Title: 2000 Arrestee Drug Abuse Monitoring: Annual Report Series: Research Report Author: National Institute of Justice Published: April 2003 Subject: ADAM (Arrestee Drug Abuse Monitoring) Program, drug abuse, and drug testing 129 pages 309,000 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 ANNUAL REPORT 2000 ARRESTEE DRUG ABUSE MONITORING ---------------------------- U.S. Department of Justice Office of Justice Programs 810 Seventh Street N.W. Washington, DC 20531 John Ashcroft Attorney General Deborah J. Daniels Assistant Attorney General Sarah V. Hart Director, National Institute of Justice 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 ---------------------------- 2000 Arrestee Drug Abuse Monitoring: Annual Report April 2003 NCJ 193013 ---------------------------- NIJ National Institute of Justice Sarah V. Hart Director For their forthright and insightful comments on the draft, we are indebted to our colleagues at the Office of National Drug Control Policy, particularly Robert Eiss; at the National Institute on Drug Abuse, particularly Lynda Erinoff; and at the Executive Office for Weed and Seed, U.S. Department of Justice, particularly Robert Samuels. We would also like to acknowledge the contributions of ADAM's data management contractor, Abt Associates Inc, and the laboratory contractor, PharmChem, Inc. 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. ---------------------------- Contents Executive Summary Part I: Drug Use and Related Behavior: Findings o Chapter I--Overall Findings and ADAM Redesign o Chapter II--Drug Dependence and Treatment o Chapter III--Alcohol Use and Alcohol Dependence o Chapter IV--Drug Markets o Chapter V--Drug Use Among Adult Female Arrestees o Chapter VI--Drug Use Among Juvenile Detainees Part II: 2000 Findings, by Site-Adult Male Arrestees Part III: Applying the New ADAM Method o Chapter VII--Implementing the New ADAM Study Design at the Local Level o Chapter VIII--"Calendaring" in ADAM: Examining Annual Patterns of Drug Use and Related Behavior o Chapter IX--Estimating Hardcore Drug Use in the Community Appendix Tables and Exhibits Tables Chapter 1 o Table 1-1: Number of Weighted Cases, by Site--Adult Male Arrestees, 2000 o Appendix Table 1-1: Drug Test Results, by Drug by Site--Adult Male Arrestees, 2000 o Appendix Table 1-2: ADAM Sample Sizes, Interviews, and Urinalyses, by Site--Adult Male Arrestees, 2000 Chapter 2 o Appendix Table 2-1: Drug Dependence and Treatment Status, by Site-- Adult Male Arrestees, 2000 o Appendix Table 2-2: Adult Male Arrestees at Risk for Drug Dependence, by Site--Adult Male Arrestees, 2000 o Appendix Table 2-3: Demographics and Sociodemographics of Adult Male Arrestees at Risk for Drug Dependence, by Site, 2000 o Appendix Table 2-4: Adult Male Arrestees Who Ever Received Drug or Alcohol Treatment, By Selected Drugs, by Site, 2000 o Appendix Table 2-5: Demographics and Sociodemographics of Adult Male Arrestees Who Received Drug or Alcohol Treatment in Past Year, by Site, 2000 o Appendix Table 2-6: Adult Male Arrestees at Risk for Drug Dependence Who Received Treatment, by Site, 2000 Chapter 3 o Table 3-1: "Heavy" Alcohol Use-ADAM Definitions o Appendix Table 3-1: Binge Drinking in Past Year and Past Month, by Site-- Adult Male Arrestees, 2000 o Appendix Table 3-2a: Binge Drinking, Past Month, by Age and Race, by Site--Adult Male Arrestees, 2000 o Appendix Table 3-2b: Binge Drinking, Past Month, by Demographic and Sociodemographic Characteristics, by Site--Adult Male Arrestees, 2000 o Appendix Table 3-3: Binge Drinking, Past Month, by Level of Drinking, by Site--Adult Male Arrestees, 2000 o Appendix Table 3-4a: Adult Male Arrestees at Risk for Alcohol Dependence, Past Year, by Age Group by Site, 2000 o Appendix Table 3-4b: Adult Male Arrestees at Risk for Alcohol Dependence in Past Year, by Demographic and Sociodemographic Characteristics, by Site, 2000 o Appendix Table 3-5: Adult Male Arrestees at Risk for Alcohol Dependence, Past Month, by Level of Alcohol Consumption, by Site, 2000 o Appendix Table 3-6: Proportions of Adult Male Arrestees at Risk for Alcohol or Drug Dependence, by Age When Drinking Began, by Site, 2000 o Appendix Table 3-7: Drug Use in Past Month, by Level of Alcohol Use- Adult Male Arrestees, by Drug by Site, 2000 Chapter 4 o Table 4-1: Method of Contacting Dealer to Obtain Selected Drugs on Cash and Noncash Basis--Averages Among Sites--Adult Male Arrestees, 2000 o Table 4-2: Reasons Attempts to Purchase Drugs Failed--Averages Among Sites--Adult Male Arrestees, 2000 o Table 4-3: Cash and Noncash Transactions, Most Active Drug Market Sites--Adult Male Arrestees, 2000 o Table 4-4: Number of Times Per Day Arrestees Obtained Drugs, Most Active Drug Market Sites--Adult Male Arrestees, 2000 o Table 4-5: Number of Days in Past Month When Arrestees Obtained Drugs, Most Active Drug Market Sites--Adult Male Arrestees, 2000 o Table 4-6: Number of Drug Transactions Per Month, Most Active Drug Market Sites--Adult Male Arrestees, 2000 o Table 4-7: Percentage of Arrestees Who Generated More Than Half the Drug Transactions, Most Active Drug Market Sites--Adult Male Arrestees, 2000 o Table 4-8: Market Size (in Dollars) of Past-month Cash-only Transactions, Most Active Drug Market Sites--Adult Male Arrestees, 2000 o Appendix Table 4-1: Drug Market Participation in Past 30 Days, by Drug by Site-Adult Male Arrestees, 2000 o Appendix Table 4-2: Drug Transaction Type (Cash, Noncash, or Combination), by Drug by Site--Adult Male Arrestees, 2000 o Appendix Table 4-3: Methods of Obtaining Drugs by Noncash Transactions, by Drug by Site-Adult Male Arrestees, 2000 o Appendix Table 4-4: Methods of Contacting Dealer to Obtain Marijuana, by Site-Adult Male Arrestees, 2000 o Appendix Table 4-5: Methods of Contacting Dealer to Obtain Crack Cocaine, by Site-Adult Male Arrestees, 2000 o Appendix Table 4-6: Methods of Contacting Dealer to Obtain Powder Cocaine, by Site-Adult Male Arrestees, 2000 o Appendix Table 4-7: Contacts with Multiple Drug Dealers for Cash Purchases, by Drug by Site-Adult Male Arrestees, 2000 o Appendix Table 4-8: Regularity of Relationship with Drug Dealer for Cash Purchase, by Drug by Site-Adult Male Arrestees, 2000 o Appendix Table 4-9: Use of Couriers/"Go-Betweens" for Cash Purchases, by Drug by Site-Adult Male Arrestees, 2000 o Appendix Table 4-10: Outdoor Drug Purchases, by Drug by Site-Adult Male Arrestees, 2000 o Appendix Table 4-11: Outside-Neighborhood Drug Purchases, by Drug by Site-Adult Male Arrestees, 2000 o Appendix Table 4-12: Failed Drug Purchases, by Drug by Site-Adult Male Arrestees, 2000 o Appendix Table 4-13: Reasons Attempts to Purchase Drugs Failed, by Drug by Site-Adult Male Arrestees, 2000 Chapter 5 o Table 5-1: Average Age of Adult Female Arrestees, by Site, 2000 o Table 5-2: Drug Transaction Types (Cash and Other) for Marijuana and Crack Cocaine, by Selected Sites--Adult Female Arrestees, 2000 o Table 5-3: Outdoor Cash Purchases of Marijuana and Crack Cocaine, by Selected Sites--Adult Female Arrestees, 2000 o Table 5-4: Outside-Neighborhood Cash Purchases of Marijuana and Crack Cocaine, by Selected Sites--Adult Female Arrestees, 2000 o Table 5-5: Failed Cash Purchases of Marijuana and Crack Cocaine, by Selected Sites--Adult Female Arrestees, 2000 o Appendix Table 5-1: Sample Size--Adult Female Arrestees, 2000 o Appendix Table 5-2: Drug Test Results, by Drug by Site-Adult Female Arrestees, 2000 o Appendix Table 5-3a: Drug Test Results for NIDA-5 Drugs and Cocaine, by Age by Site--Adult Female Arrestees, 2000 o Appendix Table 5-3b: Drug Test Results for Marijuana and Opiates, by Age by Site--Adult Female Arrestees, 2000 o Appendix Table 5-3c: Drug Test Results for Methamphetamine and PCP, by Age by Site--Adult Female Arrestees, 2000 o Appendix Table 5-3d: Drug Test Results for Multiple NIDA-5 Drugs, by Age by Site--Adult Female Arrestees, 2000 o Appendix Table 5-4a: Drug Test Results--Adult Females Arrested for Violent Offenses, by Drug by Site, 2000 o Appendix Table 5-4b: Drug Test Results--Adult Females Arrested for Drug and Alcohol Offenses, by Drug by Site, 2000 o Appendix Table 5-4c: Drug Test Results--Adult Females Arrested for Property Offenses, by Drug by Site, 2000 o Appendix Table 5-4d: Drug Test Results--Adult Females Arrested for Driving While Intoxicated, by Drug by Site, 2000 o Appendix Table 5-4e: Drug Test Results--Adult Females Arrested for Domestic Violence Offenses, by Drug by Site, 2000 o Appendix Table 5-4f: Drug Test Results--Adult Females Arrested for "Other" Offenses, by Drug by Site, 2000 o Appendix Table 5-5a: Drug Test Results by Race/Ethnicity by Drug by Site-- Adult Female Arrestees, 2000 (Whites and Blacks) o Appendix Table 5-5b: Drug Test Results by Race/Ethnicity by Drug by Site-- Adult Female Arrestees, 2000 (Hispanics and "Other") o Appendix Table 5-6: Age and Race/Ethnicity, by Site--Adult Female Arrestees, 2000 o Appendix Table 5-7: Demographics and Sociodemographics, by Site--Adult Female Arrestees, 2000 o Appendix Table 5-8: Drug Use, Past 12 Months and Past 30 Days, by Drug by Site--Adult Female Arrestees, 2000 o Appendix Table 5-9: Extent of Heavy Drinking, by Site--Adult Female Arrestees, 2000 o Appendix Table 5-10: Need for Treatment, as Measured by Risk for Dependence and Injection Drug Use, by Site--Adult Female Arrestees, 2000 o Appendix Table 5-11: Treatment for Drugs, Alcohol, or Mental Health Problems, by Site--Adult Female Arrestees, 2000 o Appendix Table 5-12a: Treatment for Drugs and Mental Health Problems Among Adult Female Arrestees Who Used Cocaine, by Type of Treatment by Site, 2000 o Appendix Table 5-12b: Treatment for Drugs and Mental Health Problems Among Adult Female Arrestees Who Used Marijuana or Heroin, by Type of Treatment by Site, 2000 o Appendix Table 5-12c: Treatment for Drugs and Mental Health Problems Among Adult Female Arrestees Who Used Methamphetamine, by Type of Treatment by Site, 2000 o Appendix Table 5-13: Drug Market Participation in Past 30 Days, by Drug by Site--Adult Female Arrestees, 2000 o Appendix Table 5-14a: Methods of Obtaining Marijuana by Noncash Means, Selected Sites--Adult Female Arrestees, 2000 o Appendix Table 5-14b: Methods of Obtaining Crack Cocaine by Noncash Means, Selected Sites--Adult Female Arrestees, 2000 o Appendix Table 5-15a: Reasons Attempts to Purchase Marijuana Failed, Selected Sites--Adult Female Arrestees, 2000 o Appendix Table 5-15b: Reasons Attempts to Purchase Crack Cocaine Failed, Selected Sites--Adult Female Arrestees, 2000 Chapter 6 o Table 6-1: ADAM Sites Where Juvenile Detainees Participated--2000 o Table 6-2: Drug Test Results, by Drug by Site--Juvenile Male Arrestees, 2000 o Table 6-3: Drug Test Results, by Drug by Site--Juvenile Female Arrestees, 2000 Chapter 8 o Table 8-1: Self-Reported Crack Use, Selected Sites, by Calendar Period-- Adult Male Arrestees, 2000 o Table 8-2: Arrest Rates in Past 12 Months, Selected Sites--Adult Male Arrestees, 2000 o Table 8-3: Adult Male Arrestees Not Included in Other Measures of Drug Use and Related Behavior, Selected Sites, 2000 Chapter 9 o Table 9-1: Estimated Number of Hardcore Drug Users in the ADAM Sites-- Adult Male Arrestees, 2000 Exhibits Chapter 2 o Exhibit 2-1: Percentages of Drug-using Adult Male Arrestees at Risk for Dependence in Past Year, by Drug, 2000 o Exhibit 2-2: Participation by Drug-using Adult Male Arrestees in Drug or Alcohol Treatment or Mental Health Treatment--Ranges Among the Sites, 2000 o Exhibit 2-3: Percentages of Drug-using Adult Male Arrestees Who Ever Received Inpatient Drug or Alcohol Treatment, by Drug--Ranges Among the Sites, 2000 o Exhibit 2-4: Percentages of Drug-using Adult Male Arrestees Who Ever Received Outpatient Drug or Alcohol Treatment, by Drug--Ranges Among the Sites, 2000 o Exhibit 2-5: Hispanic Adult Male Arrestees: Percentages at Risk for Drug Dependence and Percentages with Health Insurance, Past 12 Months, Selected Sites, 2000 o Exhibit 2-6: Percentages of Drug-using Adult Male Arrestees Who Received Drug or Alcohol Treatment, Past 12 Months, by Race, Selected Sites, 2000 o Exhibit 2-7: Percentages Who Lack Health Insurance: Drug-dependent and Drug-using Adult Male Arrestees and All Adult Male Arrestees, Selected Sites, 2000 Chapter 3 o Exhibit 3-1: Binge Drinking in Past Month Among Homeless and Nonhomeless Adult Male Arrestees, by Site, 2000 o Exhibit 3-2: Levels of Heavy Alcohol Use, Past Month--Ranges Among the Sites--Adult Male Arrestees, 2000 Chapter 4 o Exhibit 4-1: Extent of Drug Market Participation in the Past Month, by Selected Drugs-Ranges Among the Sites--Adult Male Arrestees, 2000 o Exhibit 4-2: Drug Transaction Types (Cash and Other), by Selected Drugs-- Ranges Among the Sites--Adult Male Arrestees, 2000 o Exhibit 4-3: Noncash Drug Transactions Involving Gifts, by Selected Drugs-- Ranges Among the Sites--Adult Male Arrestees, 2000 o Exhibit 4-4: Use of Two or More Drug Dealers to Make Cash Purchases, by Selected Drugs--Ranges Among the Sites--Adult Male Arrestees, 2000 o Exhibit 4-5: Most Recent Cash Purchase of Drugs from a Regular Source (Dealer), by Selected Drugs--Ranges Among the Sites--Adult Male Arrestees, 2000 o Exhibit 4-6: Outdoor Purchases of Drugs-Ranges Among the Sites, by Selected Drugs--Adult Male Arrestees, 2000 o Exhibit 4-7: Drug Purchases Made Outside the Neighborhood, by Selected Drugs--Ranges Among the Sites--Adult Male Arrestees, 2000 o Exhibit 4-8: Failed Purchases, by Selected Drugs--Ranges Among the Sites-- Adult Male Arrestees, 2000 Chapter 5 o Exhibit 5-1: Drug Test Results--Ranges Among the Sites--Adult Female Arrestees, 2000 o Exhibit 5-2: Extent of Drug Market Participation in the Past Month, by Selected Drugs--Ranges Among the Sites--Adult Female Arrestees, 2000 Chapter 8 o Exhibit 8-1: Annual Rates of Arrest, by Selected Drugs, by Selected Sites-- Adult Male Arrestees, 2000 o Exhibit 8-2: Percentages of Adult Male Arrestees Who Used Heroin or Cocaine in Past Year, by Level of Use, New York--Adult Male Arrestees, 2000 o Exhibit 8-3: Patterns of Heroin Use for 3 Adult Male Arrestees in Year Before Arrest, New York--2000 o Exhibit 8-4: Heroin Use by "User B" in Context of Treatment and Involvement in Criminal Justice System, New York--ADAM Data, 2000 Chapter 9 o Exhibit 9-1: Hardcore Drug Users Are Assumed to All Have the Same Arrest Rate o Exhibit 9-2: The Estimate Requires Creating a Model of the Arrest Process o Exhibit 9-3: The Basic Logic of the Estimation Model Illustrated o Exhibit 9-4: Introducing Measured Heterogeneity into the Estimation o Exhibit 9-5: Changing the Calculation to Accommodate Measured Heterogeneity o Exhibit 9-6: Introducing Unmeasured Heterogeneity into the Estimation o Exhibit 9-7: Average Annual Arrest Rate of Hardcore Drug Users in a County o Exhibit 9-8: Hardcore Users in the Community Per Hardcore User in the Booking Population ---------------------------- Executive Summary When the National Institute of Justice (NIJ) established the Drug Use Forecasting (DUF) program in 1988, it was the first time an objective drug testing method would be routinely used to assess the validity of self-reported drug use among people charged with crime. DUF demonstrated that it is possible to conduct research on drug use among arrestees in the jail setting, and for many years the program provided information to policymakers and practitioners about drug use in the at-risk population of arrestees. Evaluations of DUF led NIJ to decide to