Beginning in 1996, the National Institute of
Justice (NIJ) initiated a major redesign of its multisite
drug-monitoring program, the Drug Use Forecasting (DUF) system (DRUG USE
FORECASTING IN 24 CITIES IN THE UNITED STATES, 1987-1997 [ICPSR 9477]).
The program was retitled Arrestee Drug Abuse Monitoring (ADAM). ADAM
extended DUF in the number of sites and improved the quality and
generalizability of the data. The redesign was implemented in the first
quarter of 2000. The original goal remained the same -- to determine the
extent of drug use in the booked arrestee population (that is, arrestees
brought to fixed booking facilities where digital or ink fingerprinting
and other processing took place) in a defined area at specified points
each year. However, the redesigned sampling protocol and instrument
extended ADAM's goals in the following ways: (1) to provide a suitable
probability-based sample of jails and arrestees to support prevalence
estimates of drug use and related behaviors in each ADAM site, (2) to
provide accurate estimates with confidence intervals that permit tests
of the significance of drug use trends, (3) to create a standardized
dataset on arrestees in multiple jurisdictions to allow cross-site
comparisons, (4) to expand the scope of DUF data to include other areas
of concern (treatment history, dependency/abuse assessment, drug
markets), (5) to provide a platform for distinguishing between arrest
and drug use practices and for drawing inferences about the total
population of hardcore or heavy drug users, including those not in the
current ADAM sample, (6) to provide data for policy responses to
substance abuse issues both locally and nationally, (7) to investigate
drug markets or purchases, including data on characteristics of the
market, conditions of purchase or exchange, and prices paid, (8) to
assess risk of alcohol and/or drug dependency, and drug and mental
health treatment experiences, and (9) to use common definitions and,
where possible, identical questions and response categories to allow
meaningful links between ADAM and other national data systems.
The ADAM program in 2003 used an expanded adult
instrument that was first implemented in 2000. This instrument was used
in adult booking facilities for male (Part 1) and female (Part 2)
arrestees. The ADAM program also continued to use probability-based
sampling for the adult male population, a procedure initiated in 2000.
Therefore, the adult male sample includes weights, generated through
post-sampling stratification of the data. The shift to sampling of the
adult male population in 2000 required that all sites move toward a
common catchment area definition, generally a county. The 39 ADAM sites
for the current year included Albany, New York (Capital Area),
Albuquerque, New Mexico (Bernalillo County), Anchorage, Alaska
(Anchorage Borough), Atlanta, Georgia (Atlanta), Birmingham, Alabama
(Jefferson County), Boston, Massachusetts (Suffolk County), Charlotte,
North Carolina (Charlotte Metro), Chicago, Illinois (Cook County),
Cleveland, Ohio (Cuyahoga County), Dallas, Texas (Dallas County),
Denver, Colorado (Denver County), Des Moines, Iowa (Polk County),
Honolulu, Hawaii (Oahu), Houston, Texas (Harris County), Indianapolis,
Indiana (Marion County), Las Vegas, Nevada (Clark County), Los Angeles,
California (Pasadena County), Minneapolis, Minnesota (Hennepin County),
Miami, Florida (Miami-Dade County), New Orleans, Louisiana (Orleans
Parish), New York, New York (Manhattan Borough), Oklahoma City, Oklahoma
(Oklahoma County), Omaha, Nebraska (Douglas County), Philadelphia,
Pennsylvania (County of Philadelphia), Phoenix, Arizona (Maricopa
County), Portland, Oregon (Multnomah County), Rio Arriba, New Mexico
(Rio Arriba County), Sacramento, California (Sacramento County), Salt
Lake City, Utah (Salt Lake County), San Antonio, Texas (Bexar County),
San Diego, California (San Diego County), San Jose, California (Santa
Clara County), Seattle, Washington (King County), Spokane, Washington
