennett, T.H., Drugs and Crime: The Results of Research on Drug
Testing and Interviewing Arrestees, Home Office Research Study No. 183,
London: Home Office, 1998.

3. Custody blocks are temporary detention facilities to which people are
brought when first arrested.


4. In the ADAM program, the urine specimens are not tested for LSD or
alcohol.

5. There are two main types of technology for drug testing: immunoassays,
which are used primarily for drug screening, and gas chromatography
(GC), which is used primarily for drug confirmation following screening.
The former are less expensive but also less reliable. Both ADAM and NEW
ADAM screen (conduct preliminary tests) to detect drugs in urine. ADAM
uses the immunoassay Enzymes Multiplied Immune Testing (EMIT) for
screening and does confirmation testing only for amphetamines. In other
words, for all cases that screen positive for amphetamines, a confirmation
test is conducted to determine if methamphetamine was used. NEW
ADAM uses a similar screening test, online Kinetic Interaction of
Microparticles in a Solution (KIMS), but no confirmation tests. Both the
EMIT and KIMS screening processes either detect the drug itself, or the
assay detects the metabolites of the drug (compounds that result from the
breakdown of the drug by the body) that indicate the drug was used. The
assays have a screen accuracy rate of 97 to 100 percent and, when
confirmed by a scientifically accepted alternative urine testing technology
(GC/MS [mass spectrometry], for example), an accuracy rate of virtually
100 percent. In some cases, the screening process is very specific to a drug,
while in others, it is general to a class of drugs that includes illegal
substances. For example, there are specific "markers" for marijuana that a
screening test can detect, but there is no specific marker for heroin.
Instead, a screening test detects byproducts that can be indicative of not
only heroin use but codeine use as well. In other words, screening tests are
general to opiates, not specific to heroin. For cases in which a screen
indicates a class of drugs but not a specific drug, a confirmation test can be
done.

6. See Bennett, Drugs and Crime.

------------------------------

DUF/ADAM Research

The DUF (Drug Use Forecasting) program, the forerunner of ADAM,
came about as a result of a 1984 study whose aim was to monitor the
behavior of arrestees released before trial. Based at the Manhattan Central
Booking Facility, the study was to compare pretrial misconduct of
arrestees found drug positive with those found drug free. A key question
was whether urinalysis, relatively new to the criminal justice system, could
be used in this setting to measure drug use. The project was successful in
that compliance rates were high: 95 percent of arrestees approached
consented to be interviewed. Moreover, urinalysis proved to be a feasible
method of testing for drug abuse: Of the arrestees who agreed to the
interview, 84 percent provided a urine specimen.[a] 

The value of urinalysis. Two years later, the researchers replicated the
study at the same site and again succeeded in achieving similarly high
response rates from the arrestees. A major but unintended outcome of this
initial study had been the revelation of a high level of drug use detected by
urinalysis at a time when self-reports were indicating much lower levels. In
the second study, there was another revelation: a substantial increase in the
use of cocaine (especially crack cocaine) since the first study (42 percent in
1984 compared with 83 percent in 1986). The researchers had identified a
trend in cocaine use more than a year before it was detected by any other
indicator of drug abuse in the United States (for example, new treatment
admissions, overdose deaths, and emergency room admissions).[b]

These two studies showed the feasibility of using urinalysis to test for drug
use among arrestees at the site where they are brought into custody. On the
basis of this finding, and because urinalysis was detecting higher levels of
drug use than was the traditional self-report method, the National Institute
of Justice established DUF in 1987 as a way to track drug-abuse trends in
this at-risk population.

Method and findings. On a quarterly basis, DUF/ADAM used urinalysis to
test for drug use by arrestees held in booking facilities at 23 sites
throughout the United States. The ADAM program collects data from 35
sites. Annual reports present findings for each site on arrestees' use of 3 of
10 main drug types (cocaine, opiates, and marijuana), along with use of
"any drug" and "multiple drugs." The reports include breakdowns of the
findings by age, gender, and race, as well as by type of charged offense.

The DUF/ADAM program continues to show a high level of drug use
among arrestees: In a majority of sites, more than 60 percent of all adult
arrestees test positive for one or more drugs. Between 40 percent and 60
percent of all adult arrestees test positive for cocaine and about 30 percent
test positive for marijuana. For opiates, by contrast, the majority of sites
report rates of less than 10 percent. These findings indicate not only the
magnitude of drug use in various urban areas of the United States, but also
trends in drug use. For example, in 1996 DUF clearly identified the decline
in cocaine use in New York City (Manhattan).[c] DUF data for that year
show rates of cocaine use peaked among males in the first quarter of 1995
and then fell fairly steadily in each quarter to a low of below 50 percent in
the third quarter of 1996. In view of the strong link between drugs and
crime, the findings of the DUF surveys have helped provide a more
balanced explanation of the recent rapid decline of crime in New York
City. 

Notes

a. Wish, E.D., and Gropper, B.A., "Drug Testing by the Criminal Justice
System: Methods, Research, and Applications," in Drugs and Crime, ed.
M. Tonry and J.Q. Wilson, Chicago: University of Chicago Press, 1990.

b. Ibid.

c. 1996 Drug Use Forecasting: Annual Report on Adult and Juvenile
Arrestees, Research Report, Washington, D.C.: U.S. Department of
Justice, National Institute of Justice, June 1997.

------------------------------

Research in England on Arrestee Drug Abuse

Although no surveys of arrestee drug use in England comparable to the
current study have previously been conducted, there have been surveys
based on interviews with prisoners and interview-based and observational
research involving arrestees.

Admitted drug use by prisoners. One of the largest surveys of prisoners
revealed information about drug use in the period before arrest. This study,
conducted in 1988, was based on a representative sample of 1,751
prisoners from 17 prisons in England and Wales.[a] Drug use was only one
of several issues about which the prisoners were asked. The study provided
some evidence of the proportion of prisoners who admitted drug use in the
2-day period before arrest (which approximates the 2- to 3-day window
covered by urinalysis). It revealed that 12 percent admitted to prearrest
marijuana use, 7 percent to opiate use, 4 percent to amphetamine use, and
2 percent to cocaine use.

Arrestee alcohol-related offenses and drunkenness. Among the few studies
of arrestees conducted in England, one focused on the prevalence of
alcohol-related offenses and drunkenness. Researchers observed the arrival
of people brought to 7 custody blocks in London in a 5-month period in
1993 and found that 22 percent of the 2,708 arrestees could be classified as
drunk on arrival (which would be comparable to testing positive for
alcohol).[b]

Arrestee drug involvement. A 1994 Manchester-based study aimed to find
information about drug involvement of arrestees brought into custody.
Officers working in the custody blocks of three police divisions completed
questionnaires for each arrest that probed for information about possession
of drugs, requests for medications while in custody, possession of
drug-using equipment, and other indicators of drug use.[c] As measured on
these criteria, the findings indicated that 19 percent of all arrestees were
deemed to be involved with drugs.

Other than these exceptions, research conducted in England on arrestee
drug use has been limited to indirect measurement and restricted to a small
number of drug types. The information obtained from these studies
suggests much lower rates of prearrest drug use than have been reported
by the DUF/ADAM program in the United States. However, until the
current research, the extent to which this discrepancy is the result of real
differences between the two countries or of different measurement methods
has not been clear.

Notes

a. Maden, A., M. Swinton, and J. Gunn, "A Survey of Pre-arrest Drug Use
in Sentenced Prisoners," British Journal of Addiction 87 (1992): 27-33.

b. Robertson, G., R. Gibb, and R. Pearson, "Drunkenness among Police
Detainees," Addiction 90 (1995): 793-803.

c. Chatterton, M., G. Gibson, M. Gilman, C. Godfrey, M. Sutton, and A.
Wright, Performance Indicators for Local Anti-Drugs Strategies: A
Preliminary Analysis, Police Research Group Crime Detection and
Prevention Series: Paper No. 62, London: Home Office, 1995.

------------------------------

STUDY METHOD--MATCHING THE SAMPLES

A major concern in developing the analytic methods was to devise a
procedure that produced roughly comparable datasets for the United States
and England. We were resigned to the fact that it would not be possible to
retrospectively generate wholly comparable survey samples, but we
believed we could move some way toward creating datasets similar enough
to produce useful comparisons. The main goal of the analysis was to be
able to make reasonable statements about the rate of drug use and the
nature of drug use in the two countries among detained arrestees
possessing roughly comparable characteristics. Specifically, we aimed to
determine whether there were differences in drug use among detained
arrestees in the two countries after we controlled for the various
differences in research locations and arrestee characteristics.

The samples for the two countries were to some extent comparable at the
outset. This was because although the NEW ADAM program predated the
creation of the I-ADAM uniform data-collection standards, it was designed
to match very closely the methods and procedures used in the U.S. ADAM
program. A number of steps were then taken to enhance this comparability.
Because of the various data manipulations, the results of the study are
slightly different from those reported in the 1996 DUF/ADAM annual
report and the 1998 Bennett report of NEW ADAM research.[1]

Ensuring Eligibility of Study Subjects

The rules for selecting study subjects--the detained arrestees--were
basically the same in the two programs. To be included in the study sample,
an individual had to have been arrested and detained in a specific city
"catchment area" in 1996 (or early 1997) and to have been booked for an
"eligible" offense (described below). Certain categories of detainees were
excluded from the sample:

--Those unfit for the interview because they had consumed alcohol,
drugs, and/or medication.

--Those considered mentally disordered.

--Those who were potentially violent.

--Those who were in custody more than 48 hours.

--Those deemed ineligible for other reasons, at the discretion of the
jail/custody sergeant.