strengthen the program by making the sampling procedure more scientifically sound, standardizing data collection, and instituting other changes. After several years of development and testing, the restructured program was fully implemented in 2000 as Arrestee Drug Abuse Monitoring (ADAM). Probability-based sampling was adopted, the interview instrument (questionnaire) was enhanced to cover several new areas of drug use and related behavior, and the number of sites was increased. The 2000 annual report reflects these changes. That means it departs from previous years' reports in some ways. As in the past, it presents information about arrestees' drug use, both overall and site by site. This year the report also features a series of chapters that examine in depth some of the new topics that are now a routine part of the questionnaire. The emphasis is on adult male arrestees, because probability-based sampling is currently used only for this population. As in the past, the report includes a summary table of data from each site, but this year the tables also show risk for drug and alcohol dependence, admissions to treatment, and drug market participation. Another series of essays documents the new ADAM method and explores possible new ways to apply it. The "audiences" for ADAM data are the same as in the past. For policymakers, there is a broad overview of drug use among the population at risk for crime. For practitioners in the justice system who deal day-to-day with drug use and related crime, ADAM offers information useful for planning control strategies. Practitioners in the ADAM sites can compare the drug-use profile of their jurisdiction with that of other sites. For researchers, the ADAM data offer myriad possibilities for investigating the drug-crime link. Overall findings and ADAM redesign In 2000, drug use continued to be common among adult male arrestees, as in previous years. The ADAM redesign strengthens the reliability of the findings and makes it possible to explore new areas of drug use and related behavior. o In half the 35 ADAM sites, urinalysis indicated that 64 percent or more of adult male arrestees had recently used at least one of five drugs: cocaine (undistinguished between crack and powder), marijuana, opiates, methamphetamine, or PCP (phencyclidine). Marijuana was the drug most commonly used, followed by cocaine. o The transition from DUF to ADAM in 2000 completed a major redesign of the program. One component of the redesign included enhancing the data collection instrument (the interview questionnaire) to ask about alcohol use, risk for dependence on drugs and alcohol, substance abuse treatment, and drug market participation, including how and where drugs are obtained. The number of sites in the ADAM program increased from 23 to 35 (including two "affiliated" sites1). Arguably the most important change was the adoption, at all ADAM sites, of probability-based sampling for selecting adult male arrestees. Drug dependence and treatment As part of the redesigned program, adult male arrestees' risk for dependence on drugs is measured, and they are asked about their experiences with treatment. o Between about one-fourth and one-half of all adult male arrestees in the ADAM sites were found to have been at risk for dependence on drugs. o Although a large percentage of adult male arrestees had not only used drugs but also were at risk for drug dependence, few had received treatment. Among the ADAM sites, the range in the proportions who said they were treated on an inpatient basis in the year before their arrest for either drugs or alcohol was 4 percent to 17 percent, and the range of those who had received outpatient treatment was 2 percent to 15 percent. o With few exceptions, adult male arrestees who were treated for drug or alcohol use in the year before their arrest were more likely than not to have no health insurance. Alcohol use and alcohol dependence Alcohol abuse can be associated with behavioral problems, including crime. ADAM asks adult male arrestees about their use of alcohol and measures their risk for dependence on it. o Adult male arrestees drank heavily. Among the sites, the proportions who had five or more drinks on at least one occasion in the month before their arrest ranged from a low of 35 percent to a high of 70 percent. Drinking at the level defined as "heaviest" was not uncommon: The proportions who had five or more drinks on one occasion on 13 or more days in the month before their arrest ranged from 10 percent to 24 percent. o Risk for alcohol dependence was measured by a special set of questions, or "screen." By this measure, more than four in five of the "heaviest" drinkers were at risk. In half the sites, 85 percent or more were at risk, with the range among the sites 67 percent to 91 percent. o The heaviest drinkers were also likely to have used illicit drugs. On average, 71 percent of them had used at least one drug in the month before their arrest. Drug markets The ADAM redesign makes it possible to obtain information about drug markets from a large number of buyers at the local level in many sites nationwide. Adult male arrestees were asked about the extent of their participation in drug markets, how and where they acquired drugs, what difficulties they encountered trying to do so, how often they obtained drugs, and the dollar value of the drugs. o In the 23 sites analyzed,[2] the market for marijuana was the largest, as measured by percentage of adult male arrestees who participated. Much smaller percentages participated in the markets for crack cocaine, powder cocaine, heroin, and methamphetamine. o Many arrestees participated in one or more drug markets. The majority reported little difficulty completing a drug transaction, saying such obstacles as police activity and lack of drug availability were not a problem. o Fairly large proportions of adult male arrestees did not rely solely on cash to obtain drugs, whether marijuana, crack cocaine, or powder cocaine. These noncash exchanges most commonly took place at a social setting or at work. In many sites, when arrestees paid cash for marijuana, the most common method of obtaining it was by using a phone or pager, and for crack cocaine it was by approaching a dealer in a public place. o In four high-volume sites (Miami, Phoenix, Seattle, and Tucson), the number of transactions in the crack market was much larger than in the powder cocaine and marijuana markets. In these sites, the estimated size (measured in dollars) of the crack cocaine market in a 30-day period was 2 to 10 times larger than the size of the powder cocaine and marijuana markets. The range among these sites in the market size of crack cocaine was about $226,000 to $1,400,000. Drug use among adult female arrestees Although only about one in five people arrested in the United States is a woman, and the proportion of women who commit drug offenses is even smaller, the number of women charged with drug offenses is not inconsequential. Research on women's involvement in drugs has been relatively limited, but the ADAM redesign offers the opportunity to expand research on their drug use and drug-related behavior.[3] o As in previous years, urinalysis revealed that a large percentage of women arrestees had used drugs. Cocaine was the drug for which the proportion testing positive was highest, with marijuana coming in second. o Of the women arrestees who used drugs or alcohol, about half were found at risk for drug dependence. o Only very small percentages of women arrestees had been treated for drug or alcohol use the year before their arrest. The average among the sites was 11 percent. Drug use among juvenile detainees Data on drug use were collected from male and female juvenile detainees in 8 of the 35 ADAM sites (Birmingham, Denver, Los Angeles, Phoenix, Portland, San Antonio, San Diego, and Tucson). Data were also gathered in Cleveland, but for juvenile male detainees only. The samples were not probability-based, nor were the interviews conducted with the expanded ADAM questionnaire.[4] o Juveniles were more likely to test positive by urinalysis for marijuana than any other drug. o Cocaine came in a distant second; the percentages testing positive for methamphetamine were also low. Implementing the new ADAM study design at the local level Implementing the new, probability-based ADAM study design involved adopting standardized data collection procedures among 35 sites. This entailed redefining the catchment areas (the area from which arrestees are drawn to participate in the program) to make them uniform among the sites, and designing sampling plans at the county level and the level of each facility to ensure that all arrestees have some probability of being included among those participating in the program. o In DUF, the definition of the catchment area varied from site to site, and often consisted of a single jail. In ADAM the catchment area was redefined as the county for all sites. o Data collection was redesigned to account for variations among the sites in the structure and size of local criminal justice systems and processes. The county-level sampling model adopted was flexible enough to be applied to the specific counties/sites. o The transition from DUF to ADAM showed that standardized protocols and probability- based sampling can be implemented in the dynamic environment of the jail. o Within one year of introducing the new sampling method, almost all the ADAM sites had successfully implemented it. That means they can now develop reliable prevalence estimates for a variety of drug-related issues, including the proportions of arrestees who test positive for drugs and those who need treatment. "Calendaring" in ADAM: examining annual patterns of drug use and related behavior A new feature in the ADAM interview instrument in 2000 is "calendaring," which permits analysis of drug use and related behavior for the period of a full year. Through memory aids built into the questionnaire, arrestees' behavior is examined month by month for the entire 12-month period of the survey. The technique can increase accuracy in arrestees' recall of drug use and related behavior. o Data from selected sites, when broken down by different periods of time in the year, demonstrated that recent drug use is not always a good measure of longer-term, more typical use. o The annual rates of arrest for individual arrestees can vary by type of drug used. o The ADAM redesign permits the data to be "crosswalked" with other annually conducted national surveys of drug use and related behavior. Analysis indicates that some of these surveys do not cover the subpopulation reached by ADAM. o The proportions of arrestees who used heroin and cocaine at least 15 days a month in every month of the year were higher than the proportions who used them less frequently (for example 1 to 7 days a month in each month). Estimating hardcore drug use in the community ADAM is developing a method that can be used to estimate the prevalence of hardcore drug use in the sites. Made possible by the adoption of probability-based sampling, the method infers prevalence in the community from the count of adult male hardcore users who are arrested and booked at the ADAM sites. Arrest rates are therefore key to the calculations. o Preliminary estimates indicate that, in most ADAM sites, there are 750 arrests and bookings a year for every 1,000 hardcore drug users and that the number of hardcore users ranges from just over 1,500 (Minneapolis) to almost 126,000 (New York). For sites where sampling takes place in several jail facilities, the numbers are likely underrepresentations, by perhaps as much as half. o Once the method has been refined, the ADAM sites should be able to use it to make their own calculations. Notes 1. ADAM's two affiliated sites--so called because they are funded by sources other than NIJ--are Charlotte/Mecklenburg County, North Carolina, and Albany/New York Capital Area. 2. This analysis was confined to the 23 sites where the markets for all three heavily used drugs--marijuana, crack cocaine, and powder cocaine--were most active. 3. Because the number of women arrested is much lower than the number of men, fewer are available for participation in ADAM. Some ADAM sites do not include women arrestees. The expanded ADAM questionnaire was used to interview the women arrestees, but probability-based sampling does not yet include them. 4. Juvenile detainees are interviewed with the DUF instrument (questionnaire), but the program is considering designing a new interview instrument for them, to collect information about drug treatment and participation in drug markets. ---------------------------- Part I: Drug Use and Related Behavior: Findings I. Overall Findings and ADAM Redesign With this year's annual report, the transition from the Drug Use Forecasting (DUF) program to the Arrestee Drug Abuse Monitoring (ADAM) program is complete. The findings reported here are from the redesigned ADAM program. ADAM was changed to make it more scientifically rigorous and to generate more information. In 2000 the changes were fully implemented. The goal is the same as before: to track drug use and related behavior among arrestees in many of the Nation's largest cities. ADAM remains the only program that does so by using urinalysis as an objective and accurate measure. The transition to ADAM involved major changes. To select participating adult males, probability-based sampling was adopted, and all ADAM sites now use standardized procedures to collect data. Several new topics were added to the questionnaire, and although that was done before on an ad hoc basis, these new areas of inquiry will continue. Finally, the number of sites is now 38, up from 23. The changes make this annual report different from those of previous years. As in the past, the report updates findings on arrestees' use of drugs, but this year it also explains how the new ADAM method was used to analyze the 2000 data, and in a series of essays the report examines some of the new topics (Part I). Information about arrestee drug use is presented site by site, as in previous annual reports (Part II). Another set of essays documents the new ADAM method and explores possible further ways to use it (Part III). If ADAM has changed dramatically, the "audiences" remain the same. For policymakers, ADAM offers a broad overview of drug use by people at risk for crime. For the police and other criminal justice practitioners at the individual sites who deal with drug use on a day-to-day basis, ADAM offers data useful for planning control strategies; and they can compare their site with the others. For researchers, ADAM offers a wealth of topics for investigating the drug-crime link. Extent of drug use as detected by urinalysis As in previous years, the levels of drug use detected were high. The urinalysis test used in ADAM can identify any of 10 substances, but the analysis focuses on the "NIDA-5" drugs (cocaine, opiates, marijuana, methamphetamine, and PCP).