(Spokane County), Tampa, Florida (Hillsborough and Pinellas Counties),
Tucson, Arizona (Pima County), Tulsa, Oklahoma (Tulsa County),
Washington, District of Columbia (Washington, DC), and Woodbury, Iowa
(Woodbury County). The core instrument was supplemented by a facesheet,
which was used to collect demographic and charge information from
official records. Core instruments were used to collect self-report
information from respondents. Instruments were administered to persons
arrested and booked on local or state charges relevant to the
jurisdiction (i.e., not federal or out-of-county charges) within the
past 48 hours. Trained interviewers used a paper and pencil instrument
in a face-to-face setting in a secure and reasonably private area of the
booking facility. Interviews lasted a median of 17 minutes. Responses
were recorded by the interviewer at the time of the interview. At the
completion of the interview, the arrestee was asked to voluntarily
provide a urine specimen. The adult male and female data reflect all the
arrestees selected for an interview from the booking logs, including
those for whom only facesheet information was collected. The final
sample for each adult data file, however, is the subset of arrestees
that accepted and completed an interview. An external lab used the
Enzyme Multiplied Immunoassay Testing (EMIT) protocols to test for the
presence of ten drugs or metabolites of the drug in the urine sample.
All amphetamine positives were confirmed by gas chromatography/mass
spectrometry (GC/MS) to determine whether methamphetamine was used.
Local booking facilities provided a census of all adult males arrested
in each facility collecting data for the time period of data collection
in the target county. The census data are not in the public file but
were used to develop sampling weights for the male data.
A probability sampling plan was used for adult male
collection in all sites in 2003, which assured that the data truly
represented the male arrestee population, not simply an unspecified
proportion of that population. The goal of sampling was to represent
with known probability the likelihood that a male arrestee was selected
for an interview and to use that information to weight each sample case.
Additionally, ADAM's goal was to represent all days of the week and all
times of the day so as to avoid biasing the male sample against those
types of arrests and arrestees who are brought in during the period
interviewers were not collecting data (morning, after midnight, "slower"
days of the week). The final sampling goal was to represent all the
facilities in the target county -- small, large, suburban, urban, quick
release, etc. -- again to represent all types of offenders arrested and
booked on local and state charges within the past 48 hours. Each ADAM
site adopted one of four designs for sampling jails. ADAM resource
constraints, the number of jails in each county, and how male arrestees
were processed through those jails dictated the resulting plan for each
site. The single jail design applied to sites where all arrestees were
booked into a single jail and were being held pending pretrial release
or trial. In the single jail design, the site collected its entire male
sample in the single booking facility in the county. For counties with a
few booking facilities (typically six or fewer), a stratified jail
design was used. ADAM interviewers sampled arrestees in each of those
jails and were assigned to jails so that the site's male sample was
distributed across all booking facilities in the county and was roughly
proportionate to size based on bookings. For counties that had many
jails, ADAM adopted a stratified cluster sample design through which
facilities in the county were clustered by size into a small number of
strata. The site's sample was distributed across one or two facilities
in each cluster, proportionate to size. This design affords estimates
for all jails even though only some jails were included in the sample.