Drug cutoff level. The amount of a drug in the urine below which the
arrestees were not considered drug positive was made comparable in both
countries. This was done by adjusting the English "cutoff" levels to match
those used in the United States. (See table 1.) Because of this adjustment,
the levels used in this study are different from those used in the 1998
Bennett study cited earlier, and as a result the findings reported here will
differ somewhat from those in that study.

Selecting Matched Sites

It is reasonable to assume that the nature and pattern of drug abuse and
other crime will vary with the type of location where the crime is
committed: a large city, a small city, or a rural area. Hence, comparison
between arrestees in London and Omaha, for example, may reveal more
about differences in drug use between these two cities than between drug
use in England and the United States. In other words, the greater the
differences in the cities being compared, the greater the likelihood that any
differences in drug use between the two countries will be a result of
city-level rather than country-level factors.

Criteria for matching the cities. A variety of methods were considered for
matching 5 of the 23 U.S ADAM sites to the 5 English sites, with the
ultimate decision to use population density alone as the criterion. The other
methods were rejected because of the number of problems they posed. One
of these alternative methods (discussed in appendix A) expands the
matching criteria from population density alone to an additional eight
measures (five more demographic variables and three measures of crime).

Although the eight-measure alternative produces some different pairs of
matching cities, it does not substantively change the results. The main
bivariate and multivariate results of urinalysis using the five cities finally
selected are nearly identical with the results obtained using the alternative
five matched cities. In fact, when urinalysis findings from all 23 U.S.
ADAM sites were examined, they were also nearly identical to those of the
5 matched cities used in this report and to the 5 alternative cities. The basic
findings from urinalysis therefore seem to be robust and not particularly
sensitive to the criteria for matching cities. The matched sites (whose
locations are shown on map 1, "NEW ADAM Sites," and map 2, "ADAM
Sites") are listed in table 2:

With the five matched U.S. cities identified, the data on the remaining cities
were removed from the original U.S. data file (leaving data for five English
cities and five matched U.S. cities).

Selecting Matched Study Subjects

In much the same way as the cities, the study subjects in both countries--
the detained arrestees--had to be made comparable. The first task in
matching the study subjects in the two datasets was to exclude certain
categories of arrestees that could not be matched. The samples then were
weighted to further refine the match.

Exclusions. Not surprisingly, the crimes for which suspected offenders
could be held in England were slightly different from those in the United
States. The list of English offenses[2] contained none for which someone
could not be arrested and detained in a jail in the United States. However,
some cases in the U.S. data file involved offenses for which a person would
not be detained in an English lockup facility. These cases, numbering
1,081, were excluded from the outset as ineligible for this study. The
excluded categories, all covering relatively less serious offenses, were
"release on own recognizance violation," "flight from jail," "possession of
liquor," "trespassing," and "violation of an Order of Protection." 

Interviews conducted in a language other than English were also excluded
from the data files. In England, arrestees who do not speak English and
who required an interpreter were deemed ineligible and were not
interviewed. The 206 interviews in another language (typically Spanish)
found in the U.S. dataset were therefore omitted to more closely match the
English dataset. 

Weighting the sample. Retrospective statistical weighting was used to
ensure that the distribution of various demographic and related factors in
these two samples was similar. Gender, age, race, and offense type were
the factors chosen because they were expected to be important predictors
of drug use. Four proved to be about the maximum number that could be
used as a basis for weighting the data.[3]

Before the samples were weighted, they differed from each other in the
proportions of arrestees charged with various crimes. In the U.S. sample,
for instance, 35 percent of the detained arrestees had been charged with
personal crimes (that is, violent crimes such as robbery); in the English
sample, that figure was only 16 percent. In the U.S. sample, 37 percent had
been charged with property crimes; in the English sample, that figure was
much higher at 53 percent. (The full breakdown of offenses by country is
presented in table 3.)

Demographically the two samples were also different: 79 percent of the
U.S. sample was nonwhite, compared with 15 percent of the English
sample. Men constituted 81 percent of the U.S. sample and 86 percent of
the English sample. The arrestees in England and the United States also
differed in age. (See table 4.)

The U.S. data were weighted to match the percentage distribution of cases
in the English dataset. For example, if 10 percent of the English sample
consisted of white males under 30 years old who were arrested for personal
crimes, the weighting system would create that same 10 percent
distribution of white males with the same characteristics in the U.S. sample.

The final, postexclusion, postweighting count was 4,470 detained arrestees
for the 5-city sample from the United States and 839 detained arrestees for
the 5-city sample from England.

Does the Sampling Method Affect Comparability?

At least in part, the data in both countries were collected using a system of
nonprobability-based sampling. (Three of the five surveys in England used
nonprobability sampling and two used probability sampling). Strictly
speaking, this type of sampling violates the assumptions of standard
significance tests because it is not technically possible to calculate standard
errors or confidence intervals for the estimated coefficients and
consequently not possible to determine whether any difference between
two samples is the result of sampling error or is due to other factors.

However, it is possible to estimate the extent to which, using the
procedures described above, nonprobability sampling generates a sample of
arrestees representative of the population of all arrestees in each country. If
it could be shown that such a method generated a sample closely matching
the population from which it was drawn, it might be reasonable to use
significance tests as a rough guide to the nature of the differences
observed.

Sampling methods of ADAM and NEW ADAM. There is some evidence to
suggest that samples selected in the ADAM and NEW ADAM programs
were fairly representative of their populations. In a study specifically
designed to investigate the method used in the DUF program, the
researchers concluded, " ...the current DUF procedures appear to select a
sample of interviewees that is highly representative of arrestees who are
detained in the particular booking centers where the DUF program
operates."[4] However, they noted that because the DUF selection
procedures rule out minor offenses, the samples are not wholly
representative of all arrestees reported in the FBI's Uniform Crime Reports.

A study of the convenience sampling used in two of the three English sites
examined whether the method affected the representativeness of the data.
The researcher found that the samples were very similar demographically to
the populations of the sites from which both were drawn. There were,
however, more substantial differences between the sample and the
population it was drawn from in the time of day of arrest and (to some
extent as a result) the nature of the offense.[5]

Although some forms of nonprobability sampling (such as quota sampling
of passersby in a shopping center) might produce quite unrepresentative
samples of the general population, other forms are more likely to produce
fairly representative samples of the populations under investigation. The
studies of the DUF method and the method used in the two English sites
tend to show that the kind of nonprobability sampling used in the United
States and English programs produces fairly representative results.

Tests of statistical significance. Reassured by these findings, we conducted
the analyses on the assumption that (after excluding ineligible cases) the
samples are fairly representative of the populations studied. Therefore, we
felt it appropriate to use tests of statistical significance to identify which
differences between the two countries might be considered meaningful. As
an added precaution, we sought primarily to identify only the more
substantial distinctions between the two as likely to represent "real"
differences.

Do Sample Size and Related Factors Affect Comparability?

As noted earlier, the size of the samples became smaller after the area-level
and individual-level matching procedures excluded certain categories of
arrestees. In the U.S. dataset, the remaining number of cases was generally
large enough to provide sufficient statistical power. Although a large
number of U.S. cases were lost through the exclusions, the English sample
sizes were smaller still, and for this reason the U.S. dataset was adjusted
(often by eliminating cases) to match the English dataset. (Table 5 shows
the numbers of cases left after the various exclusions.)

In the final, combined dataset for the two countries, the largest proportion
of cases are from the five U.S. sites. (See table 6.)[6] Because the samples
in each of the U.S. and English surveys were large enough to generate
sufficient statistical power to guard against Type II errors,[7] the fact that
they are not the same size is not especially problematic. However, because
the U.S. sample is larger, the U.S. data have more statistical power and are
more likely to generate significant differences in demographic subgroups
than are the English data.

Differences in the proportion of interviewees who agreed to provide a urine
specimen might also differentiate the two samples and thus affect the
results. The U.S. database includes only detained arrestees who provided a
urine specimen, because detained arrestees who agreed to be interviewed
but refused to provide a specimen were excluded. However, the English
database includes detained arrestees who agreed to be interviewed but
refused to provide a specimen. Of the 839 detained arrestees who agreed to
be interviewed, 74 percent also agreed to provide a urine specimen, for a
total of 621 cases available for urinalysis.

To what extent were the English arrestees who gave a urine specimen
different from those who did not? Comparison of those who did and did
not give urine samples revealed some small differences: Females were
significantly less likely than males to give a specimen and nonwhites were
significantly less likely than whites to do so.[8] Age made no difference.

These small differences suggest that the inclusion of nongivers in the
English sample might make it to some extent unequivalent with the U.S.
sample (although these differences would have been adjusted to some
extent by weighting). Clearly, it is important that information about the
characteristics of respondents and nonrespondents is recorded to allow for
more accurate adjustments when making comparisons of this kind. This
information was collected only in Nottingham and Sunderland in England.
Procedures have been adopted to collect it in the United States.

Finally, because not all the survey questions were answered by every
arrestee interviewed, there are varying numbers of missing values for some
of the survey items. In most cases, however, the number missing was small.

Notes

1. 1996 Drug Use Forecasting: Annual Report on Adult and Juvenile
Arrestees, Research Report, Washington, D.C.: U.S. Department of
Justice, National Institute of Justice, June 1997, NCJ 165691; and Bennett,
T.H., Drugs and Crime: The Results of Research on Drug Testing and
Interviewing Arrestees, Home Office Research Study No. 183, London:
Home Office, 1998.

2. In Bennett, Drugs and Crime: 100-101.

3. When the number of variables was expanded to five or more, the number
of categories with missing values in one or both countries grew
unacceptably high. Even this conservative number of variables (gender [2
values], age [2 values], race [2 values], and offense [5 values]) resulted in
40 cells. Four of these 40 cells (nonwhite females over 30 years old who
were arrested for alcohol/drug crimes; for disorderly offenses; or for
"other" offenses; and nonwhite females under 30 years old who were
arrested for alcohol/drug crimes) contained some cases in the U.S. dataset
(n=324), but no cases in the English dataset.