[1] (See "ADAM Drug Testing--the Procedure, the Drugs" for details of these drugs.) In half the ADAM sites that reported data, 64 percent or more of the adult male arrestees[2] had recently used at least one of these drugs. Use ranged from 52 percent of arrestees (Anchorage) to 80 percent (New York) (See Appendix Table 1-1.) For each drug there were major variations among the sites and regions. These are explored here. In each site there were also distinctive patterns, examined in the section profiling the sites. An analysis that combined data from many regions of the country into a nationwide picture of drug use by arrestees would mask these differences. The differences revealed by ADAM suggest a one-size-fits-all approach to controlling drug use may not be the optimal one, and policies and strategies for enforcement and treatment are best tailored to specific user groups and locations. Of the 10 drugs analyzed by ADAM through urinalysis, four--cocaine (both crack and powder), marijuana, methamphetamine, and opiates (heroin, for example)--were the ones used most often by adult male arrestees in most sites. Of these, marijuana was most commonly used, followed by cocaine, opiates, and methamphetamine, in that order. In half the sites at least 40 percent of the adult male arrestees tested positive for marijuana. Use was lowest in Laredo (29 percent testing positive), with Oklahoma City at the top of the range (57 percent testing positive). Large percentages of adult male arrestees recently used cocaine (undistinguished here between crack and powder). In half the sites, at least 31 percent tested positive, with the range between 11 percent (Des Moines) and 49 percent (Atlanta and New York). Many sites where the proportions testing positive for cocaine were relatively low (under 20 percent) were on the West Coast and in the Pacific Northwest. These include Sacramento and Salt Lake City (both 18 percent), Honolulu (16 percent), Spokane and San Diego (both 15 percent), and San Jose (12 percent). For methamphetamine, the West is where the proportions of adult male arrestees who used this drug were highest. In several Midwestern States as well, substantial proportions of arrestees tested positive for this substance. Confirmatory urinalyses[3] indicated the highest methamphetamine use (20 percent or more of adult male arrestees) was in Honolulu (36 percent), Sacramento (29 percent), San Diego (26 percent), San Jose (22 percent), Portland (21 percent), and Spokane (20 percent). Double-digit rates also showed up in Phoenix and Des Moines (both 19 percent), Las Vegas (18 percent), Salt Lake City (17 percent), and Oklahoma City and Omaha (both 11 percent). In some sites, urinalysis indicated no recent methamphetamine use. These sites, 8 in number, are largely in the eastern part of the country (Albany/New York Capital Area, Chicago, Detroit, Fort Lauderdale, Laredo, Miami, New York, and Philadelphia). In nine other sites, only between one-tenth of 1 percent and 1 percent of adult male arrestees tested positive. These two groups of sites, 17 in all, where 1 percent or fewer arrestees tested positive for methamphetamine, lower the median for all the sites.[4] Although that midpoint is only 2 percent (in half the sites, 2 percent or fewer tested positive), it does not obscure the fact that in 12 sites more than 10 percent of the arrestees were positive for methamphetamine. Only in a few sites were opiates used extensively. In most sites, few adult male arrestees tested positive for these substances (in half the sites, the proportion was 7 percent or fewer). The range was 2 percent of arrestees (Charlotte-Metro, Fort Lauderdale, and Omaha) to 27 percent (Chicago). In addition to Chicago, sites with double-digit opiate-positive rates were New York (21 percent), New Orleans (16 percent), Portland (14 percent), Philadelphia and Albuquerque (both 12 percent), and Birmingham, San Antonio, Laredo, and Seattle (each 10 percent). This distribution suggests no geographic pattern. PCP was used by only a small percentage of arrestees in most of the sites (in half the sites, the proportion who used it was 0.3 percent or less). This low rate is consistent with the findings of earlier DUF and ADAM reports. In only two sites in 2000 did 5 percent or more of the adult male arrestees test positive for PCP (Cleveland, 8 percent, and Oklahoma City, 5 percent), and in 12 sites no arrestees tested positive. Most adult male arrestees tested positive for only one of the five drugs. In half the sites, 21 percent or more tested positive for polydrug use, with the sites ranging from 10 percent of arrestees (Anchorage and Albany) to 34 percent (Chicago). For polydrug use the evidence should be interpreted cautiously, because the test detects only recent use. Studies have consistently shown past year or past month polydrug use the norm,[5] with users substituting one drug for another when the drug of choice is scarce, or mixing drugs to counter or moderate the effects of one or the other. The ADAM interviews can add to the information from urinalysis and reveal whether arrestees are using different types of drugs in the period of a month[6] or a year (and how frequently they are used). The new ADAM method The redesigned ADAM program provides better estimates of drug use and related behavior than it did previously.[7] Data collection is now based on probability sampling. The sample of arrestees at any site is selected in such a way that the findings become an accurate estimate of the proportion of all arrestees in the county who would test positive for drugs had all of them been interviewed and tested. This also means data for use in research projects at each site are stronger. And because the sites will be able to place the numbers within confidence intervals, trend analysis (year-to-year comparisons) will be more reliable and more easily interpreted than in the past. The year 2000 was the first time these probability-based samples were obtained for adult male arrestees. Some sites were unable to implement the new procedures as quickly and effectively as others. But at most ADAM sites, beginning in 2000, the data collected constituted statistically reliable estimates of the proportion of all male arrestees in the area who had used drugs within a specified time period. Plans are to develop probability-based sampling plans for female arrestees as well. Ensuring a representative sample The new sampling procedure ensures a representativeness not possible under the DUF program and during the first years of the ADAM program.[8] In each city, data were generally collected at only one lockup facility--the largest--and interviews were conducted with volunteers who had been arrested no more than 48 hours previously. DUF and ADAM staff tried to gain access to the facilities at times during the day when there was a large number of arrests, though these times varied considerably from site to site. As a result, the representativeness of the time period of data collection and of the resultant sample was unknown, and standard errors for the samples could not be calculated. With the introduction of probability sampling in 2000, which refined the procedures for when and where data collection would take place, ADAM gained greater scientific rigor in estimating drug use. Sample sizes and weighting The findings reported here come from 35 of the 38 ADAM sites--those able to collect data during at least one calendar quarter in 2000. In general, the ADAM sites are very successful in convincing arrestees to participate. That was true in 2000, when at least 81 percent of adult male arrestees in half the sites agreed to be interviewed (Appendix Table 1-2). The refusal rate ranged from a low of 6 percent (Fort Lauderdale) to a high of 40 percent (Charlotte-Metro area). The vast majority of arrestees interviewed also agreed to provide a urine specimen for analysis. In half the sites, 89 percent or more agreed, with a low of 75 percent (Albany) to a high of 98 percent (Oklahoma City). (See Appendix Table 1-2.) In half the sites, 600 or more interviews were "complete" (that is, an interview was conducted and a urine sample obtained), with the range from 109 (Charlotte-Metro area) to 1,534 (Phoenix). A number of factors contributed to the variation in sample size (See "Why Sample Sizes Vary from Site to Site--and the Implications"), and when numbers were very small, they were not used in some analyses presented here. The number of adult male arrestees selected for inclusion in the sample averaged close to 300 per calendar quarter for each site. On the whole, these samples (the unweighted data) were more than adequate to allow data analysis and a reasonable interpretation of the results. With the adoption of probability-based sampling, the numbers can be converted by weighting to represent all arrestees in a given county/site--many more than in the original sample. The 2000 sample, when weighted, represents a large number of arrestees, from 921 in Laredo, Texas, to 18,037 in New York City. In more than half the sites the weighted sample size is more than 4,000. (See Table 1-1.) Refining the catchment area-where data are collected ADAM sites are typically named for the largest city in an area (the "primary city"). However, in most sites the catchment area has been redefined by ADAM to encompass a substantially larger geographic area than the urban center. The standard catchment area--the geographic region from which samples are drawn--is now the county in all the sites. The organization of booking facilities (jails), where arrestees are interviewed for the ADAM program, varies considerably by county. Some have a single, large facility where arrestees are brought by both city and county law enforcement agencies. Others have numerous smaller jails throughout the county. Generally, however, the jurisdictional reach of law enforcement agencies does not extend beyond county lines. Defining the sites by the county where a major metropolitan center is located (but does not necessarily encompass) means the primary unit of analysis for ADAM coincides with the standard government jurisdiction in which law enforcement's jurisdiction is generally defined. There are now 38 sites in 26 States and the District of Columbia. How the samples are now selected The sampling "frame" for ADAM data collection is now the total number of adult males arrested in a county in a two-week period, regardless of charge. The probability-based sampling has two stages: drawing samples of booking facilities and, within the facilities, drawing samples of arrestees. To allocate ADAM resources efficiently, a sampling simulation exercise is initially used to choose the optimal sampling design, select the booking facilities to be sampled, and distribute interviewer resources in each site. The overall goal of the design is to minimize the standard error of estimates for each site while recognizing the real-world constraints within which the program operates. The precision of estimates varies somewhat from site to site; it may be lower in some site where more than one facility is included. The specific goal is to generate estimates of drug use and related behavior that have no more than a .05 standard error overall for all sites. Selecting the booking facilities. In the first stage, a sample of booking facilities is drawn at each site from all facilities where people are arrested. The method of selection varies by site, depending on the number of facilities in the county and the number of arrestees booked into each. For sites that have only one booking facility, all cases are drawn from it. Sites with a small number of facilities (2 to 5) are stratified by size, and cases are sampled proportionate to the size of the facility. For sites having many facilities, the facilities are clustered, principally by size, and those in each cluster are sampled proportionate to size. In a few counties, a more complex sampling model that recognizes movement of arrestees within the county is required.[9] Selecting the arrestees. Once the facilities are selected, the second step is to draw a sample of arrestees from each. The sampling method in every facility is the same. An attempt is made to select cases systematically. Some arrestees are selected during the time of day when the volume of arrestees ("arrestee flow) is highest. In order to include a sample of arrestees booked when interviewers are not on site ("arrestee stock"), others are randomly selected during the rest of each 24-hour period. Arrestees who cannot be interviewed because they were released early are represented through statistical imputation. Sites are given a target number of interviews to complete each calendar quarter. It is based on an assumption of the number of interviews completed by one interviewer who works a regular shift each day of the week for a 1- or 2-week period. The probability of selection and the assignment of case weights are calculated by examining data on all arrestees booked at each facility in the two-week arrest/interview period. The new interview instrument The interview is a key component of the ADAM program--the source of information that cannot be obtained from official records or urinalyses. The interview process itself remains the same as in the past. Interviews are conducted among arrestees who volunteer to participate, and the process conforms to stringent Federal confidentiality regulations. Privacy is ensured because these regulations prohibit linking the interview to the arrestee's name and using the information for or against the arrestee during booking or adjudication. No record is kept of arrestees' names or other personal identifiers. Only a common ID number is assigned to the interview form and the urine specimen container so that these data can be linked. The interview-process and administration As in the past, interviews are conducted four times a year among male and female adult arrestees and juvenile detainees who have been in a booking facility less than 48 hours. They take place typically during a 4- to 8-hour period every day for one to two weeks. At each site, data collection proceeds on a staggered schedule, with collection periods for any single population (males, females, or juveniles) generally lasting one to two consecutive weeks. In most sites, more than 80 percent of the people asked to be interviewed agree. At each site, data collection is managed by a local team that includes a site director and site coordinator.