Finally, for situations in which a large number of jails quickly
transfer a selected group of arrestees to a central holding facility,
ADAM adopted a feeder jail design wherein interviewers sampled arrestees
as they were booked into the central facility. Interviewees selected at
the central facility represented arrestees at each of the "feeder"
jails. However, only certain types of offender (typically those charged
with serious crimes) were transferred. Therefore, interviewers also went
to selected feeder jails to sample male arrestees who did not get
transferred. ADAM created a process to sample male arrestees within a
jail that were booked at any time of the day or any day of the week with
a known probability of selection by splitting the booked population into
two parts. The stock comprised males who had been booked before the
interviewer arrived at the jail. Interviews were, in general, conducted
from 4 p.m. to midnight. The flow comprised males who were booked while
the interviewer was stationed at the jail. Flow data collection began
the moment the data collection team entered a facility and represented
the period of the day when bookings were at the highest point. Cases
were selected throughout the period as they were available from booking,
with the interviewer selecting the case booked closest to when his/her
previous interview was completed. This method ensured that the
interviews moved throughout the shift and thus represented the full time
period. For the same reason, when an interview target number was reached
before the end of the shift, interviewing continued until the time
period was over. Flow cases were selected from booking log or records
data maintained by law enforcement in the facility. The booking log was
also the source of the stock sampling. The interviewers in this case
arrayed the male arrestees listed as booked during the non-interview
times chronologically and took cases on an interval determined by the
target number of stock cases for that day. Facesheets were filled out
for all males who would be in the sample regardless of whether they were
eventually interviewed. Arrestees selected in the sampling were not
always still in the facility, making those remaining a potential biased
estimate of the true male population characteristics. This bias was
addressed in ADAM through weighting of cases. A convenience sample was
used when collecting data from the adult female (32 sites) population.
The sample of sites was not a probability-based sample. In other words,
both DUF and the subsequent ADAM sites were not sampled from a list of
counties in the United States. They were selected through applications
of sites that were interested in participating.
All persons arrested and booked on local and state charges
(i.e., not federal and out-of-county charges) in any of the 39 ADAM
catchment areas in the United States during 2003.
Individual arrestees.
A double-sided facesheet was used to collect information
from administrative records on all adult arrestees selected for an
interview. The ADAM adult interview instrument was used to record
information from voluntary, anonymous, and confidential interviews with
all male and female adult arrestees in the sample available for an
interview within 48 hours of the time of arrest. Urine tests were used
to collect clinical records data in order to detect the presence of
several drugs in specimens provided by the interviewee at the conclusion
of the interview.
administrative records data
clinical data
medical records
survey data
For the adult data (male and female, Parts 1 and
2), variables from the facesheet include arrest precinct, ZIP code of
arrest location, ZIP code of respondent's address, respondent's gender
and race, three most serious arrest charges, sample source (stock,
flow, other), interview status (including reason an individual
selected in the sample was not interviewed), language of instrument
used, and the number of hours since arrest. Demographic information
from the core instrument include respondent's age, ethnicity,
residency, education, employment, health insurance coverage, marital
status, and telephone access. Variables from the calendar provide
information on inpatient and outpatient substance abuse treatment,
inpatient mental health treatment, arrests and incarcerations, heavy
alcohol use, use of marijuana, crack/rock cocaine, powder cocaine,
heroin, methamphetamine, and other drug (ever and previous 12 months),
age of first use of the above six drugs and heavy alcohol use, drug
dependency in the previous 12 months, characteristics of drug
transactions in past 30 days, use of marijuana, crack/rock cocaine,
powder cocaine, heroin, and methamphetamine in past 30 days, 7 days,
and 72 hours, heavy alcohol use in past 30 days, and secondary drug
use of 15 other drugs in the past 72 hours. Urine test results are
provided for 11 drugs -- alcohol, marijuana, cocaine, opiates,
phencyclidine (PCP), benzodiazepines (Valium), propoxyphene (Darvon),
methadone, barbiturates, amphetamines, and methamphetamine. The
files also include several derived variables. The male data also
include four sampling weights, and stratum IDs and percents.
Among the ADAM-eligible male arrestees randomly
selected for interview, 56.7 percent agreed to an interview, 11.7
percent declined, 31.4 percent were not available to be interviewed at
the same time of selection due to prior release from custody, court
appearance, or other logistical reasons, and 0.3 percent were available
but not approached. Of the male arrestees who were interviewed, 91.4
percent provided a urine sample. Among eligible female arrestees
selected for interview on a convenience basis, 59.2 percent agreed to be
interviewed, 10.4 percent declined, and 30.4 percent were not available.
Of the female arrestees who were interviewed, 93.1 percent provided a
urine specimen.
None.