4. Chaiken, J.M., and M.R. Chaiken, "Understanding the Drug Use
Forecasting (DUF) Sample of Adult Arrestees," unpublished report
prepared for the National Institute of Justice, Washington, D.C.: 1993: 45.

5. Bennett, T.H., Drug Testing of Arrestees in England and Wales: The
Effect of Convenience Sampling on the Representativeness of the Results
Obtained in Cambridge, Cambridge, England: Institute of Criminology,
1997; and Bennett, T.H., Drug Testing of Arrestees in England and Wales:
The Effect of Convenience Sampling on the Representativeness of the
Results Obtained in Hammersmith, Cambridge, England: Institute of
Criminology, 1997.

6. Table 6 shows the distribution of cases after implementation of all the
various area-level and individual-level matching procedures (4,470 cases
for the U.S. sites and 839 cases for the English sites). It should be noted
that after all cases excluded from the study were filtered out, but before the
data were weighted, the 4,470 cases from the U.S. database were
distributed as follows among the five U.S. cities: New York=1,088 cases;
Washington, D.C.=967 cases; Ft. Lauderdale=864 cases; Birmingham=852
cases; and Miami=699 cases. Only after weighting the data was the
distribution of cases presented in table 6 achieved.

7. A Type II error occurs when a researcher fails to reject a null hypothesis
if it is actually false.

8. Bennett, Drugs and Crime.

FINDINGS--SIMILARITIES AND DIFFERENCES

The major finding of this study has to do with the prevalence of drug use
among detained arrestees in the United States and England. (The term
"prevalence" is used here to refer to the proportion of the detained arrestee
population that used drugs in a specified period of time.) Prevalence was
examined by different measures--urinalysis and self-reports[1]--and in
different time periods--3 days before the arrest, in the past month, and in
the past year.

Results of Urinalysis

The starting point was an analysis of the raw, unadjusted data.[2] It
revealed that a large percentage of the detained arrestees in both countries
consumed drugs in the 3-day period before arrest. (See chart 1.) The data,
presented for all 23 ADAM sites and the 5 NEW ADAM sites, indicate
that for most of the selected drugs, larger proportions of detained arrestees
in England than in the United States tested positive (marijuana:
X[squared]=35.7, p<.001; opiates: X[squared]=57.9, p<.001;
benzodiazepines: X[squared]=19.8, p<.001; and methadone:
X[squared]=48.3, p<.001).

For certain drugs, that overall picture of higher use in England changes.
For use of amphetamines and for multiple drugs, there were no statistically
significant differences between detained arrestees in the two countries
(amphetamines: X[squared]=0.5, multiple drugs: X[squared]=0.4). Cocaine
was the only drug for which prevalence was higher in the United States
than England (X[squared]=311.5, p<.001). At 40.4 percent, the U.S.
cocaine rate was more than four times higher than England's 8.7 percent. In
fact, the higher cocaine rate contributed substantially to the higher rate in
the United States (66.3 percent, compared with England's 59.1 percent) for
consumption of any of the six drugs tested (X[squared]=13.6, p<.001).
Although arrestees in England have a larger "working repertoire" of drugs
(for four of the six selected drugs, their use rate is higher), the overall
percentage of arrestees who use "any" drug is higher in the United States.

One of the basic goals of this report is to explore whether the differences in
prevalence are "real" or are artifacts of the two countries' different
research methods. As discussed in the chapter "Study Method--Matching
the Samples," a number of steps were taken to create roughly comparable
datasets for the United States and England. Cases deemed ineligible in
either the U.S. or English databases were excluded from the combined
dataset. The data from the two countries then were weighted by gender,
age, race, and type of crime to make them as similar as possible.

The resulting, adjusted prevalence rates (see chart 2) are fairly similar to
the unadjusted rates, with some distinctions. In England, both the adjusted
and unadjusted rates were higher than those of the United States for
marijuana (X[squared]=8.7, p<.01), opiates (X[squared]=48.3, p<.001),
and methadone (X[squared]=17.4, p<.001). However, amphetamine use,
which was the same in both countries when the unadjusted data were used,
became significantly higher in England (X[squared]=71.8, p<.001) when
the adjusted data were used. The prevalence of benzodiazepines, higher in
England with the unadjusted data, is no different in the two countries when
the adjusted data are used (X[squared]=0.46). The unadjusted rates
showed cocaine and "any drug" use to be higher in the United States than
in England. With the data adjusted, the rate of cocaine use and use of "any
drug" remained higher in the United States (X[squared]=290.8, p<.001 and
X[squared]=20.1, p<.001, respectively), but the rate of multiple drug use
(the same in both countries when unadjusted data are used) became higher
in the United States (X[squared]=8.6, p<.01). (The adjusted rates for each
drug in each of the 23 U.S. sites and each of the 5 English sites are
presented in appendix B, table B-1.) 

Differences in drug use by demographic characteristics and type of crime.
To find out whether there were differences in drug use among various
subgroups of the detained arrestees, the results of the urinalysis were
broken down by gender, age, race, employment status, and type of crime
for which the person was arrested. (See table 7.) As noted in the chapter
"Study Method," because of the effect of sample size on statistical power,
it is easier to find these subgroup differences in the United States dataset,
which is larger (4,470) than the English dataset (839). The emphasis in this
section is on identifying subgroups whose drug use prevalence clearly
exceeded the national averages in each country.

Gender. Studies based on self-reports and official studies based on criminal
justice processing tend to show that females are generally less deviant than
males (with some exceptions). Some of the early analyses of DUF data
showed, however, that female detained arrestees were just as likely as male
detained arrestees (and sometimes more likely) to test positive for certain
types of drugs.[3] Because the DUF research was conducted solely in the
United States, the extent to which similar gender differences would be
found among arrestees in England was unknown.

The current study reveals that in both countries (and for most types of
drugs) female arrestees were just as likely as or more likely than males to
test positive. The only specific drug for which the proportion of male
arrestees testing positive was higher than the proportion of females testing
positive was marijuana, which was true for both countries (in the United
States, 43.8 percent of the men and 21.7 percent of the women; in
England, 49.3 percent of the men and 26.8 percent of the women). Men
were also more likely than women to use any of the six drugs for which
urinalysis was conducted (in the United States, 69.0 percent of the men and
63.6 percent of the women; in England, 59.9 percent of the men and 52.1
percent of the women).

Overall, the two countries were found to be very similar with respect to
drug use and gender; of the specific drugs, only benzodiazepines showed a
measurable difference between the two countries. In the United States, the
rate of benzodiazepine use was higher among females than males (15
percent and 8 percent, respectively; X[squared]=29.4, p<.001). In England,
benzodiazepine use by females and males was similar (7 percent and 8.4
percent, respectively; X[squared]=0.16 [nonsignificant]). For multiple
drugs, there was a difference between the two countries; in England,
women arrestees had a slightly higher rate than did men (29.6 percent,
compared with 20.8 percent; although the difference is not statistically
significant), while in the United States, women had slightly lower rates than
did men (24.5 percent, compared with 27.6 percent; again, the difference
was not statistically significant).

Age. For age, the pattern is clear. In both countries, older arrestees (age 21
or older) were more likely than younger ones (age 20 or younger) to test
positive for most drug types. Only for marijuana use were younger people
more likely than older people to test positive. Thus, in the United States,
64.8 percent of younger arrestees compared with 34.8 percent of older
arrestees tested positive for marijuana (X[squared]=256.8, p<.001); in
England, 56.2 percent of younger arrestees compared with 43.4 percent of
older arrestees tested positive for this drug (X[squared]=8.3, p<.01). In
neither country was there a statistically significant difference in the
proportion of younger and older arrestees on the measure of "any drug." In
the United States, benzodiazepines and multiple drugs were the exceptions,
with no statistically significant differences between young and old arrestees
(benzodiazepines, X[squared]=3.3, NS; multiple drugs, X[squared]=2.0,
NS). 

Race. Only small differences by race were detected in the two countries.
The only statistically significant racial difference identified in both countries
was for marijuana use. In the United States, 40.2 percent of white arrestees
and 44.3 percent of nonwhite arrestees used this drug (X[squared]=3.7,
p<.05); in England, 44.1 percent of white arrestees and 58.8 percent of
nonwhite arrestees used it (X[squared]=5.2, p<.05). No other common
significant racial difference in drug use was found in both countries. When
the data from each country were examined separately, some significant
racial differences were found, however. Thus, in England but not in the
United States, white arrestees were significantly more likely than nonwhite
arrestees to have used amphetamines (6.2 percent and 0 percent,
respectively: X[squared]=8.1, p<.01). In the United States but not 
in England, white arrestees were significantly more likely than nonwhite
arrestees to test positive for benzodiazepines and multiple drug use.

Employment status. There is a strong association between employment
status and use of various types of drugs. In both countries, unemployed
arrestees were significantly more likely than employed arrestees to test
positive for a range of drugs. The only exceptions were marijuana and
amphetamine use in the United States, for which no significant difference
between unemployed and employed arrestees was found, and amphetamine
use in England, for which there was also no difference by employment
status. 

The size of the effects of unemployment are especially large in England.
For example, more than four times as many unemployed arrestees as
employed arrestees (22.6 percent, in contrast to 4.9 percent) tested positive
for opiates, and more than three times as many unemployed arrestees as
employed arrestees tested positive for cocaine (10.8 percent, in contrast to
3.0 percent). In the use of benzodiazepines, the difference between
unemployed and employed arrestees in England was particularly striking:
18 times as many unemployed arrestees as employed arrestees tested
positive for this drug (10.8 percent, in contrast to 0.6 percent). Three times
more unemployed than employed arrestees tested positive for multiple
drugs.