[10] A pool of interviewers administers the interviews and collects the urine specimens. All interviewers must successfully complete a 3-day training course. At all sites, local data collection staff are trained in interview techniques and in administering the ADAM interview instrument. The same, standardized training materials are used at all sites. Training is conducted just before data collection so that new skills can be applied immediately to field conditions and so that interviewers can be observed by the trainers. All interviewers also must take enhancement training every quarter. The new design From 1987, the year the DUF program was established, through 1999, a relatively limited amount of information could be obtained during the interviews. It included the types of drugs arrestees used, arrestees' perceived dependence on drugs, and arrestees' perceived need for alcohol or drug treatment or both. Because the offense was known, the relationship between type of offense and drug use could be analyzed. Demographic and related information were also obtained during the interview. As part of the ADAM redesign, the interview instrument (questionnaire) has been enhanced significantly and a great deal more information is collected. The newly designed instrument, which takes about 10 minutes longer than previously (approximately 25 minutes) to administer, preserves the key measures of drug use and thus ensures comparability of data from year to year. The new features extend the usefulness of the information obtained: o Greater focus on the NIDA-5 drugs and patterns of use in the year before the arrestees were interviewed. o A screen for identifying arrestees' risk for drug dependence and clinically defined drug "abuse." o Questions about arrestees' participation in inpatient and outpatient drug and alcohol treatment and mental health treatment. o Questions about arrest history. o Questions about drug acquisition and recent use patterns. The latter feature offers insights into the dynamics of not only drug markets but also drug use and drug sharing. The new instrument is structured to permit crosswalks to other national datasets on drug use, such as the National Household Survey on Drug Abuse (NHSDA), the Treatment Episode Data Set (TEDS), the System to Retrieve Information from Drug Evidence (STRIDE), and the Uniform Crime Reports (UCR). All data are available for use by anyone who has a bona fide research project. (See "Availability of 'Raw' ADAM Data.") Notes 1. The ten drugs for which arrestees are tested in the ADAM program are cocaine, opiates, marijuana, methamphetamine, phencyclidine (PCP), methadone, benzodiazepines, methaqualone, propoxyphene, and barbiturates. The first five are the "NIDA-5," established as a standard panel of commonly used illegal drugs by the National Institute on Drug Abuse. 2. An adult is defined here as anyone brought to an adult lockup facility. 3. Urinalysis can detect drugs in the amphetamine group, but only a confirmatory test indicates whether the drug is methamphetamine. The confirmation is also necessary because several cold and diet medications contain amphetamines, which would produce false positives. 4. Unless indicated otherwise, all averages are expressed as medians. 5. U.S. Department of Health and Human Services, Office of Applied Studies, SAMSHA, National Household Survey on Drug Abuse--Main Findings, Washington, DC: 1998. 6. Throughout this report, "past month" and "past 30 days" are used interchangeably to refer to the 30 days before the arrestees were interviewed. 7. See Chapter 7 for an in-depth discussion of the ADAM redesign. 8. A detailed discussion of the method used to collect ADAM data is in Methodology Guide for ADAM, by D. Hunt and W. Rhodes. Prepared by Abt Associates Inc. in May 2001, it can be downloaded from the ADAM Web page (http://www.adam-nij.net) on the NIJ Web site (http://www.ojp.usdoj.gov/nij). 9. For more details, see Methodology Guide for ADAM. 10. Accountability from all data collection sites is ensured by the contractor that manages ADAM for NIJ. The contractor provides centralized oversight for such matters as fiscal management, rigorously standardized data collection procedures, and minimum requirements for interviewers. ---------------------------- Why Sample Sizes Vary from Site to Site--and the Implications In general, this report presents findings from all the ADAM sites. Of the 38 sites, findings are reported from all those (35 in number) where data were collected in at least one calendar quarter of 2000. Although the new procedure ensures representativeness of the sample, its adoption introduced complexities that affect comparability of findings from site to site. The findings should be read with an understanding that some data are missing and that in some cases changes were made to increase the representativeness of what data were available. Sampling difficulties Although 24 of the 35 sites were able to collect data in all four quarters, others were not. Six sites collected data in three quarters, 3 sites collected data in two quarters, and 2 sites collected data in only one quarter. (See Appendix Table 1-2.) In some sites, not enough data from arrestee case flow were obtained to permit weighting and thus these sites did not report data in the quarters when this information was missing. Some sites collected information from different populations from quarter to quarter. Findings reported here have not been adjusted for the missing quarters of data. A site-by-site breakdown reveals the difficulties: o Minneapolis and Philadelphia: Because they began data collection in the second quarter of the year, they reported data for only three quarters. o Los Angeles: After several years of collecting data at the Los Angeles Police Department's main facility, this site lost access in 2000. The site staff spent the year re-establishing authorization. Therefore, this report does not contain information about Los Angeles. o Albuquerque: Staffing problems in the jail prevented this site from collecting data in the fourth quarter. o Dallas: Data are presented for only three quarters, because the site team went on hiatus status to resolve sampling difficulties. o Houston and Fort Lauderdale: In these sites, staffing changes on the site team reduced to two the number of quarters when data were collected. o Miami: Here, staffing changes reduced to three the number of quarters in which data were collected. o Albany and Charlotte-Metro area: These two sites became part of the ADAM program as "affiliates" and did not collect data in all four quarters. Albany began collection in the second quarter and Charlotte-Metro in the fourth quarter. A few other sites encountered major obstacles to obtaining the census data needed to weight their samples, which in turn limited the number of quarters weighted data were available: o Chicago and Detroit: Data collection took place at these sites for more than one quarter, but both sites could provide adequate census data for only one quarter. o Atlanta: At this site it was impossible to obtain census data for all facilities in the sample. The findings are from Fulton County only, although data were collected from both Fulton and DeKalb counties. Making the data more representative As a result of these difficulties, changes were made to increase the representativeness of the data. As the examples of Houston, Dallas, and New York illustrate, in some cases the changes were dramatic. o Houston: In the first quarter, data were collected at the jails operated by the Houston Police Department and in the second quarter at a jail operated by the Harris County Sheriff's Department. This meant the first-quarter data reflect people arrested within the Houston city limits, while the second-quarter data reflect people arrested throughout Harris County. o Dallas: Collection had taken place in the main county jail, expanding to other booking facilities only in the fourth quarter (after a hiatus in the third quarter). As a result, fourth-quarter data are more representative of all arrestees in Dallas County than are first- and second-quarter data. o New York: Data collection, which had taken place in all five boroughs in the first quarter, was reduced to one borough--Manhattan--for subsequent quarters because of difficulties in sampling and obtaining census data from the other four. In some sites where there were several jails (Atlanta, Birmingham, Cleveland, Dallas, Des Moines, Detroit, Phoenix, San Antonio, and Seattle), the sampling plans used a stratified cluster model (explained in the Methodology Guide for ADAM. See note 8.) This required obtaining case flow data for all arrestees in the county. However, the data from these sites were weighted to the facilities in the site sampling plans--not to the county as a whole. Weights will be refined annually to reflect the countywide arrestee population; that is, the statistical inflation factor will be applied once all data are obtained. ---------------------------- Availability of "Raw" ADAM Data The ADAM data are both a research product and a resource to be used in future research. The National Institute of Justice recognizes the need to preserve and make available these and other machine-coded data collected with public funds. All archived ADAM data files are stored with the Inter-University Consortium for Political and Social Research (ICPSR), at the University of Michigan. Researchers who would like to obtain the raw data files may contact the ICPSR (by phone at 800-999-0960 or 734-998-9825 or on the Web at http://www.icpsr.umich.edu/NACJD/. NIJ's policy on use of ADAM data is on ADAM Web page (http://www.adam-nij.net), which can be accessed via the Web site of the National Institute of Justice (http://www.ojp.usdoj.gov/nij). In general, ADAM data for a particular year are available for public use after they have been presented in the ADAM annual report for that year. ---------------------------- ADAM drug testing--the procedure, the drugs Drug testing by urinalysis is a unique and important component of the ADAM program. ADAM uses an immunoassay (EMIT (Enzyme Multiplied Immunoassay Testing), to screen for the presence of drugs in urine. EMIT tests have been shown to be one of the most consistently accurate drug testing methods, with greater than 95 percent accuracy and specificity for most drugs. The procedure At the conclusion of the ADAM interview, arrestees are asked to provide a urine sample. Over the years of the program, approximately 80 percent agree to be interviewed, and of those more than 80 percent also agree to give a sample. Arrestees who have complete interviews (that is, they have been interviewed and have also given a urine sample) are given an incentive (for example, candy bars, gift certificates, or a soft drink). The urine specimens are removed daily from the ADAM site facilities. A positive result from the EMIT assay (or "screen") indicates that the drug for which the test is performed is present in the urine sample at a level above or equal to a specified cutoff point. A negative result means either there is no drug in the urine sample or the level is below the cutoff point. Because ADAM tracks the epidemiology of drug use over time, it is not necessary or cost-effective to take other steps to confirm the presence of drugs. A confirmatory test is performed only when it is necessary to detect a particular subclass of a drug. For instance, all amphetamine positives are confirmed by gas chromatography/mass spectrometry (GC/MS) to determine whether methamphetamine was used. Specimens from all the sites are screened at a central laboratory. The drugs detected by ADAM ADAM detects as many as 10 drugs, but the focus of the program is the "NIDA-5," so called because the National Institute on Drug Abuse has identified them as a standard panel of commonly used illegal drugs. They are cocaine, marijuana, methamphetamine, opiates, and phencyclidine (PCP). The other five are methadone, benzodiazepines, methaqualone, propoxyphene, and barbiturates. Amphetamines A positive EMIT screen result indicates the presence of one or more drugs in the amphetamine group. Drugs that produce an amphetamine-positive screen include: o d - Amphetamine o d - Methamphetamine o Methylenedioxyamphetamine (MDA) o Methylenedioxymethamphetamine (MDMA). When a test conducted to detect methamphetamine is positive, that means amphetamines are in the urine. In this country, most amphetamine use represents legal or illegal ingestion of manufactured products containing the substance. Several over-the-counter cold and diet medications, as well as drugs used to treat ADD, can trigger a positive EMIT result. By contrast, most methamphetamine use represents consumption of an illegal substance. To determine whether the substance detected is in fact methamphetamine, screens that indicate the presence of amphetamines are subjected to a confirmatory, GC/MS test. The percentage of a dose of amphetamine excreted from the body unchanged into a metabolite varies with the pH of the urine, with the range 2 percent (alkaline pH) to 68 percent (acidic pH). Typically, 20 to 30 percent of the substance is excreted as unchanged amphetamine and 25 percent as benzoic acid and a simple compound (hippuric acid). Methamphetamine is excreted primarily unchanged, with a small fraction as amphetamine (44 percent and 6 percent, respectively). Barbiturates A barbiturate screen detects drugs in the barbiturate group. A positive screen indicates the presence of any metabolites of the group. The EMIT screen process is most efficient at detecting secobarbital in the urine. However, depending on the concentration of drug, the screen will also detect other commonly encountered barbiturates, including butalbital, pentobarbital, alphenal, amobarbital, aprobarbital, barbital, cyclopentobarbital, 5-ethyl-5-(4-hydroxyphenyl) barbituric acid, butabarbital, phenobarbital, talbutal, and thiopental. Benzodiazepines Most benzodiazepines are metabolized extensively in the liver and excreted through the urine as metabolites. The EMIT assay is best at detecting oxazepam, a common metabolite of benzodiazepines. However, the assay can be positive for many other benzodiazepines and/or metabolites, such as the compounds alprazolam, bromazepam, chlordiazepoxide, clobazam, clonazepam, clorazepate, clotiazepam, demoxepam, N-desalkylflurazepam, N-desmethyldiazepam, diazepam, flunitrazepam (Rohypnol), flurazepam, halazepam (Halcion), a-hydroxyalprazolam, 1-N-hydroxyethylflurazepam, a-hydroxytriazolam, ketazolam, lorazepam, medazepam, midazolam, nitrazepam, norchlordiazepoxide, prazepam, temazepam, tetrazepam, and triazolam. Cocaine Cocaine is metabolized extensively by liver and plasma esterases,* and only 1 percent of the dose is excreted in the urine unchanged. The primary metabolite of cocaine, benzoylecgonine, is easily identified in a urine specimen. Therefore, the EMIT assay was specifically designed to detect benzoylecgonine. Marijuana Delta-9-tetrahydrocannabinol (THC) is the primary psychoactive ingredient in marijuana. THC is one of approximately 30 compounds known as cannabinoids. Almost no THC is excreted in the urine unchanged into a metabolite. The primary metabolite of THC is 11-nor-D9-THC-9-carboxylic acid. Other major metabolites detected by EMIT assay, and which indicate marijuana use, include: o 11-nor-D9-THC-9-carboxylic acid o 8-b-11-hydroxy-D9-THC o 8-b-hydroxy- D9-THC o 11-hydroxy- D8-THC o 11-hydroxy-D9-THC. Methadone The EMIT assay is specific to methadone. Unchanged methadone is detectable in the urine. Methaqualone Methaqualone is metabolized extensively. Less than 1 percent of the dose is excreted unchanged in the urine, while 25 percent is excreted as hydroxylated metabolites. The assay detects the following compounds: o Methaqualone o Macloqualone o 3'-hydroxy-methaqualone o 4'-hydroxy-methaqualone o 2'-hydroxymethyl-methaqualone. Opiates Opiates are a broad class of drugs that include heroin, morphine, codeine, and semisynthetic derivatives of morphine. Heroin is rapidly broken down in the body, first to 6-monoacetylmorphine, which is metabolized to morphine. Both heroin and 6-monoacetylmorphine disappear rapidly from the blood. Codeine is also metabolized to morphine. Because heroin and codeine break down to morphine, and the unique metabolite of heroin (6-monoacetylmorphine) disappears rapidly from the body, the EMIT opiate assay is designed to detect morphine and its metabolites. A positive screen on the EMIT assay indicates only that the substance might be heroin; use of other opiate drugs cannot be ruled out with the screen alone. Someone who has used morphine or codeine legally (morphine after surgery, for example, and codeine in a prescription drug, for example) might reasonably be expected to screen positive for opiates. The EMIT assay can detect the following common compounds in the that belong to the class of opiates: o Morphine o Morphine-3-glucuronide o Codeine o Dihydrocodeine o Hydrocodone o Hydromorphone o Levallorphan. Morphine is metabolized extensively, with only 2 to 12 percent excreted unchanged in the urine. Large amounts (60 to 80 percent) of the conjugated metabolites (glucuronides) are excreted. In terms of quantity excreted, the most important metabolite of opiates is morphine-3-glucuronide-67 to 70 percent of the dose is excreted in the urine. The pattern of urinary excretion of morphine from heroin is similar to that of pharmaceutical morphine: 7 percent is excreted unchanged and 50 to 60 percent as conjugated morphine (glucuronides). Codeine is metabolized extensively, primarily to conjugated 6-codeine-glucuronide, while 10 to 15 percent of the dose forms morphine and norcodeine. Phencyclidine (PCP) The EMIT assay for PCP is designed to detect the following metabolites: o Phencyclidine o N, N-diethyl-1-phenylcyclohexylamine (PCDE) o 1-(4-hydroxypiperidino) phenylcyclohexane o 1-(1-phenylcyclohexyl) morpholine (PCM) o 1-(1-phenylcyclohexyl) pyrrolidine (PCPy) o 4-phenyl-4-piperidinocyclohexanol o 1-(1-(2-thienyl)-cyclohexyl) morpholine (TCM) o 1-(1-(2-thienyl)-cyclohexyl) piperidine (TCP) o 1-(1-(2-thienyl)-cyclohexyl) pyrrolidine (TCPy). The body produces all these metabolites by consuming PCP. Only about 10 percent of a PCP dose is excreted unchanged in the urine. About 40 percent of the substances in a urine specimen containing PCP have not been identified by science. Propoxyphene Propoxyphene is classified as a narcotic analgesic, used for pain relief, that includes the trade name Darvon. The EMIT process detects the following compounds that indicate propoxyphene use: o Propoxyphene o Norpropoxyphene. * An esterase is an enzyme that speeds the splitting up of an ester--a molecule consisting of an acid and an alcohol. ---------------------------- Immunoassays and what they detect An immunoassay is a test that uses antibodies to detect the presence of drugs and other substances in urine. Each immunoassay is designed to detect one particular drug or drug class. In some cases, the EMIT assay used by ADAM detects the drug itself, while in other cases it detects the metabolites of the drug. Metabolites are compounds produced by the breakdown of a drug in the body. The drug-metabolite distinction is important. There is no specific EMIT heroin assay, for example. Instead, EMIT detects metabolites common to all opiates, including heroin and codeine. When a screen detects a class of drugs, such as opiates, a confirmation test can be performed to identify the specific drug. ---------------------------- II. Drug Dependence and Treatment by Christine R. Crossland and Henry H. Brownstein* DUF and ADAM have revealed that people who come to the attention of the criminal justice system by being arrested are more often than not users of drugs and/or alcohol.[1] What is not known is the extent to which they have become dependent on these substances.[2] Nor is it known to what extent they need treatment or even have access to treatment.[3] Nonetheless, dependence and access to treatment, particularly for this at-risk population, are serious social and public health problems[4] about which data are often limited. Many communities have historically lacked the data needed to identify arrestees' treatment needs, because such users are typically undercounted in drug-use surveys (for example, the National Household Survey on Drug Abuse). Two additions to the ADAM survey instrument were designed to promote understanding of arrestee dependence and treatment needs as a means to address the resultant public health problems. First, the instrument now includes a screening tool to assess risk for drug and alcohol dependence--a measure of need for treatment. Second, questions about arrestees' treatment history have been added in an attempt to determine whether arrestees have ever received drug or alcohol treatment and whether they received such treatment recently-- specifically, in the year before they were arrested.[5] With the redesigned ADAM program, many communities now have access to data on treatment and can use it to develop evidence-based policies that can help local and national policymakers acquire or target treatment resources. Measuring drug dependence/treatment need In response to the debate among researchers and policymakers about the distinction between physical and psychological dependence, sociologist Erich Goode has suggested that such distinction is "largely irrelevant."[6] He contends that chronic users of drugs that cause psychological dependence behave in much the same way as individuals who are addicted to drugs that cause physiological dependence. For example, while cocaine dependence is not the same as heroin addiction,[7] the profound psychological need felt by cocaine users produces similar behavioral outcomes. Because the behavioral effects of physiological and psychological addiction are similar, the emphasis in the ADAM screener for dependence is on behavior rather than on classic physiological markers, such as tolerance or withdrawal. (For details about the development of the screener and the screener itself, see "Screening Arrestees for Drug and Alcohol Dependence/Need for Treatment.") Arrestees at risk for dependence on drugs Overall, among all adult male arrestees in the ADAM sample, between 27 percent (Houston and San Antonio) and 47 percent (Chicago) were found to be at risk for dependence on drugs. (See Appendix Table 2-1.) While in no site were more than half the arrestees found to be drug dependent, neither was there any site where less than one-fourth were drug dependent. Risk for dependence by type of drug Among users of marijuana, crack, powder cocaine, heroin, and methamphetamine, more than half were found to have been at risk for dependence in the past year. (See Appendix Table 2-2.) The proportions at risk varied by drug. In general, heroin users were more likely than users of other drugs to be at risk for dependence. The figure was 88 percent or more in half the sites, with a range of 50 percent (Charlotte-Metro) to 100 percent (Birmingham, Des Moines, and Indianapolis). At 56 percent in half the sites, the proportion of marijuana users at risk for dependence was much lower than for all other drugs. The range was 45 percent of drug-using arrestees (Denver) to 69 percent (Des Moines). For users of the other drugs, the proportions at risk for dependence lay between the rates for heroin and marijuana. Thus, for crack cocaine, 80 percent of drug-using arrestees in half the sites were at risk; the figure for powder cocaine was 74 percent, for methamphetamine, 76 percent, and for other drugs, 74 percent. (Exhibit 2-1 shows these relative averages.[8]) Demographics and sociodemographics of those at risk Among drug-using arrestees at risk for dependence, there was some variation by site in age, race, ethnicity, employment status, level of education, marital status, and whether or not the arrestee had health insurance. (See Appendix Table 2-3.) For example, the proportion of arrestees who scored at risk for drug dependence and were under 21 ranged from less than 10 percent (Denver and Las Vegas) to more than 35 percent (San Antonio). Of arrestees at risk for dependence, in Atlanta, Birmingham, Chicago, Detroit, and New Orleans, more than 75 percent were black; in Albuquerque, Honolulu, Laredo, Phoenix, Salt Lake City, San Antonio, and Spokane, fewer than 13 percent were black. This breakdown may, of course, reflect the racial and ethnic composition of all adult male arrestees and all people living in the particular county.[9] Thus, in the same way, in a number of southwestern sites, the proportion of arrestees who were both drug-dependent and Hispanic was relatively high (for example, 64 percent in Albuquerque; 93 percent in Laredo; 71 percent in San Antonio; and 42 percent in Tucson). This reflects the high percentage of adult male arrestees in these sites who said they were of Hispanic heritage (Albuquerque, 60 percent; Laredo, 96 percent; San Antonio, 68 percent; and Tucson, 42 percent). Prevalence of treatment among drug users The adult male arrestees who said they used drugs were asked whether they had participated in inpatient and outpatient treatment for drugs or alcohol, both in the past year and in their lifetime. Fewer than one in ten said they had received inpatient drug or alcohol treatment (for example, in detox, rehab, a therapeutic community, or a hospital) in the past year (9 percent or less, in half the sites). The range was 4 percent (Birmingham) to 17 percent (Albany/New York Capital area). (See Appendix Table 2-1.) The proportion who had ever been in inpatient treatment was higher: In half the sites, at least 29 percent of drug-using arrestees said they had ever been treated on an inpatient basis. (See Exhibit 2-2 for a visual illustration of the comparative percentages.) This may reflect the large numbers who have participated in (inpatient) detox programs. Mental health treatment was much less common, with 10 percent or less in half the sites saying they ever received such treatment. The proportions ranged from 2 percent (Charlotte-Metro) to 20 percent (Spokane). Treatment by type of drug The proportion of drug users who ever received treatment varied by type of drug used. For inpatient treatment, marijuana was the drug for which the proportion of arrestees was lowest (28 percent or less in half the sites). (See Exhibit 2-3.) Among drug users who ever used marijuana,[10] the proportion who ever participated in inpatient treatment ranged from 16 percent (New Orleans) to 46 percent (Albany). (See Appendix Table 2-4.) The proportions who ever received outpatient treatment for this drug were somewhat lower, with the range 11 percent (New Orleans) to 42 percent (Albany). Arrestees who had used crack at some point in their lives were typically more likely than marijuana users to have ever received either type of treatment. In half the sites, 48 percent had received inpatient treatment and 31 percent outpatient treatment. For inpatient treatment, the range was 28 percent (New Orleans) to 73 percent (Albany); for outpatient treatment, it was 17 percent (New Orleans) to 66 percent (Albany). The proportions of arrestees who ever used the other drugs--heroin, powder cocaine, or methamphetamine--and said they had ever been in treatment were relatively high, with figures varying somewhat by site. In all sites except four (Atlanta, Chicago, Dallas, and New Orleans), half or more of all arrestees who ever used heroin also said they had received inpatient drug treatment at some point in their lives. Overall, at 61 percent, the proportion of heroin-using arrestees who had ever received inpatient treatment was higher than for those who used any of the other drugs. (See Exhibit 2-4.) The same was true of heroin users who received outpatient treatment, although the differences among the drugs were less dramatic. In nine sites (Albany, Anchorage, Des Moines, Detroit, Minneapolis, New York, Portland, San Diego, and Seattle), half or more of the arrestees who ever used powder cocaine said they had received inpatient treatment at some time in their lives. Because the proportions who received treatment were in some instances relatively high, they suggest overall that many adult male arrestees who used drugs have at one time or another availed themselves of treatment but remain drug users. Demographics and socio-demographics Among adult male arrestees who had participated in drug or alcohol treatment in the year before their arrest, there were few demographic differences by site. The average (median) age of those who had participated in inpatient treatment in the 12 months before their arrest was 34 years; among those participating in outpatient treatment it was 32. Among those who had participated in inpatient treatment, the proportion who did not have health insurance was high: In half the sites, at least 66 percent said they currently lacked health insurance. The range was 29 percent (Birmingham) to 85 percent (Indianapolis). The proportions who participated in outpatient treatment but lacked health insurance were also high: 64 percent or more in half the sites, with the range 31 percent (Omaha) to 91 percent (Charlotte-Metro). (See Appendix Table 2-5.) The situation of Hispanic arrestees is particularly notable. High percentages were at risk for drug dependence, and among them the proportions who had the health insurance coverage needed to address the problem were relatively low. (See Exhibit 2-5.) As with other demographic characteristics, race appears to make a difference in likelihood of being treated. Black arrestees were far more likely than whites and others to have said they had received treatment for drug or alcohol use in the year before they were arrested. (See Exhibit 2-6.) Drug-dependent arrestees--treatment needs The redesigned ADAM revealed notable proportions of adult male arrestees at risk for drug dependence and thus in particular need of treatment. Many had received treatment at some point in their lives, but the vast majority had not been treated recently (in the past year). (See Appendix Table 2-6.) In the year before their arrest, as few as 6 percent of drug-dependent arrestees (in Atlanta and Chicago) and rarely more than 20 percent (in Albany, Charlotte-Metro, Laredo, Minneapolis, and Portland) were treated on an inpatient basis, and in more than half the sites less than 10 percent received outpatient treatment. There may be a number of reasons that, despite an evident need, arrestees do not receive treatment. One barrier may be lack of health insurance. In half of the sites, at least two-thirds of these at-risk arrestees lacked any type of health insurance. (See Appendix Table 2-3.) The proportion lacking health insurance reached 50 percent in every site. For the most part, the proportion of drug-dependent arrestees who lacked insurance exceeded the proportions of drug-using arrestees who lacked insurance. (See Exhibit 2-7.) Toward more in-depth investigation ADAM has shown that not only is there considerable drug use among adult male arrestees, but there is also considerable risk for drug dependence--an index of need for treatment. As the ADAM program continues to expand and evolve, additional questions about drug dependence and treatment needs can be investigated. In addition to the new questions about need for treatment and types of services received (whether inpatient or outpatient), it may be possible to explore treatment settings, modalities, and types of interventions. In the near future, by adding to the interview instrument a more substantive module addressing treatment, the ADAM program will be able to offer practitioners, researchers, and policymakers more detailed information about arrestees' need for services. And it will be possible not only to identify treatment needs in particular areas at a particular time and to compare sites but, as data are collected from year to year, it will also be possible to track changes in specific sites. * Christine R. Crossland is a Program and Policy Analyst with the Drugs and Crime Research Division of the National Institute of Justice (NIJ); Henry H. Brownstein, Ph.D., is Director of the Drugs and Crime Research Division, NIJ, and Executive Director of the ADAM Program. Notes 1. See, for example, Arrestee Drug Abuse Monitoring Program, 1999 Annual Report on Drug Use Among Adult and Juvenile Arrestees, Research Report, Washington, DC: U.S. Department of Justice, National Institute of Justice, June 2000, NCJ 181426. 