Type of crime. The offenses for which arrestees were charged were
categorized as personal crimes, property crimes, alcohol/drug offenses,
public disorder offenses, and other offenses. The relationship between drug
use and type of crime was different in the two countries. In England, with
two exceptions, those charged with property crimes tended fairly
consistently to be the group among whom the proportions of drug-positive
tests were highest, but in the United States, with one exception, the highest
proportions of positive tests were among those charged with alcohol/drug
offenses. That single exception in the United States was marijuana: Among
those who tested positive for marijuana, arrestees charged with an "other"
offense had the highest use rate (49.8 percent). Arrestees charged with an
alcohol/drug offense were the second highest marijuana-using group (47.9
percent tested positive).

Differences in the use of each drug. Whereas the previous section examined
differences in drug use among various demographic groups and by types of
crime, this section "profiles" various categories of arrestees. It focuses
particularly on groups showing the highest prevalence in each country for
each drug. Thus, in the previous section, drug use rates of men and women
in both countries were examined; in this section, these two groups are
further broken out by age, race, employment status, and type of crime.
(The data are drawn from the following tables: four breakdowns of these
trivariate results [appendix B, tables B-2 through B-5] and a summary of
the national averages for both countries [table 8, based on chart 2], which
presents the highest prevalence rate for each drug from each bivariate
category indicated [from table 7], and a summary of the group with the
highest prevalence rates for each drug from each of the trivariate
subcategories in appendix B, tables B-2 through B-5.)

The picture that emerges is of few subgroups in either country that clearly
eclipse their respective country averages. For example, in England,
methadone use is about 6 percent overall, but among female arrestees ages
31-35, the rate is more than 30 percent. In the United States, cocaine use is
40.7 percent overall, but for arrestees ages 31-35 who are unemployed, the
vast majority are using cocaine (80.9 percent). In the United States, for six
of the eight drug categories (from table 8's "Highest Use Category"),
arrestees ages 31-35 were found to be the highest drug-consuming group.
In fact, nearly all detained arrestees in the United States in this age group
were using at least one of the six specific drugs (84.6 percent). Arrestees in
the United States who are in this age group and are unemployed are in the
highest risk group for consuming cocaine (more than 80 percent), opiates
(almost 30 percent), any of the six selected drugs (almost 90 percent), and
multiple drugs (slightly less than 55 percent).

In England, for five of the eight drug categories (from table 8's "Highest
Use Category"), the highest drug-consuming group is slightly younger
(ages 26-30) than that in the United States. However, for four of the eight
drug categories (from table 8's "Highest Use Subcategory"), English
arrestees ages 31-35 are the highest drug-consuming group when this
group includes females or nonwhites. English female detained arrestees
ages 31-35 were the subcategory with the highest use of opiates (slightly
less than 45 percent), methadone (33 percent), and multiple drugs (nearly
45 percent). English nonwhite detained arrestees ages 31-35 were the
subcategory with the highest use of consuming any of the six tested drugs
(more than 90 percent).

The groups showing the highest use of each drug are profiled in the
sections that follow.[4]

Marijuana. In the United States, overall marijuana use among detained
arrestees was 40.6 percent. However, as noted earlier, almost 65 percent of
arrestees ages 15-20 in the United States tested positive for marijuana. The
rate of marijuana use for this age group is even higher among arrestees
charged with certain offenses. Of those ages 15-20 charged with public
disorder offenses, 89.7 percent used marijuana, and of those charged with
"other" offenses, 84.5 percent used this drug. In England, the overall rate
of marijuana use was 46.9 percent. As in the United States, younger
arrestees ages 15-20 were among the groups with the highest use (56.2
percent--the subgroup with the second highest prevalence rate in England).
The category with the highest overall rate of marijuana use in England was
nonwhite detained arrestees (58.8 percent). Within this group, the highest
rates were among older detained arrestees--nonwhites ages 31-35 (81.8
percent). This high number could be due to the small number of cases in
this category (9 of the 11 people in this category tested positive for
marijuana). Nonwhite arrestees charged with public disorder offenses were
the subcategory having the second highest rate of marijuana use in England
(75.0 percent).

Opiates. In the United States, overall opiate use among detained arrestees
was 8.4 percent. The category with the highest rate of opiate use was older
arrestees ages 31-35 (16.4 percent). Within this age group, those who were
unemployed had even higher rates (at 28.4 percent, this was more than
three times the U.S. average). In England, the overall rate of opiate use
among detained arrestees was 17.9 percent. As in the United States, the
category with the highest opiate use was older arrestees. However, in
England this older age group was 26-30 (33.7 percent of whom tested
positive for opiates), not those 31-35 years of age. Among the
subcategories examined, the highest rate of opiate use was among
women ages 31-35 (44.4 percent), with women ages 26-30 (40.0 percent)
a close second.

Cocaine. In the United States, 40.7 percent of the detained arrestees
overall tested positive for cocaine (including "crack"). The highest rates
were among the category of older arrestees ages 31-35 (64.6 percent). In
this age group, the subcategory of those who were unemployed had even
higher rates (80.9 percent). In England, a much lower percentage--8.7--of
the detained arrestees overall tested positive for cocaine. The highest rate
of cocaine use in England was among arrestees ages 26-30 (17.3 percent).
Of all subcategories, women charged with one of the least serious offenses
(the "other offense" category) had the highest cocaine use rate (33.3
percent).

Amphetamines. The overall rate of amphetamine use among detained
arrestees in the United States was 0.5 and was not much higher for any
subcategory. For arrestees ages 31-35, the rate was highest, at 1.6 percent.
However, examining the subcategories reveals that the overall highest
prevalence was among women in this age group, 13.6 percent of whom
tested positive for amphetamines. In England, 5.3 percent of detained
arrestees overall tested positive for amphetamines, with higher rates among
the subcategory of arrestees charged with an alcohol or drug offense (14.8
percent). The subcategory with the highest rate of amphetamine use in
England was female arrestees charged with an alcohol or drug offense
(50.0 percent).

Benzodiazepines. In the United States, 9.0 percent of detained arrestees
overall tested positive for benzodiazepines. The highest rate of
benzodiazepine use in the United States was, again, the category of older
arrestees ages 31-35 (17.4 percent). Notably, examination of the
subcategories reveals that the overall highest rate of benzodiazepine use
was among younger arrestees ages 15-20 who were charged with one of
the least serious offenses (the "other offense" category; 32.8 percent of
them tested positive). In England, 8.2 percent of detained arrestees overall
tested positive for use of benzodiazepines. However, of those ages 26-30,
15.4 percent tested positive for these substances. Within this age group, the
rate was even higher among those who were unemployed (20.5 percent).

Methadone. In the United States, 2.8 percent of detained arrestees overall
tested positive for methadone, with the highest prevalence, 6.8 percent,
among the oldest group (36 years of age or older). Within this age group,
those who were unemployed had even higher use rates (13.5 percent). In
England, 6.3 percent of detained arrestees overall tested positive for
methadone. Among English arrestees ages 26-30 the rate was 14.4 percent.
The highest methadone prevalence was among female arrestees ages 31-35
(33.3 percent).

Used any of the six drugs at least once. Among detained arrestees in the
United States, 68.3 percent overall tested positive for one of the six drugs,
with the highest rates among older arrestees ages 31-35 (84.6 percent). In
this age group, the rates were not much higher among any particular
combination of subcategories (although at 88 percent, the rate among
arrestees ages 31-35 who were unemployed was slightly higher). In
England, 59.1 percent of detained arrestees overall tested positive for use
of any of the six drugs. The category of English arrestees ages 31-35 had
the highest positive rate at 65.9 percent. The subcategory of English
detained arrestees with the highest prevalence on this measure was
nonwhites ages 31-35 (90.9 percent).

Multiple drugs. The overall rate of multiple drug use (measured by two or
more positive tests) in the United States was 27.2 percent. The category
with the highest rate (41.0 percent) was older arrestees ages 31-35. In this 
age group, those who were unemployed had even higher rates of multiple
drug use (54.6 percent). In England, 21.7 percent of all detained arrestees
tested positive twice, with the highest rate (35.6 percent) among those ages
26-30. The subcategory with the highest prevalence was female arrestees,
44.4 percent of whom, ages 31-35, tested positive at least two times.

Does the Country Make a Difference?

The analyses revealed that the level of drug use varies substantially among
demographic groups and depending on the type of offense. It is difficult to
discern from these analyses whether, independent of gender, age, and the
like, there would be differences between the two countries. In other words,
it would be illuminating to find out what additional influence the variable
"country" has on drug use rates, after controlling for demographic
characteristics and type of offense.

To find out if "country" makes a difference, separate logistic regression
models were developed for each of the eight drug measures. (See table 9.)
They revealed statistically significant differences between England and the
United States in arrestee use of opiates, cocaine, amphetamines,
methadone, any drug, and multiple drugs. Thus, the finding, discussed
above (see chart 2), that use of opiates, amphetamines, and methadone was
significantly lower in the United States than in England was confirmed by
the findings of the logistic regression analysis. The finding that cocaine use,
use of any drug, and multiple drug use were significantly higher in the
United States also was supported by the logistic regression analysis.

The bivariate finding of no difference between the two countries in
benzodiazepine use also was confirmed at the multivariate level. When the
logistic regression was run for benzodiazepines, the finding of no difference
between the two countries was confirmed (in the bivariate analysis, 9.0
percent of arrestees in the United States tested positive, as did 8.2 percent
in England, X[squared]=0.46, NS; in the multivariate analysis, Beta=0.14,
NS).

Marijuana was the sole exception, as the bivariate and multivariate analyses
yielded different results. For this drug, the multivariate analysis had
indicated that "country" was not statistically significant (Beta=-0.14, NS),
while the bivariate analysis had shown marijuana rates to be significantly
lower in the United States than England (40.6 percent and 46.9 percent,
respectively) (X[squared]=8.7, p<.01). The logistic regression did not
sustain the bivariate finding. The bivariate finding of lower rates of arrestee
marijuana use in the United States was probably the product of a measured
or unmeasured demographic characteristic or type of crime that
independently explained the relationship.

Other than for marijuana, the analysis supports all the differences between
the two countries revealed at the bivariate level. Specifically, for the use of
cocaine, any drug, and multiple drugs, rates were higher in the United
States, while for the use of opiates, amphetamines, and methadone, rates
were higher in England. The multivariate test confirmed the bivariate
finding of no difference between the two countries in benzodiazepine use
by detained arrestees.

Comparing Urinalysis and Self-Reports

Previous research has shown there is often a discrepancy between
self-reporting and urinalysis as measures of recent drug use. To investigate
whether or to what extent there is such a discrepancy for detained arrestees
in the United States and England, the results of both types of measures
were compared. Specifically, the results of the urinalysis were
compared with the arrestees' report of drug use in the 3 days before the
interview. (See table 10.)

Overall, for more than 90 percent of the detained arrestees in the United
States and England (90.3 percent and 91.3 percent, respectively) the
findings of the self-report survey and the urinalysis were in agreement. The
rest either underreported or overreported drug use (as measured by
urinalysis).

In the United States, the rate at which drug use was underreported (that is,
failure to report drug use when the urinalysis was positive) was higher than
in England. Overall, 7.8 percent of detained arrestees in the United States
underreported drug use compared with 1.9 percent who overreported (that
is, they reported using drugs, but the urinalysis was negative).[5]
Underreporting in the United States was especially evident for marijuana
(16.6 percent) and cocaine use (16.6 percent). In England, 3.7 percent of
arrestees underreported drug use, while 5.0 percent overreported. In
England as in the United States, the greatest disparity between self-reports
and urinalysis was for marijuana. Slightly more than 10 percent of the
detained arrestees in England overreported marijuana use, and 7.1 percent
underreported marijuana use.

The problem of relying solely on self-reports is highlighted in the findings
on benzodiazepine use. In the United States, underreporting of
benzodiazepine use was substantial (the urinalysis indicated a rate of 9.0
percent; the self-reports 4.4 percent). There was some overreporting of
benzodiazepine use in England (the self-reports indicated a rate of 10.3
percent; the urinalysis 8.2 percent). The net result of the two discrepancies
is that there is no difference in benzodiazepine use by arrestees in the two
countries as measured by urinalysis; but as measured by self-reports,
arrestees in England had significantly higher benzodiazepine use than their
counterparts in the United States.

For the other five drugs, the "direction" of the results did not change as it
did for benzodiazepine, and the results of the urinalysis and self-reports
were compatible. For example, as measured by both urinalysis and
self-report data, England had higher rates than the United States for
marijuana (bivariate results only), opiates, amphetamines, and methadone
but lower rates than the United States for cocaine.

The main difference between the two sets of results lies in the magnitude of
the difference between the countries. For example, as noted, cocaine use
was higher in the United States whether measured by urinalysis or
self-reports. However, urinalysis indicated that 40.7 percent of U.S.
arrestees and 8.7 percent of English arrestees tested positive for cocaine,
while the self-reports indicated that 25.2 percent of U.S. arrestees and 7.9
percent of English arrestees tested positive for cocaine.

Why the discrepancy? There are a number of reasons for a discrepancy
between urinalysis and self-reports. One is based on the argument that
urinalysis is more accurate. It includes the assumption that interviewees
might be unwilling or unable to disclose precisely the amount of drugs
consumed at various times. A second reason is based on the argument that
urinalysis and self-reports measure different things and that neither is more
accurate. Urinalysis can measure only drugs that have been consumed
within a specific period and have reached a certain point in the body's cycle
of metabolism. This argument is highlighted most clearly in the case of
marijuana use, which might be detected by urinalysis as long as a month
after consumption. A third reason is based on the argument that urinalysis
is less accurate than self-reports. According to this argument, technical
matters related to the cross-reactivity and specificity of the tests affect the
outcome.


Results of Self-Reported Drug Use 

Self-report surveys of arrestees' drug use can be important checks on the
results of urinalysis. However, they can be much more. They can supply
information about the history of drug use and can measure the use of types
of drugs not tested with urinalysis (for example, barbiturates, LSD, and
inhalants). Self-reports also can help distinguish between, and provide
additional information about, the use of crack cocaine and powder cocaine.

Arrestees were asked whether they used any of the drugs tested by
urinalysis and also were asked about their use of barbiturates, LSD,
inhalants, and alcohol (for a total of 10 drugs). These self-reports revealed
much the same as urinalysis: A majority of detained arrestees in both
England and the United States used drugs. In each country, more than 80
percent of the detained arrestees had used 1 of the 10 selected drugs at
least once in their lifetime. (See table 11.) More than 65 percent of
arrestees in each country said they used at least one of the selected drugs in
the past year and more than 60 percent had used a drug in the past month.
In England, more than 55 percent of detained arrestees said they had used
at least one drug in the 3 days before the interview, and slightly less than
50 percent of arrestees in the United States said they had used at least one
drug in the same period.

In England, three of the drugs had been used by more than half the
arrestees at some point in their life (81.6 percent had used marijuana, 64.6
percent had used amphetamines, and 54.6 percent had used LSD). In the
United States, marijuana was the only drug that more than half the
arrestees admitted having used at some point in their life (79.2 percent),
although powder cocaine, used by 47.3 percent of arrestees at some point
in their life, was a close second.

The self-reports were fairly consistent overall throughout the various time
periods--lifetime, annual use, monthly use, and use in the past 3 days. In
the main, if arrestees in one country reported significantly higher rates of
use for a particular drug category than those in the other country, they
were higher for all four time periods. The only drugs for which these
findings were not statistically significant were marijuana (lifetime use),
powder cocaine (use in past year), inhalants (use in past month and past 3
days), and alcohol (use in past 3 days).

In general, rates of use as measured by self-reports were higher in England
than the United States for 7 of the 10 types of drugs (marijuana, opiates,
amphetamines, methadone, benzodiazepines, LSD, and inhalants) as well as
for alcohol. Note, however, that the much higher rate of self-reported
marijuana use in the past 3 days in England (47.0 percent, compared with
30.1 percent in the United States) might be partially explained by the fact
that arrestees in the United States underreported marijuana use by 9.5
percentage points more than English arrestees. For crack cocaine, powder
cocaine, and barbiturates, rates were higher in the United States.[6] For
three measures of powder cocaine use--lifetime use, use in the past month,
and use in the past 3 days--the differences between the two countries were
statistically significant beyond the .001 level.

For drugs measured by both self-reports and urinalysis and whose findings
could therefore be compared, the self-reports revealed much the same as
did urinalysis. That is, larger percentages of detained arrestees in England
than in the United States used marijuana (according to the bivariate
analysis only), opiates, and methadone, while larger percentages of
detained arrestees in the United States than in England used cocaine.

Extent of injection drug use. The biggest difference between the two
countries in self-reported injecting of drugs was in amphetamines. Among
detained arrestees in England, 16.3 percent said they had injected
amphetamines at some time in their life, compared with only 1.7 percent 
of those in the United States. (See table 12.) When the data were
recalculated to examine only arrestees who reported use of the drug, they
revealed that 25 percent of the amphetamine users in England had injected
it at some point in their life compared with 7 percent in the United States.

When it came to drugs other than cocaine, opiates, and amphetamines,
detained arrestees in England were also significantly more likely than those
in the United States to say they had injected drugs (3.7 percent, compared
with 1.7 percent). Cocaine, however, was much more likely to have been
injected by arrestees in the United States than in England (11.3 percent and
8.0 percent, respectively). When the cocaine data were recalculated to
include only self-reported cocaine users, the difference between the two
countries disappeared: 21 percent of the cocaine users in the United States
and the same percentage in England had injected cocaine in their lifetime.

For heroin injection, in the sample as a whole, there was no significant
difference between the two countries. But again, the data were recalculated
to include only self-reported heroin users. This time they showed that a
higher percentage of heroin users in the United States than in England
injected the drug (57 percent, compared with 37 percent)
(X[squared]=97.8***).

Drug treatment--extent and need. The detained arrestees in both countries
were asked whether they had ever been treated for drug abuse and for
alcohol abuse and whether they wished to be treated.[7] The questions
were asked of those who had reported using drugs and those who had
reported using alcohol at least once in their lifetime. Overall, the analysis
revealed that in both countries a substantial proportion of detained
arrestees were treated for drug abuse in the past and a slightly higher
proportion said they currently needed treatment. Of particular importance
is the fact that one-third of arrestees in the United States and one-fifth of
those in England felt they currently need treatment.

It is worth noting that there was no difference between England and the
United States in the percentage of detained arrestees who said they had
been treated for drug abuse. More than one-fourth of those in both
countries (28.2 percent in the United States and 26.0 percent in England)
reported having been treated for drug abuse at some point in their life
(X[squared]=1.6, NS). (Again, these findings were based on the number of
arrestees who reported using drugs at least once in their lifetime). For
alcohol the findings were similar; that is, there was no difference between
the two countries in the percentage of detained arrestees who said they had
been treated for alcohol abuse some time in their life. In the United States,
12.4 percent of detained arrestees who reported using alcohol also reported
having been treated for alcohol abuse, and in England, 11.0 percent said
they had received treatment (X[squared]=1.5, NS). 

When it came to detained arrestees' reported need for treatment, the two
countries were significantly different. In the United States, 33.3 percent of
the detained arrestees who said they had used drugs also reported that they
currently needed drug treatment, in contrast to 22.2 percent of those in
England (X[squared]=38.2, p<.001). In the need for alcohol treatment, no
statistically significant difference was found between the two countries
(X[squared]=1.3, NS). Among the detained arrestees who said they used
alcohol, 14 percent of those in the United States and 13 percent in England
felt they currently needed to be treated for alcohol abuse.