2. See Chen, H.T., et al., "Problems and Solutions for Estimating the Prevalence of Drug Abuse Among Arrestees," Journal of Drug Issues 27 (1997): 689-701; and Goode, E., Drugs in American Society, New York: McGraw Hill, 1993. 3. See Harrison, L., "The Revolving Prison Door for Drug-Involved Offenders: Challenges and Opportunities," Crime and Delinquency 47 (July 2001): 462- 484; and Hser, Y.I., D. Longshore, and M.D. Anglin, "Prevalence of Drug Use Among Criminal Offender Populations: Implications for Control, Treatment, and Policy," in Drugs and Crime-Evaluating Public Policy Initiatives, ed. D.L. Layton and C.D. Uchida, Thousand Oaks, CA: Sage, 1994:18-41. 4. Horgan, C., K.C. Skwara, and G.S., Substance Abuse-The Nation's Number One Health Problem, Princeton, NJ: The Robert Wood Johnson Foundation, 2001; and Office of National Drug Control Policy, The National Drug Control Strategy: 2001 Annual Report, Washington, DC: Executive Office of the President, 2001. 5. In this report, 12 months and one year are used interchangeably. 6. Goode, Drugs in American Society: 33. 7. U.S. Sentencing Commission, Cocaine and Federal Sentencing Policy, Washington, DC: U.S. Sentencing Commission, 1995: 22-28. 8. Unless indicated otherwise, averages are expressed as medians. 9. For most demographic characteristics, the proportions of adult male arrestees who scored as at risk for dependence mirrored the overall rates for the entire ADAM sample. When differences occurred, they were among sites rather than between arrestees who were drug-dependent and those who were not. 10. The arrestees were asked about each drug separately; thus a single arrestee could be included in each group of users of a specific drug. For example, an arrestee who used marijuana might also be among the cocaine and/or methamphetamine users. This overlap should be kept in mind in interpreting the findings. ---------------------------- Screening Arrestees for Drug and Alcohol Dependence/Need for Treatment As part of the redesign, ADAM added to the questionnaire a "screener" that generates information about risk for dependence on drugs and alcohol and consequent need for treatment. The screener was developed from a subset of questions derived from the Substance Use Disorder Diagnostic Schedule (SUDDS-IV), a clinical assessment based in turn on criteria for dependence in the American Psychiatric Association's DSM-IV.[a] This series of questions in the ADAM interview makes it possible to estimate the number of arrestees who are likely to be at risk for alcohol and/or drug dependence. The information from the new series of questions can also aid in responding to the problem. Examining the use of specific drugs can help promote the development of strategies and planning policies to address new or emerging problems. For instance, if the number of heroin users increases, if that increase was recent, and if the proportions found at risk for dependence have increased, this information can be used by providers to assess the need for resources (for example, whether more methadone treatment is needed). To measure substance abuse and risk for dependence, arrestees who said they used alcohol or drugs in the 12 months before their arrest are asked six questions. Pilot tests conducted in three cities[b] revealed these particular questions best predicted risk for dependence and abuse. o Have they spent more time drinking or using drugs than they intended? o Had they neglected their usual responsibilities because of drug or alcohol use? o Had they wanted to cut down on drinking or drug use? o Had anyone, during the past 12 months, objected to their use of drugs or alcohol? o How frequently had they found themselves thinking about using drugs or alcohol? o Had they had used drugs or alcohol to alleviate feelings such as sadness, anger, or boredom? Arrestees who answered yes to only one or none of the six questions were considered at no risk for either drug abuse or dependence. A combination of two affirmative responses indicated risk for abuse, unless the two responses were to the questions about using drugs and alleviating negative emotions. Risk for abuse was also indicated when an arrestee answered yes to three or more questions, as long as thinking about using drugs or alcohol or alleviating negative emotions was among the three. A combination of three or more affirmative responses indicated risk for dependence, provided that either thinking about using drugs or alcohol or alleviating negative emotions was one of the three. In addition, if both thinking about using either substance and alleviating negative emotions were the only two affirmative responses, the person was considered at risk for dependence. a. DSM-IV refers to the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders, compiled and published in 1994 by the American Psychiatric Association. It is used by psychiatrists for diagnoses and is widely used by others. See Hoffmann, N.G. and P.A. Harrison, SUDDS-IV: Substance Use Disorder Diagnostic Schedule-IV, St. Paul: New Standards, Inc., 1995; Hoffmann, N.G., et al., "UNCOPE: A Brief Substance Dependence Screen for Use with Arrestees," in Drug and Alcohol Dependence, forthcoming; and Hunt, D. and W. Rhodes, Methodology Guide for ADAM, Washington, DC: U.S. Department of Justice, National Institute of Justice, May 2001. The Guide can be downloaded from the ADAM Web page (http://www.adam-nij.net) on the NIJ Web site (http://www.ojp.usdoj.gov/nij). b. Hoffmann, et al., "UNCOPE." ---------------------------- III. Alcohol Use and Alcohol Dependence by Natalie Lu* As part of the redesigned ADAM program, arrestees are now asked about alcohol use. Since drug use is higher among arrestees than among the general population, it is no surprise that the same is true of alcohol use. About half of all Americans age 12 and older drink alcohol at least once a month and about 20 percent have five or more drinks on one occasion in a month.[1] By contrast, 61 percent or more of the arrestees, on average,[2] said they drank alcohol heavily in the past year, and 52 percent on average said they drank heavily in the past month.[3] Heavy alcohol use among adult male arrestees seems to be unrelated to most demographic indicators examined here. And large proportions of these arrestees who drink most heavily are at risk for dependence on alcohol and are more likely to have used drugs than those who are not heavy drinkers. Why measure heavy alcohol use Alcohol is the most widely used psychoactive drug in the United States.[4] It is legal and for most people does not cause health problems. Light or moderate alcohol use may even confer some health benefits, particularly for the cardiovascular system.[5] Some people, however, consume alcohol in quantities large enough to cause problems for themselves or others.[6] Chronic heavy drinking has been linked to brain damage, hypertension, stroke, certain cancers, and harm to the fetus during pregnancy;[7] it is a contributing factor in workplace and automobile accidents and increases the likelihood of homicide and suicide[8] and has been implicated in sexual assault and domestic violence.[9] Although alcohol is like illicit drugs in producing profound effects, it also differs in many respects.[10] Alcohol has more complicated effects on the brain. While most illicit drugs affect only a few brain neurotransmitters, alcohol affects many, and the outcomes differ from person to person. And unlike some illicit drugs, alcohol is toxic to most body organs. To enhance the understanding of alcohol use and alcohol-related behavior, ADAM asks arrestees[11] about alcohol use and their experiences with treatment[12] and also measures their risk for dependence on alcohol. Overall findings Alcohol is heavily used by arrestees. Various levels of "heavy" drinking are defined here, with the level depending on the number of days a month the arrestee had five or more drinks.[13] (Definitions are presented in Table 3-1.) Large percentages of arrestees drank heavily in the year and the month before their arrest. Past-year heavy drinking (defined as "binge drinking,") ranged from a low of 47 percent of arrestees (Philadelphia) to a high of 82 percent (Albuquerque). In half the sites, 61 percent or more said they engaged in binge drinking (that is, had five or more drinks on at least one occasion in a one-month period) the year before their arrest. Figures for past-month binge drinking ranged from a low of 35 percent (Philadelphia) to a high of 70 percent (Albuquerque). In half the sites, 52 percent or more engaged in binge drinking in the past month. (See Appendix Table 3-1.) Age and other demographic and sociodemographic characteristics Overall, there appear to be few differences between younger and older adult male arrestees in extent of binge drinking. Among the youngest (those under 21), at least 45 percent in half the sites said they had five or more drinks on one occasion at least once in the month before they were interviewed; among the oldest arrestees (over 35) the median was 53 percent--not that much greater. (See Appendix Table 3-2a). Within some age groups, however, there was considerable variation by site. Thus, among the youngest arrestees, the rates of binge drinking ranged from a low of 17 percent of arrestees (New Orleans) to a high of 66 percent (Albuquerque). Similarly, among arrestees ages 21 to 25, the range was 24 percent (New Orleans) to 75 percent (Albuquerque). In the overwhelming majority of sites (32 of the 35), more white arrestees than blacks said they had five or more drinks on one occasion at least once in the past month. Employment status, education level, and whether or not the arrestee has health insurance seem to play minor roles in explaining binge drinking. (See Appendix Table 3-2b.) The one factor other than race that made a difference was homelessness. (See Exhibit 3-1.) In 29 of the 35 sites, homeless arrestees were more likely to say they binged the month before they were arrested than those who were not homeless. In sites such as Fort Lauderdale, the difference was notable, with past month binge drinking among homeless arrestees approximately 92 percent, while for arrestees who were not homeless it was 51 percent. Levels of heavy alcohol use The proportion of adult male arrestees who were the heaviest drinkers (had five or more drinks on a single occasion on at least 13 days in the month before their arrest--or every other day of the month) ranged from 10 percent (Miami) to 24 percent (Tucson). (See Appendix Table 3-3.) In half the sites, 17 percent or more could be placed in this category of heaviest drinkers A relatively small proportion of arrestees (median 6 percent) were classified as heavier drinkers (had five or more drinks on a single occasion on 8 to 12 days in the month before the arrest), while the proportion classified as heavy drinkers (had five or more drinks on a single occasion on 1 to 7 days in the past month) was the largest (median 27 percent). There appears to be little middle ground in the drinking patterns of ADAM male arrestees who consume alcohol heavily. The proportions of arrestees who were heavy and heaviest drinkers were higher than the proportions who drank at the middle or heavier level. (See Exhibit 3-2.) Lowest and highest percentages for each category are represented by the "tails" of the box plot. Alcohol dependence The use of alcohol (or drugs) does not necessarily mean abuse or dependence. Level of alcohol consumption varies dramatically--from casual to frequent to very frequent, heavy use. For some moderate drinkers, even a small amount of alcohol can create problems, while for some people who drink heavily the social and/or health problems may not materialize right away. Because of these differences, clinicians are able to diagnose alcohol abuse and dependence only by determining whether they have resulted in health and/or relationship problems. This is done through an extensive series of questions based on criteria established by the American Psychiatric Association's DSM-IV.[14] The result is a clinical diagnosis of either alcohol abuse or alcohol dependence. Beginning in 2000, the ADAM interview instrument included questions that screen for drug and alcohol abuse and dependence. The screen consists of six questions from the Substance Use Disorder Diagnostic Schedule (SUDDS-IV), an instrument based on dependency criteria in the DSM-IV. The screen does not produce a clinical diagnosis, but rather an indication of risk for dependence.[15] (A more detailed discussion of this screen is in Chapter 2.) Risk for alcohol dependence is discussed here. In employment status, education level, and health insurance status, there were few differences in the proportions of adult male arrestees at risk for dependence on alcohol. (See Appendix Tables 3-4a and 3-4b.) There were differences by age. Among the youngest adult male arrestees, 23 percent on average were at risk for alcohol dependence; by contrast, among the oldest group the percentage was 35. The difference was even more notable in homelessness. Homeless arrestees were much more likely than those who were not homeless to report behavior that would classify them as at risk for alcohol dependence (46 percent, on average, compared to 30 percent). This mirrors the pattern for binge drinking by arrestees, noted above: Whether or not they were at risk for alcohol dependence, arrestees who were homeless were more likely than those who were not homeless to be binge drinkers. If alcohol dependence is not measured by level of use, is there any relation between level of use and dependence? An examination of the data reveals there is: Among arrestees who were the heaviest drinkers, on average more than four in five scored as at risk for alcohol dependence. (See Appendix Table 3-5.) The range among the sites was 67 percent (Omaha) to 91 percent (Charlotte), with 85 percent or more of the heaviest drinkers in half the sites at risk for dependence. The proportions at risk for dependence declined with the levels of drinking. Thus, among the heavier-drinking group, 72 percent on average were at risk, with the range 39 percent (Charlotte-Metro) to 89 percent (Cleveland). And among the heavy-drinking group (the lowest level), the average at risk for dependence was still lower, at 59 percent, with the range 39 percent (Omaha) to 72 percent (Spokane). Given the easy accessibility and low cost of alcohol, and the fact that drinking often precedes illicit drug use, alcohol is sometimes referred to as a "gateway drug" for young people.