Drug-using "careers." The detained arrestees were asked how old they
were when they used drugs for the first time and how many years they had
been using drugs. For five of the ten categories of drugs the arrestees were
asked about (marijuana, powder cocaine, barbiturates, methadone, and
benzodiazepines), those in the United States began their careers at a
younger age than those in England. (See table 13.) For the most part,
however, these differences in age of initiation between the two countries
were only modest (the difference was only 1 or 2 years for each drug
category).

As expected, in both countries alcohol was, on average, the first controlled
substance that arrestees tried in their lifetime (age of first use was 14.0 in
England and 14.7 in the United States). Among the illicit drugs, marijuana
was, on average, the first that arrestees in the United States tried in their
lifetime (age 15.1), while inhalants were the illicit drug that English
arrestees tried first (age 14.1). The drug that arrestees in the United States
began to use latest in their lives was crack cocaine: The average age of
crack initiation was 23.6. Among English arrestees, crack was the drug
they used second to last: The average age of first use was 21.5.
(Methadone, with an initiation age of 22.7, was the drug that English
arrestees began to use latest in life.) Because the age of initiation of crack
was late, its "career" was shorter than that of any other drug among
arrestees in the United States (6.7 years) and that of any other drug except
methadone in England (4.8 years for both drugs).

Income sources and expenditures on drugs. Detained arrestees in both
England and the United States were asked to estimate their total annual
income from all legal sources, their total annual income from all illegal
sources, and the amount of money they spent annually on illicit drugs.
Other than converting British to U.S. currency,[8] this analysis posed a
number of methodological challenges.[9] One was the non-normal
distribution of each of the three measures (legal income, illegal income, and
amount of money spent on illicit drugs). Since parametric tests of statistical
significance all are based on the assumption of normal distribution, to
approximate normality the analysis was based on the log value of each of
these measures.[10]

The findings revealed differences between detained arrestees in England
and the United States on all three measures. English arrestees had higher
illegal incomes (log mean 2.8; unlogged mean $9,760) than U.S. arrestees
(log mean 1.7, unlogged mean $8,888) (F=351.9, p<.001). English
arrestees also spent more on illicit drugs (log mean 2.6, unlogged mean
$6,346) than U.S. arrestees spent (log mean 2.0, unlogged mean $4,629)
(F=82.7, p<.001). U.S. arrestees had higher legal incomes (log mean 3.6;
unlogged mean $13,469) than English arrestees (log mean 3.3, unlogged
mean $4,889) (F=44.5, p<.001). Notably, English arrestees had higher
illegal incomes than legal incomes, while the reverse was the case for U.S.
arrestees.

Arrestees who tested positive for drugs (by urinalysis) were compared with
those who tested negative to find out if there were any differences between
the two countries in their legal and illegal incomes and the amount they
spent on drugs. This analysis revealed few within-country differences in
total legal income among arrestees who tested positive for drugs and
arrestees who tested negative for drugs in both countries. (See table 14.)

Legal income. In England, marijuana use was the only statistically
significant predictor of arrestees' legal income, and in the United States,
the only statistically significant predictor of arrestees' legal income was
cocaine use. Arrestees who did not use marijuana had the highest legal
income (log value 3.47) of all those in the English sample, and arrestees
who used marijuana had the lowest legal income (log value 3.15). In the
United States, cocaine use was a good marker for an arrestee's legal
income level. Arrestees who tested negative for cocaine had the highest
legal income (log value 3.76) of those in the U.S. sample, and arrestees
who tested positive for cocaine had the lowest legal income (log value 3.4).

Arrestees in both countries who tested positive for at least one of the six
drugs for which urinalysis was conducted had lower legal incomes and
higher illegal incomes than arrestees who tested negative for all six drugs.
Arrestees in both countries who tested positive for two or more of the six
tested drugs had lower legal incomes than arrestees who tested negative for
multiple drug use.

Illegal income. In general, arrestees in both countries who tested positive
for drugs had higher illegal incomes and spent more money on illicit drugs
than arrestees who tested negative. The only exception was amphetamines.
In both countries, there was no difference in the amount of illegal income
they said they earned between arrestees who tested positive for
amphetamines and those who tested negative. There were also no
differences in expenditure on drugs between arrestees in the United States
who tested positive for amphetamines and arrestees who tested negative.
However, arrestees in England who tested positive for amphetamines spent
significantly more money on drugs than those who tested negative.

Methadone users among arrestees in England had the highest illegal
incomes in the English sample and spent the most money on illicit drugs.
The highest illegal income among arrestees in the United States was earned
by those who tested positive for cocaine (log value 2.24), and the highest
drug spending among arrestees in the United States was by those who
tested positive for opiates (log value 3.21). Arrestees in both England and
the United States who tested positive for amphetamines had the lowest
illegal incomes of all drug-using groups (log values 1.47 and 3.03,
respectively) and spent the least amount of money on illicit drugs (England
log value 2.99; U.S. log value 1.99).

Notes

1. See "Urinalysis Versus Self-Reports" for a summary discussion of the
advantages and disadvantages of each method of determining the level of
drug use.

2. The raw, unadjusted data are reported because this was the approach
used in the 1998 Bennett report of NEW ADAM research. Bennett, T.H.,
Drugs and Crime: The Results of Research on Drug Testing and
Interviewing Arrestees, Home Office Research Study No. 183, London:
Home Office, 1998.

3. Wish, E.D., and Gropper, B.A., "Drug Testing by the Criminal Justice
System: Methods, Research, and Applications," in Drugs and Crime, ed.
M. Tonry and J.Q. Wilson, Chicago: University of Chicago Press, 1990.

4. The data are in the following tables and charts: overall rates of drug use
(chart 2); use of each drug by gender, age group, race, employment status,
and type of crime (table 7); male and female arrestee use of each drug, by
age, race, employment status, and type of crime (appendix B, table B-2);
use of each drug by younger and older offenders, broken down by race,
employment status, type of crime, and gender (appendix B, table B-3); use
of each drug by whites and nonwhites, broken down by gender, age,
employment status, and crime type (appendix B, table B-4); and use of
each drug by employment status, broken down by gender, age, race, and
crime type (appendix B, table B-5). Each presents the data for both
countries.

5. Overreporting could occur because arrestees exaggerate their drug use
or "telescope" their use over a long period into use over a shorter period,
or, alternatively, urinalysis fails to detect low levels of recent marijuana
use.

6. Note, however, that the difference between the two countries in
past-year use of powder cocaine (22.9 percent in the United States
compared with 20.3 percent in England) was not statistically significant.

7. We would like to have examined issues related to the nature of drug
treatment, including length and type of treatment. With the new I-ADAM
survey instrument, these issues can be explored.

8. The conversion rate was 1.653.

9. One problem noted was the lack of reliability/validity of the
DUF/ADAM data on the question of income. In the U.S. sample, 14.2
percent of the arrestees reported no legal income (compared with the
English rate of 3.8 percent). When illegal income was factored in, the U.S.
figure for no income was 6.0 percent (compared with the English rate of
1.0 percent).

10. Normal distributions will have values for skewness and kurtosis that are
close to zero. See Norusis, M., SPSS for Windows-Base System User's
Guide, Release 5.0, Chicago: SPSS, Inc., 1992: 167. The legal income
source variable (for both countries combined) had a skewness value of 21.1
and kurtosis of 669, the illegal income source variable (for both countries
combined) had a skewness value of 10.3 and kurtosis of 337, and the
money spent on drugs variable (for both countries combined) had a
skewness value of 7.7 and kurtosis of 74.3. Because the statistical
technique used to analyze these variables is based on normal distributions,
the log value of each of these measures was taken to create normality.
After the logarithmic transformations, the legal income source variable had
a skewness value of -1.5 and kurtosis of 1.2, the illegal income source
variable had a skewness value of 1.1 and kurtosis of -0.4, and the money
spent on drugs variable had a skewness value of -0.5 and kurtosis of -1.3.
The apparent normality of the data following the logarithmic
transformations permitted conducting analysis of variance tests.

------------------------------

Urinalysis Versus Self-Reports

Both methods of measuring drug use have advantages and disadvantages.
In combination, the two can provide a fuller picture of drug use than either
would separately.

Urinalysis

The advantage of urinalysis is that it is an objective measure and does not
rely on respondent recall or honesty. However, it typically measures drug
use only in a period of 48 to 72 hours after consumption (with the
exception of marijuana, which may be detected in heavy users as late as a
month after consumption). It is unable to detect drugs used for longer
periods, and it cannot detect how often a drug was used.

Self-Reports

This measure, obtained from interviews, has the advantage of being able to
measure drug use in different periods--whatever period of time the
interviewer asks about. The disadvantage is that self-reports depend on
respondents' ability to accurately and truthfully recall their use of drugs.

------------------------------

DISCUSSION--TOWARD UNDERSTANDING THE WORLDWIDE
DRUG PROBLEM

The aim of this study was to compare drug use prevalence and patterns of
drug abuse by detained arrestees interviewed and subjected to urinalysis as
part of the ADAM program in the United States and the NEW ADAM
program in England and Wales. The study report began by noting the
widely held belief that crime rates and drug abuse are substantially higher in
the United States than in England and also that while for certain types of
crimes recent research has called this supposition into question, very little is
known about whether for drug abuse the belief is well founded. The current
research has shown that for detained arrestees, the belief is in most cases
unfounded.

Differences Depend on Type of Drug

For opiates/heroin, methadone, and amphetamines, drug use prevalence
among detained arrestees tends to be higher in England than the United
States. There is no significant difference between the two countries in
arrestees' use of benzodiazepines and marijuana (the latter at the
multivariate level, as revealed by the logistic regression [table 9]). Only for
cocaine/crack were prevalence rates found to be significantly higher in the
United States. The percentage of detained arrestees in the United States
who tested positive for cocaine was more than four-and-one-half times that
of their counterparts in England, and this in turn drives the overall arrestee
drug use rate of the United States to a higher level than in England--68.3
percent for consumption of any of the six selected drugs, compared with
59.1 percent.

Several notable correlations were found between rates of drug use and
various characteristics of the arrestees and the type of offense with which
they were charged. In a number of instances, the subgroups exhibiting the
highest rate of drug use were the same in both countries. Thus, in both
countries, older arrestees tended in general to be the group with the highest
rates of drug abuse, marijuana was more likely to be used by younger
arrestees, female arrestees were as likely as or more likely than males to
use certain types of drugs, and unemployed arrestees were more likely than
those who were employed to test positive for a range of drugs.

Various other aspects of drug abuse and the "lifestyles" of drug-abusing
arrestees were examined. One aspect was injection drug use. The analysis
revealed that in both countries, moderately high proportions of arrestees
used injection as a method of administering drugs, although there were
some distinct differences between the two countries depending on the type
of drug. There was also little difference between the two countries in the
age of initiation of drug use, although again, there were some differences
depending on the type of drug. The analysis of income sources and amount
spent on drugs revealed that English arrestees tended to spend more money
on drugs and to report higher levels of illegal income than their
counterparts in the United States. There was no difference between the two
countries in the proportion of arrestees who said they had ever been treated
for drug abuse, although more arrestees in the United States said they
currently needed treatment.

Possible Explanations 

That detained arrestees in England are significantly more likely than those
in the United States to have recently used opiates, methadone, or
amphetamines while those in the United States are more likely to have
recently used cocaine is a significant finding of the study. Assuming that
these differences are real, how can they be explained? Future research
could focus on supply-based and demand-based explanations.

The preference of arrestees in the United States for cocaine is possibly a
product of the proximity of that country to the major source of supply
(South America). It is also possible that the preference of arrestees in
England for heroin is a product of that country's geographic position on the
traditional trade routes of the main source countries of heroin and other
opium-based products (Burma, Pakistan, Laos, Cambodia, and Iran).

It is more difficult to identify supply-based reasons for the higher rates of
methadone use in England. In both countries, the main source of
methadone is treatment programs. To explain the higher rates in England
would require demonstrating either that legal methadone (prescribed to the
user) is more readily available than the illegal variety (purchased on the
black market) or that the most desirable form of methadone (linctus or
injectable ampoules) is more readily available. Although methadone
treatment programs are common in both countries, there is no readily
available information on their number. It may be that during the 1960s and
1970s, injectable methadone was more likely to be prescribed in England
than in the United States. This form of treatment has since declined,
however, and most methadone programs in England now prescribe
methadone linctus. Another unknown is whether it is more difficult to
smuggle the substance out of United States methadone clinics than out of
English clinics.

Supply-based reasons for higher rates of amphetamine use among detained
arrestees in England also are difficult to explain. Illegal use of
amphetamines increased rapidly in England in the 1950s and 1960s partly
because of the irresponsible prescribing practices of some physicians.
Amphetamines continue to be one of the classes of drugs most frequently
seized by the police and by Customs and Excise officers. They may have
been more widely available in this period in England than in the United
States, although since then the laws in England governing prescribing, and
the drug laws in general, have become more prohibitive.

On the demand side, cultural differences between the two countries may
possibly explain the preference of English arrestees for certain drugs and of
arrestees in the United States for others. This, too, would constitute a topic
for future research.

This attempt to explain the differences between the two countries is
admittedly speculative. It is intended simply to illustrate the types of
explanations that future research can explore. As difficult as it is to
assemble comparable international datasets, the task of explaining observed
differences may be even more complex.

Next Steps in Crossnational Research

In summarizing the main findings of the first attempt at crossnational
comparison of drug abuse by detained arrestees, this report demonstrated
that there is considerable potential in developing an international database
on drug abuse that can generate research-based information relevant to
both public policy and future fundamental research on the nature of drug
use by criminal populations. Hopefully this study will be followed by
further comparative analyses of drug abuse by these populations in the
United States and England and additional analyses from some of the other
countries that are collecting I-ADAM data.

Revealing differences between and among countries in drug abuse patterns
suggests the many uses to which the I-ADAM program can be applied.
This comparison of two countries is only one example. Fundamentally, I-
ADAM was designed to serve as a knowledge "platform" on which to build
greater understanding of the nature of the drug problem worldwide.
Identifying invariant factors that predict drug use in various countries may
bring us closer to understanding the nature of the drug epidemic. This type
of research might show, for example, that certain market conditions must
exist for the drug epidemic to thrive. The findings could then be used by
countries not experiencing the epidemic to plan for prevention to avert the
problem altogether.

For such research to proceed requires that I-ADAM expand to new sites.
With five countries (England, Australia, Chile, Scotland, and South Africa)
having secured internal funding to participate in I-ADAM and some having
begun data collection, the program has a base from which to build. The
planning time necessary to launch a new I-ADAM site often takes more
than 6 months. Aside from securing funding, the political will to embark on
this type of research project has to be mustered. Given the lead time
required for all this, mobilization and outreach have begun. The hope is
that sites can be developed worldwide; to achieve that end, I-ADAM will
be reaching out to a variety of international organizations for support.

------------------------------

APPENDIX A

Testing Alternate Criteria for Selecting Matched U.S. and English Sites

As explained in the chapter "Study Method--Matching the Samples,"
conducting the study required finding 5 cities from among the 23 ADAM
sites in the United States that best matched the 5 English cities. Population
density was selected as the criterion for finding matches, and the five U.S.
sites most closely aligned with the five English sites on that criterion then
were chosen. However, a set of eight variables also was considered as
alternate criteria and subjected to analysis. The set consisted of five
demographic factors (plus population density) related to population
characteristics and three related to crime. (See table A-1.) The
demographic factors were as follows:

--Population density (number of people per square mile).

--Percentage male.

--Percentage living in owner-occupied dwellings.

--Percentage white.

--Percentage ages 16-29.

--Percentage of adult males who were unemployed.

These demographic data were gathered for 1990 in the United States and
for 1991 in England--in both countries the year of the most recent national
census.

The types of crime considered as part of the set of eight alternate criteria
were number of burglaries, robberies, and vehicle thefts per 100 residents
in the population. These three types were selected because they were the
main types of crime that were defined in a similar way by the two countries.
These data were collected for the year 1996 (the period in which the
arrestee surveys were conducted in both countries) from FBI data in the
United States and from local police records in England.

Assessing the Alternate Criteria

A number of issues raised by the eight alternate measures led to the
decision to rely solely on population density as the criterion for matching
the sites. First, for the criteria percentage male and percentage ages 16-29,
there was very little variation among the sites, and when the analysis of the
alternate criteria was conducted, these two variables had no effect. In most
of the sites, about 48 percent of the population was male (with a standard
deviation of only 1 percent). In most of the sites, about 23 percent of the
population was between 16 and 29 years of age (with a standard deviation
of only 3 percent).

Definitional differences. For some of the measures, there were differences
in definitions between the two countries. How crime is defined was one.
The U.S. measure of crime rates is based on the number of offenses
reported to the police. In at least one of the English cities (Nottingham),
however, the three crime rate figures were based on a combination of
police reports and the findings of local victimization surveys. This means a
city such as Nottingham has artificially higher rates of crime than it would
if the rates were measured exclusively by police reports (such as the FBI's
Uniform Crime Reports). At the time this report was prepared, it was not
possible to obtain solely police-based crime rates for the smaller, research
catchment area of the English sites (the area from which the data were
collected).

The definitions of an owner-occupied dwelling also were incompatible. The
measure considered for use in this study was calculated by dividing the
number of households occupied by owners by the total number of occupied
households. The problem arose because the denominators are slightly
different in the two countries. In the United States, the definition of
occupied household excludes several types of living quarters, some of
which would be included in England: dormitories, bunkhouses, barracks,
and quarters in predominantly transient hotels and motels. In England, it
appears as though some types of transient hotels and motels could count as
owner occupied, while in the United States, they would not. This variation
could artificially inflate the English owner-occupancy rates.

In assembling unemployment data, a major problem was that they were
available at the catchment area of the five English cities for males only.
That made it possible to calculate only the male unemployment rate--
obviously a problem because the sample consists of women as well as men.
Moreover, of particular concern was the large differences between male
and female detained arrestees in drug use in the two samples--differences
that might obscure the result if a male-only unemployment rate were used.

The measure of the percentage of the population that was white was
defined as 1 minus nonwhite. The problem posed by this measure was that
in England and the United States, being nonwhite means different things. In
many U.S. cities, the nonwhite population includes sizeable proportions of
Hispanics and blacks. In the five English cities, the nonwhite population is
made up of many people of Indian and Pakistani backgrounds. Moreover,
belonging to one of these minority groups has different implications for
crime in the two countries. For one thing, in the United States, blacks are
disproportionately represented in arrest statistics, but in England, members
of racial minorities are not. Therefore, using percentage nonwhite as a
criterion for selected matched cities could produce dubious results.

Census limitations. Aside from the definitional problems, one of the main
barriers to conducting this matching exercise was the lack of readily
available census data at the relevant level for the English cities. The English
surveys were conducted in each case not for the entire city but only for
small subsections of it. Thus, census data were needed not for all London,
but only for the smaller subsection Hammersmith within London; for
Manchester, the subsection Trafford; for Nottingham, the subsection
Nottingham City Centre Division; for Sunderland, the subsection
Northumbria; and for Cambridge, the Southern Division of the
Cambridgeshire Police Force area.