[16] That raises the question of whether there is a relationship between dependence on alcohol or drugs later in life and the age at which someone first starts drinking. Are people who become dependent on alcohol or drugs more likely to have started drinking at an early age? The ADAM data suggest they are. Compared to those who had their first drink after age 21, adult male arrestees who started drinking at 13 or younger were twice as likely to be classified as at risk for alcohol dependence. (See Appendix Table 3-6.) Similarly, if not more dramatically, compared to those who began drinking later in life, arrestees who had their first drink at 13 or younger were twice as likely to be at risk for drug dependence. To more definitively determine whether alcohol is a gateway drug would require an analysis beyond the scope of this report. The ADAM data are presented to suggest areas for further study. Is alcohol use related to use of illicit drugs? For some people, alcohol use is the primary substance abuse problem, while for others, it may be only one of several high-risk behaviors.[17] One of them may be drug use. This raises the question of whether for some people the two types of substance abuse are related. Perhaps not surprisingly, the heaviest drinkers were also likely to have used illicit drugs. Compared to arrestees who did not binge drink at all, those in the heaviest drinker category were more likely to say they used at least one NIDA-5 drug. In half the sites, 71 percent or more of the heaviest drinkers used at least one drug. (See Appendix Table 3-7.) (It should be kept in mind that arrestees could say they used more than one drug. Therefore, if an arrestee who was among the heaviest alcohol users also used marijuana, it is possible that he might also have used cocaine, heroin, methamphetamine, and/or PCP.) Overall, more than half the arrestees who were among the heaviest drinkers in the month before their arrest also reported marijuana use in the same period. And among the heaviest drinkers, the proportion who used crack cocaine was almost three times higher than among those who did not binge drink (28 percent compared to 10 percent). * Natalie Lu, Ph.D., is a Drug Testing Technology Specialist with the National Institute of Justice. Notes 1. Substance Abuse and Mental Health Service Administration, The 1999 National Household Survey on Drug Abuse, Rockville, MD: U.S. Department of Health and Human Services, 2000. 2. These percentages are medians. Unless otherwise indicated, averages are expressed as medians throughout this report. 3. "Month" and "30 days" are used interchangeably, as are "year" and "12 months." 4. Horgan, C., Substance Abuse--The Nation's Number One Health Problem, Princeton, NJ: Robert Wood Johnson Foundation, 2001. 5. Agarwal, D.P. and L.M. Srivastava, "Does Moderate Alcohol Intake Protect Against Coronary Heart Disease?" Indian Heart Journal 53 (March- April 2001): 224-30; Marques-Vidal, et al., "Relationships Between Alcoholic Beverages and Cardiovascular Risk Factor Levels in Middle-Aged Men: The PRIME Study," Atherosclerosis 157 (August 2001): 431-40; and Puddey, I.B., V. Rakic, S.B. Dimmitt, and L.J. Beilin, "Influence of Pattern of Drinking on Cardiovascular Disease and Cardiovascular Risk Factors: A Review," Addiction 94 (May 1999): 649-663. 6. Hoffmeister, H., et al., "The Relationship Between Alcohol Consumption, Health Indicators and Mortality in the German Population," International Journal of Epidemiology 28 (December 1999):1066-1072; and Muntwyler, et al., "Mortality and Light to Moderate Alcohol Consumption After Myocardial Infarction," Lancet 12, 352 (December 1998):1882-18825. 7. Iribarren, C., T. et al., "Cohort Study of Thyroid Cancer in a San Francisco Bay Area Population," International Journal of Cancer 93 (September 2001):745-750; Van Der Leeden, M., et al., "Infants Exposed to Alcohol Prenatally: Outcome at 3 and 7 Months of Age," Annals of Tropical Pediatrics 21 (June 2001):127-134; Hard, M.L., T.R. Einarson, and G. Koren, "The Role of Acetaldehyde in Pregnancy Outcome After Prenatal Alcohol Exposure," The Drug Monitor 23 (August 2001): 427-434; Ajani, U.A., et al., "Alcohol Consumption and Risk of Type 2 Diabetes Mellitus Among U.S. Male Physicians," Archives of Internal Medicine 160 (April 2000):1025-1030; and Berger, K., et al., "Light-to-Moderate Alcohol Consumption and Risk of Stroke Among U.S. Male Physicians," New England Journal of Medicine 341 (November 1999):1557-1564. 8. Martin, S.E., K. Bryant, and N. Fitzgerald, "Self-Reported Alcohol Use and Abuse by Arrestees in the 1998 Arrestee Drug Abuse Monitoring Program," Alcohol Research and Health 25 (2001): 72-79; Parker, R.N. and K. Auerhahn, "Alcohol, Drugs, and Violence," Annual Review of Sociology 24 (1998): 291-311; and Spunt, B.J., et al., "Alcohol and Homicide: Interviews with Prison Inmates," Journal of Drug Issues 24 (1994):143-163. 9. See Aldarondo, E., and G.K. Kantor, "Social Predictors of Wife Assault Cessation," in Out of Darkness: Contemporary Perspectives on Family Violence, ed. G. K. Kantor and J.L. Jaswiski, Thousand Oaks, CA: Sage, 1997; Kaufman Kantor, G., and J.L. Jasinski, "Dynamics and Risk Factors in Partner Violence," in Partner Violence: A Comprehensive Review of 20 Years of Research, ed. J.L. Jasinski and L.M. Williams, Thousand Oaks, CA: Sage, 1998; Leonard, K., and M. Senchak, "Prospective Prediction of Husband Marital Aggression within Newlywed Couples, Journal of Abnormal Psychology 105 (1996): 369-380; Pan, H.S., P.H. Neidig, and D.K. O'Leary, "Predicting Mild and Severe Husband-to-Wife Physical Aggression, Journal of Consulting and Clinical Psychology 62 (1994): 975-981; Woffordt, S., D.E. Mihalic, and S. Menard, "Continuities in Marital Violence," Journal of Family Violence (1994):195-225. and Ullman, S.E., G. Karabatsos, and M.P. Koss, "Alcohol and Sexual Assault in a National Sample of College Women, Journal of Interpersonal Violence 14, 6 (1999): 603-625. 10. See Horgan, C., Substance Abuse. 11. ADAM does not use urinalysis to confirm arrestees' self-reported alcohol use, because alcohol can be detected in the urine for only a short time. All information on alcohol use was obtained from the self-reports. The new ADAM interview instrument also incorporates many cross-link variables that make it feasible to compare ADAM data with other national survey datasets such as the National Household Survey on Drug Abuse (NHSDA) and the Treatment Episode Data Set (TEDS). 12. Treatment is discussed in Chapter 2. 13. In the preliminary report of the 2000 ADAM findings, the NHSDA definition of heavy drinking (five or more drinks on five or more occasions in a month) was also used. See Taylor, Bruce G., et al., ADAM Preliminary 2000 Findings on Drug Use and Drug Markets--Adult Male Arrestees, Research Report, Washington, DC: U.S. Department of Justice, National Institute of Justice, December 2001:16 (NCJ189101). 14. DSM-IV refers to the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders, compiled and published in 1994 by the American Psychiatric Association. It is used by psychiatrists for diagnoses and is widely used by others. 15. See also Hoffman, N.G., et al., "UNCOPE: A Brief Substance Dependence Screen for Use with Arrestees," Drug and Alcohol Dependence, forthcoming. 16. Horgan, C., Substance Abuse. 17. Paniagua Repetto, H., et al., "Tobacco, Alcohol and Illegal Drug Consumption among Adolescents: Relationship with Lifestyle and Environment," Anales Espanoles de Pediatria 55 (August 2001):121-128; and Carol, G., et al., "Alcohol and Drug Abuse: A Preliminary Investigation of Cocaine Craving Among Persons With and Without Schizophrenia," Psychiatric Services 52 (August 2001):1029-1031. ---------------------------- IV. Drug Markets by Bruce G. Taylor and Michael Costa* Law enforcement agencies often base their strategies for controlling drug markets on tactical or anecdotal information and the experience of their officers. That approach is useful but limited. Aside from the DEA's monitoring systems, which track only a small number of communities, there are few other information resources. The ADAM redesign makes it possible for the first time to obtain information about drug markets from a large number of buyers at the local level. This information, on a wide variety of topics related to drug markets, can help criminal justice and law enforcement policymakers and practitioners to design better strategies. (For discussion of the DEA drug market monitoring systems, see "Drug Market Monitoring by the DEA.") Much previous research on drug markets was carried out as single, stand-alone studies, and include a rich tradition of ethnographic studies,[1] but the ADAM redesign makes possible multiple-site studies and analysis of trends. ADAM offers the opportunity to examine larger samples of drug markets than are available in single-site studies; systematic analysis is possible because all the ADAM sites have a uniform data collection procedure. The opportunity to explore drug markets was the result of a cumulative process that began with the addition of questions about market participation to the interview instrument fielded in 1995 in six DUF (Drug Use Forecasting program) sites.[2] Areas of focus The ADAM redesign generates information about extent of participation in drug markets, method of acquisition (whether cash or noncash), place of purchase (on the street or indoors), neighborhood of purchase, and difficulties in locating and buying drugs. The analyses presented here focus on two areas: buyer behavior and transaction dynamics. The first analysis covers the activities of buyers in the environment of the drug market. The second analysis covers the specific drugs obtained, the quantities obtained, the frequency of transactions, and the amount of money exchanged. Previous research on drug markets suggests that while they all operate according to the same general market principles,[3] the dynamics are likely to be somewhat different for each drug.[4] This necessitates examining each one separately. In most of this chapter the emphasis is on crack cocaine, powder cocaine, and marijuana because, of the drugs analyzed by ADAM, these are the ones used by the largest proportion of arrestees at the ADAM sites.[5] Extent of drug market participation Adult male arrestees were asked whether they had obtained crack cocaine, powder cocaine, marijuana, methamphetamine, and heroin in the past 30 days. (See "Asking about Drug Market Participation" for an explanation of the development and phrasing of the question.) As measured by percentages of arrestees who participated, the marijuana market was the largest among the five drugs. It is a finding consistent with earlier ADAM data. Among all sites, 44 percent of arrestees, on average (median),[6] participated in the market for this drug in the month before their arrest. The range was 31 percent (Laredo) to 51 percent (Cleveland). In every site except one (Laredo), the percentage of marijuana market participants was higher than for any of the other four drugs. (See Exhibit 4-1.) Market participation for the other drugs was much lower. An average 15 percent of adult male arrestees participated in the crack cocaine market, with the range 5 percent (San Antonio) to 26 percent (Atlanta). For powder cocaine, an average 15 percent participated, with a range of 4 percent (Sacramento) to 35 percent (Laredo). Heroin attracted 5 percent of adult male arrestees as market participants, with the range zero (Charlotte) to 24 percent (Chicago). And for methamphetamine, 3 percent of adult male arrestees participated in the market, with the range zero (Fort Lauderdale) to 32 percent (Honolulu). (See Appendix Table 4-1, which presents weighted and unweighted numbers of participants as well as percentages.) Paying for drugs The dollar value of a drug transaction can be difficult to calculate. When questions about drug acquisition were field tested by ADAM in focus groups of arrestees, the answers confirmed what ethnographers have often reported: a substantial portion of the drug trade at the street level consists of combinations of goods and services exchanged in addition to or in place of cash. For example, to buy heroin, someone might pay $25 plus a radio for five "dime bags." If only the cash part of this transaction were taken into account, the assumption would be that five bags were worth $25. In fact, they were sold for the equivalent street value of about $50 (that is, $25 plus the cash value of the radio). Other focus group participants said they received a specified amount of drugs in exchange for sexual favors or services, such as transporting drugs or messages and steering customers to the seller. The "value" of the drugs on the market remains the same; it is simply paid for differently. Because the value of goods and services must be taken into account, ADAM examines cash and noncash transactions, as well as transactions that combine the two.[7] Fairly large proportions of market participants did not rely solely on cash to obtain marijuana, crack cocaine, or powder cocaine.[8] (See Appendix Table 4-2.) This was particularly true for marijuana. Marijuana market participants at most of the sites were more likely to have used noncash only transactions than to have paid cash. In half the sites, 43 percent or more used noncash means to obtain this drug, while 34 percent, on average, used combination (cash and noncash) transactions, and 23 percent used cash-only transactions. (See also Exhibit 4-2.) Conversely, cash-only transactions were more common in the crack and powder cocaine markets. For both these drugs, the proportions who paid cash were higher than the proportions who paid cash for marijuana (in half the sites, some 40 percent or more of arrestees paid cash). The proportion of arrestees who obtained crack by noncash means was on average 17 percent among the sites. By contrast, for powder cocaine, the proportion who obtained the drug by noncash means was almost twice as large--33 percent among the sites. Cash-only transactions The marijuana market was the one least likely to involve cash-only transactions. The proportion of arrestees who paid cash for this substance was lower than the proportions who did so for crack or powder cocaine. In the marijuana market, the proportion of arrestees who paid cash exceeded one-third in only 6 of the 23 sites analyzed. (See Appendix Table 4-2.) In both the crack and powder cocaine markets, the proportions paying cash for these drugs exceeded one-third in almost all sites (17 of the 23 sites and 18 of the 23 sites, respectively). Noncash-only transactions Among the various types of noncash transactions, the most common was receiving it as a "gift" (that is, paying nothing for it). Examples of gifts are marijuana joints given or shared at a party or sharing crack. Gifts dominated noncash transactions for all three drugs. For crack, the proportions of arrestees who said they received this drug as a gift was at least 56 percent in half the sites. (See Exhibit 4-3 and Appendix Table 4-3.) Gift-giving was even more pronounced in marijuana and powder cocaine transactions. Of noncash marijuana transactions, 76 percent on average involved receiving the drug as a gift. The proportion who received marijuana as a gift was greater than 60 percent in all sites. Powder cocaine was received as a gift by about two-thirds (68 percent) of arrestees who used noncash transactions to obtain this drug. In almost all sites (20 of the 23) the proportion exceeded 60 percent. After gifts, the next most common method of obtaining drugs was to buy on credit and pay cash later.