The six demographic variables were the only ones available to the research
team at the smaller, city subsection level. That became a limitation and
prevented the use of other demographic variables related to, for example,
social disorganization (expressed, among other ways, as neighborhood
mobility or transiency), that might aid in locating better city matches. The
use of census data, which are often poor proxies for the true underlying
concepts being explored, also would be a limitation.

The value of population density as a matching measure. The best measure,
and the one selected, was population density (number of people per square
mile). In view of the problems the other measures raised, this was thought
to be the more conservative approach. It presented no crossnational
definitional problems, and data were available for both women and men.
Population density also could serve as a useful proxy for many
unmeasurable concepts. Thus, matching densely populated cities with other
densely populated cities and less densely populated cities with other less
densely populated cities could facilitate comparisons of other measures
related to the availability of and demand for drugs.

Testing the Alternate Criteria 

Despite the limitations of the demographic and crime rate variables, these
criteria were tested to determine how much (or if) the study findings would
be different using them rather than using population density alone to select
matched cities. At the very least, if the findings could be shown to be
robust using both sets of criteria for matching the sites, then the alternative
criteria for selecting the city pairs could be ruled out. The analyses were
conducted with the full knowledge that they do not completely resolve the
question of the "correct" matching criteria.

The alternate city pairs. The nine alternate matching criteria were tested by
conducting a cluster analysis, which consists of a multivariate-level analysis
that identifies the best match from a weighted average of all covariates.
The cluster analysis produces a distance measure (known as Euclidian
distance) that is the sum of the squared differences between the values for
the variables. The smaller the Euclidian distance, the closer the cities match
one another. Using the standardized z-score option within the cluster
analysis routine enabled all nine variables to play some nontrivial role in
determining the overall Euclidian distance between the city matches.[1]

The city pairs selected using population density as the matching criterion
were:

London--New York, New York

Sunderland--Washington, D.C.

Manchester--Fort Lauderdale, Florida

Cambridge--Birmingham, Alabama

Nottingham--Miami, Florida

Running the cluster analysis with the nine alternate variables retained two
of the population-density matches (New York and London; Fort
Lauderdale and Manchester) and generated three new ones. The three new
matches were:

Sunderland--San Antonio, Texas

Nottingham--Dallas, Texas

Cambridge--Omaha, Nebraska

The 18 U.S. cities remaining in the ADAM dataset were then removed and
all the individual-level matching techniques described in the "Study
Method" chapter of this report (excluding the ineligible offense types and
ineligible categories of arrestees, then weighting the alternative sample to
make it further match the English data) were applied.

Finally, the five pairs of cities matched using the population-density
criterion and the five pairs of cities matched using the alternative criteria
were compared with the 23 U.S. ADAM sites. The same individual-level
matching techniques used for this study were applied to the 23-city
sample.[2]

Findings by type of drug. The results obtained from the new matches (see
chart A-1) were very close to those that were based on population density
and that were presented in this report (and summarized in chart 2). For
marijuana, the rate among English detained arrestees, at 46.9 percent, was
higher than the rates obtained using all three U.S. samples of detained
arrestees. For the 5 original cities, the U.S. rate was 40.6 percent
(X[squared]=8.7, p<.01); for the 5 alternative cities, the rate was 41.2
percent (X[squared]=7.2, p<.01); and for all 23 U.S. cities, the rate was
41.0 percent (X[squared]=8.6, p<.01). For opiate use, in all three U.S.
samples, the rates were much lower than in the five-city English composite.
The latter rate was 17.9 percent; the rate for the 5 original cities was 8.4
percent (X[squared]=48.3, p<.001), the rate for the 5 alternative cities was
8.3 percent (X[squared]=49.2, p<.001), and the rate for the 23 U.S. city
composite was 8.2 percent (X[squared]=56.9, p<.001).

In cocaine use by detained arrestees, there was a fair amount of variation
among the three U.S. samples. For the 5 original cities, the rate was 40.7
percent (X[squared]=290.8, p<.001); for the 5 alternative cities, the rate
was 31.2 percent (X[squared]=166.7, p<.001); and for all 23 U.S. cities,
the rate was 29.5 percent (X[squared]=158.1, p<.001). 

In all three U.S. samples, the rate of cocaine use was much higher than the
8.7-percent rate of the five-city English composite. Thus, the substantive
finding that higher proportions of U.S. detained arrestees use cocaine than
English detained arrestees remains intact. Only the precise magnitude of the
effect is somewhat in doubt. Therefore, instead of being 4.7 times greater
in the United States (40.7 percent, compared with 8.7 percent), the
difference in cocaine use is perhaps slightly less (3.6 times greater,
calculated by dividing 31.2 percent by 8.7 percent). In any event, the
question of the precise magnitude of the difference between the countries
in cocaine use will never be resolved, no matter how good the matching
system.

The English detained arrestees' rate of amphetamine use, at 5.3 percent,
was significantly higher than the 0.5 percent for the five original cities
(X[squared]=71.8, p<.001) and the 2.0 percent for the five alternative
cities (X[squared]=20.9, p<.001). 

Although at 4.7 percent, the 23 U.S. cities' rate of amphetamine use was
lower than that of the English city composite, the result was not
statistically significant (X[squared]=0.4, NS [nonsignificant]). For
benzodiazepines, the situation was similar, with the two U.S. five-city
matches corresponding. Thus, there was no difference between England's
8.2 percent and the 9.0 percent in the United States as measured in the five
original cities (X[squared]=0.46, NS) or the 7.6 percent in the United
States as measured in the five alternative cities (X[squared]=3.4, NS).
However, data from the 23-city sample indicate that the United States had
a lower rate of benzodiazepine use than England (6.0 percent,
X[squared]=4.9, p<.05). For multiple drug use, there was a similar finding
of noncorrespondence with the 23-city composite sample. On the two U.S.
five-city matches, the findings corresponded. That is, detained arrestees in
the United States had significantly higher multiple drug use rates than
England. The rate in England, at 21.7 percent, was significantly lower than
in the United States, according to both the original 5-city measure (27.2
percent; X[squared]=8.6, p<.01) and the alternative 5-city measure (25.9
percent; X[squared]=5.6, p<.05), but the 23-city sample, at 24.3 percent,
indicated no difference between the two countries (X[squared]=2.1, NS).
However, as will be seen from the logistic regression, on the multivariate
level, the results do correspond for all three measures for amphetamines,
benzodiazepines, and multiple drug use.

For methadone, all three U.S. compiled samples revealed lower rates than
England's 6.3 percent. For the 5 original cities, the rate was 2.8 percent
(X[squared]=17.4, p<.001); for the 5 alternative cities, the rate was 2.2
percent (X[squared]=27.8, p<.001); and for all 23 U.S. cities, the rate was
1.4 percent (X[squared]=56.4, p<.001). The rates for "any drug" use were
significantly higher in all three U.S. samples than in the five-city English
composite. The English rate was 59.1 percent; the rate for the 5 original
cities was 68.3 percent (X[squared]=20.1, p<.001), the rate for the 5
alternative cities was 63.0 percent (X[squared]=4.4, p<.05), and the rate
for the 23 U.S. city composite was 65.1 percent (X[squared]=9.4, p<.01).
The effects of "country." Another main component of the analysis
conducted for this report were the logistic regression models (presented in
table 9). They examined the question of what remaining effects the variable
"country" has on drug use, independent of the effects of gender, age, race,
employment, and crime type. To find out whether the results of the study
were confirmed using the alternative 5 matched cities and the 23 U.S. city
sample, separate logistic regression models were estimated for all eight
drug measures.

The results of this analysis (presented in table A-2) were very close to
those presented in the report. That is, every result is in the same direction
as the results in the report and is either statistically significant or
nonsignificant in the same way as the results from the report.[3] For
example, for marijuana use, all three beta coefficients from the three
comparison samples are negative and nonsignificant. That is, in all three
cases, after controlling for the effects of gender, age, race, employment
status, and crime type, there are no differences between detained arrestees
in the United States and England in marijuana prevalence rates. There was
similar correspondence for all seven other drug types.

Usefulness of the Original Criterion Confirmed

Overall, even though use of the alternative matching criteria generates
some different city pairs, the substantive results do not change. That is, the
main bivariate and multivariate results of the urinalysis based on the five
cities selected for the study are close to or identical with the results arrived
at using the alternative five matched cities. Also, both matched samples
correspond fairly well with the 23 U.S. city sample. Some differences
emerged between the results of the bivariate analysis conducted with this
23-city sample and the results obtained from the two 5-city samples. (The
differences were for amphetamines, benzodiazepines, and multiple drug
use.) However, there was correspondence on the multivariate level in the 
logistic regression models.

Nonetheless, the analyses conducted with the alternative matches do not
definitively answer the question of whether the "best" city pairs were
selected. There is still the possibility that some other criteria not tested
could produce different results. However, the analyses demonstrated that
the results obtained from using at least one known alternative matching
system are robust.

Notes

1. We are grateful to one of the reviewers for noting the importance of
using a weighted Euclidian distance measure to ensure that all variables in
the cluster analysis play some nontrivial role in determining the city
matches.

2. Note that the estimates of drug use arrived at by using this 23-city
sample are different from those calculated by using the raw data (presented
in chart 1), because of the individual-level matching methods used.

3. To simplify the presentation, the results (see table A-2) are shown only
for the covariate "country" (the main variable of interest), although the
model was estimated using the additional covariates of gender, age, race,
employment, and type of crime.

------------------------------

For more information on the National Institute of Justice, please contact:

National Criminal Justice Reference Service
Box 6000
Rockville, MD 20849-6000
800-851-3420
E-mail: askncjrs@ncjrs.org

To access the World Wide Web site, go to
http://www.ncjrs.org

If you have any questions, call or e-mail NCJRS.