[9] It was not a close second, however. For crack, in half the sites 11 percent of the noncash transactions involved credit with cash paid later. The figures for powder cocaine and marijuana were 7 percent and 5 percent, respectively. Cash and noncash combined ADAM measures three types of "combination" drug transactions. One consists of two separate transactions, one cash and one noncash. The second combination consists of a single transaction in which the buyer simultaneously pays in both cash and noncash (for example, $5 and a watch). The third consists of two transactions, one involving noncash payment and the other both cash and noncash together.[10] Of the markets for the three drugs, crack and marijuana were those in which the proportion of arrestees who used combination transactions was highest. In the crack cocaine market, 41 percent or more of arrestees in half the sites used a combination of cash and noncash, with the range 9 percent (New York) to 53 percent (Anchorage). (See Appendix Table 4-2.) In 17 of the 23 sites, the proportion who obtained crack this way exceeded one-third. For marijuana, the proportion who obtained the drug by combination transactions was similar to crack cocaine, averaging 34 percent among the sites. In 13 of the 23 sites, more than one-third of marijuana market participants obtained the drug this way. In the powder cocaine market, the proportions who used combination transactions were generally lower than for the other two drugs. Just under one-fourth of arrestees on average obtained powder cocaine this way, with the proportion barely surpassing 30 percent in only 3 sites. The type of dominant transaction varied by site. In New York City, for example, cash-only transactions dominated the markets for all three drugs (in the crack and powder cocaine markets, 90 percent of arrestees paid cash only, and 79 percent paid cash only in the marijuana market). The same was true of three other sites--Cleveland, Fort Lauderdale, and Miami--though not by margins as wide as in New York. Noncash exchangers dominated the markets for all three drugs in only one site--Spokane. Combination exchangers did not dominate all three drug markets in any of the 23 sites. Method of contacting drug dealers Arrestees were asked how they contacted dealers to obtain drugs. The methods of contact varied, and for each of the three drug markets, there were also differences between cash and noncash exchanges. (See Table 4-1 for the averages of the sites.) Among arrestees who paid cash for marijuana, the largest proportion used a phone or pager, with the next largest proportion going to someone's house or apartment. The averages among the sites for these two types of dealer contacts were 36 percent and 25 percent, respectively. By contrast, among arrestees who used noncash exchanges to obtain this drug, the proportion who contacted the dealer at work or in a social setting was by far the largest among the various methods of contact. In half the sites, 48 percent or more contacted the dealer this way, while for the other types of contact the proportions were much lower. (See Appendix Table 4-4.) For cash purchases of crack cocaine the picture was somewhat different. In contrast to marijuana, for crack the most common method was to approach a dealer in a public place. The proportion of arrestees who paid cash for crack cocaine this way was 43 percent or more in half the sites--more than double the proportion who bought marijuana this way. The second most popular way to obtain crack with cash was by contacting a dealer by phone or pager. The average was 30 percent among the sites. Ways to contact dealers for noncash crack transactions resembled those for marijuana: Contacts were most often made at work or in a social setting, with the next most frequent method of contact approaching a dealer in public (averages were 30 percent and 23 percent, respectively, among the sites). (See Appendix Table 4-5.) Much as in the cash marijuana market, cash purchases for powder cocaine tended to be made by phone or pager. In half the sites, almost half the arrestees said they used a phone or pager to buy powder cocaine in cash transactions. Noncash transactions of powder cocaine resembled those for marijuana and crack cocaine, with the largest proportion of arrestees (44 percent among the sites, on average) saying they obtained the drug at work or social settings. (See Appendix Table 4-6.) Whereas large proportions of arrestees obtained drugs by noncash means at work or in social settings, this was not the case for cash purchases. Overall, only small proportions of arrestees paid cash for any of the three drugs at work or in social settings. (See Appendix Tables 4-4, 4-5, and 4-6). And only small proportions of arrestees engaged in noncash transactions by going to someone's house or apartment to obtain any of the three drugs. The findings on noncash methods suggest they have two identifiable characteristics. First, the noncash events were, in most cases, opportunistic; that is, they occurred when someone happened to be at a social setting or at work. In other words, they may not have been planned. Second, the arrestees who obtained drugs through noncash transactions were acquainted with those who supplied them, suggesting they may be connected to other drug market participants. The cash methods suggest a well-structured network of contacts that include knowledge of dealers, as well as their beeper numbers, phone numbers, and addresses. Some sites diverged from the patterns noted above. For example, although marijuana cash purchases were most often made by phone or pager in most sites, in some this was not the case. In eight sites, the most common method used by arrestees who paid cash for marijuana was approaching a dealer in a public place. These sites were Atlanta, Cleveland, Denver, Fort Lauderdale, Miami, Minneapolis, New Orleans, and New York. Also, while cash purchases for powder cocaine were most often made by phone or pager, this was not the case in Atlanta, Cleveland, Fort Lauderdale, Miami, New Orleans, New York, and San Jose. In these cities, approaching a dealer in a public place was the most frequent way to contact dealers. And while cash purchases of crack cocaine were most commonly made by approaching dealers in public places, in Albuquerque, Anchorage, Denver, Indianapolis, Portland, Salt Lake City, and Spokane, the most common method was to use a phone or pager. In four southwestern sites (Dallas, Oklahoma City, Phoenix, and Tucson), going to someone's house or apartment was the most common method of buying crack with cash. Relationship of buyers to sellers[11] Do arrestees who obtain drugs have a regular dealer? Do they have only one dealer or several? Does the number of dealers vary with the drug obtained? With the ADAM redesign, these and other questions about the relationships between buyers and sellers are being explored. Crack cocaine was the drug whose purchase in cash was most likely to involve two or more dealers. In half the sites, 65 percent or more of adult male arrestees said they bought crack from two or more dealers in the month before their arrest. The figures for marijuana and powder cocaine were 42 percent and 34 percent, respectively. (See Appendix Table 4-7. Exhibit 4-4 presents the proportions of arrestees who made cash purchases from two or more dealers.) This pattern is particularly evident in sites like Houston (where 70 percent of arrestees used two or more dealers to buy crack, compared to 37 percent who did so when buying marijuana and 9 percent who did so when buying powder cocaine), Phoenix (where 59 percent of arrestees used two or more dealers to buy crack, compared to 19 percent for powder cocaine), and San Jose (where 71 percent used two or more dealers to buy crack, compared to the 15 percent who did so to buy powder cocaine). The large proportions of arrestees who used two or more dealers to buy crack help explain why the average number of dealers used by crack cocaine market participants was the highest among all three drugs. On average, crack market participants used 3.2 dealers, a figure higher than the 1.9 dealers used by marijuana market participants and the 1.8 used by powder cocaine market participants. The ADAM data reveal that particularly for crack cocaine purchases made in cash, arrestees often had more than two dealers, but they also show that arrestees commonly had a regular source, rather than either someone they dealt with occasionally or a new dealer. (See Exhibit 4-5.) This was the case in the markets for all three drugs studied. In the powder cocaine market, 61 percent or more of arrestees bought from a regular source. The range was 41 percent (Minneapolis) to 75 percent (Phoenix). In the crack cocaine market, the proportion who had a regular source was 49 percent or more in half the sites, with the range 19 percent (San Jose) to 62 percent (Tucson). In the marijuana market, the proportion having a regular source was 46 percent or more in half the sites, with the range 36 percent (Salt Lake City) to 69 percent (New York). (See Appendix Table 4-8.) For all three drugs, the percentage who obtained drugs from a regular source exceeded the percentage who obtained them from an occasional source, suggesting a certain stability in the markets. The percentages of arrestees who made their most recent cash purchase from a new source were fairly similar for all three drugs studied. On average, 19 percent used a new source for crack; for marijuana the figure was 16 percent, and for powder cocaine it was 13 percent. Drug markets often have go-betweens or couriers who facilitate purchases and also serve as "layers of protection" to preserve the seller's anonymity. The ADAM analysis revealed that in none of the three drug markets studied was there extensive use of these facilitators by arrestees. On average, in the marijuana market, 3 percent of arrestees used couriers, in the crack cocaine market 3 percent used couriers, and in the powder cocaine market, 4 percent did so.[12] (See Appendix Table 4-9.) In the crack cocaine market, the use of drug couriers ranged from none (Houston) to 12 percent (Denver). In the marijuana market, the range was none (Fort Lauderdale) to 7 percent (Salt Lake City and San Diego). And in the powder cocaine market the range was none (Albuquerque, Cleveland, Minneapolis, New York, and San Diego) to 12 percent (Salt Lake City). Are outdoor purchases the norm? The emergence of crack cocaine markets in urban areas of the United States in the late 1980s and early 1990s brought the environmental context to the forefront as an important variable in drug market dynamics. Before the crack cocaine epidemic, drugs were typically sold indoors. But in many cities crack was sold in open air markets. The media was quick to report on the high levels of violence attendant on the emerging trafficking in crack cocaine.[13] Researchers who subsequently documented the violence saw it as related to the characteristics of the substance itself, the nature of the market, and the marketing of the product.[14] When violent crime in urban areas began to decline in the early 1990s, some observers suggested it was to some extent related to the changing nature of the crack markets. One change was that open air sales were being replaced by indoor transactions, which were considered safer for buyers and sellers.[15] With ADAM now collecting information about drug markets, it is possible to assess the extent to which particular drugs in particular places at particular times are sold outdoors or indoors. Extent of outdoor sales For crack, the image of the open air market is confirmed in many sites. The proportion of arrestees who bought crack outdoors was 50 percent or more in 10 of the 23 sites. (See Appendix Table 4-10 and Exhibit 4-6.) In half the sites, 44 percent or more of arrestees bought crack this way, and the range was wide: 19 percent (Spokane) to 88 percent (New York). For marijuana, by contrast, the proportion who made outdoor purchases was 50 percent or more in only three sites. In half the sites, 31 percent or more bought marijuana outdoors. For powder cocaine, the proportion making purchases outdoors was as low: In only four sites did it exceed 50 percent. The average among the sites was about the same as for marijuana. These findings may reflect differences in the operations of the market for the various drugs and differences within specific sites. In New York and Cleveland, for example, outdoor purchasing dominated the markets for all three drugs. At the other end of the continuum were several sites where the proportion of arrestees who bought drugs indoors exceeded 70 percent for all three drugs. (These are Albuquerque, Anchorage, Dallas, Oklahoma City, Phoenix, Salt Lake City, Spokane, and Tucson). Thus, irrespective of type of drug, in some sites high proportions of arrestees buy drugs outdoors and in others high proportions buy drugs indoors. These differences also illustrate the value of ADAM's focus on individual sites--differences that would be obscured in nationwide or regional analyses of drug use patterns. The drug-market neighborhood The role of the drug trade in promoting neighborhood instability has not been studied often or systematically. Community activists have noted that outsiders (people who do not live in the neighborhood) come into the community to buy drugs. The ADAM data confirm their observations and bring to light new information about drugs as a destabilizing force. For all three drugs studied here, about half of all market participants said that at least one transaction took place outside their own neighborhood.[16] (see Exhibit 4-7. Appendix Table 4-11 presents site-by-site findings.) What makes a purchase attempt fail? Considerable law enforcement resources have been spent on making it more difficult for drug users to find and obtain illicit drugs.[17] According to the ADAM data, a surprisingly high percentage of arrestees have no difficulty completing a drug transaction. (See Exhibit 4-8 for the ranges and averages and Appendix Table 4-12 for site-by-site data).[18] Marijuana is the drug for which the percentage of arrestees reporting one or more failed cash transactions was highest. In half the sites 39 percent or more said they failed in an attempt to buy marijuana, with the range 12 percent (New York) to 53 percent (Indianapolis). Crack cocaine was a close second in failed transactions. In attempting to buy this drug, 37 percent or more of arrestees in half the sites said they failed. Failure rates for crack ranged from a low of 9 percent (New York) to a high of 59 percent (Oklahoma City). In attempting to buy powder cocaine, 29 percent or more of arrestees in half the sites failed, with the range 11 percent (New York) to 39 percent (Denver). Further research is likely to reveal more insights into these failed transactions, particularly with respect to the differences among the sites and among the various drugs. Not only did relatively few transactions end in failure, but when they did, police activity was rarely cited as the reason. (See Appendix Table 4-13.) The proportion of arrestees who said the presence of the police had deterred them from buying drugs was generally low. For marijuana, 6 percent or fewer of arrestees in half the sites cited the police as a deterrence; for both crack and powder cocaine the figure was 11 percent. (See Table 4-2.) There are a few notable exceptions to the evident ease with which drugs are obtained. In Miami, for example, where more than one-fourth of the arrestees said their transactions for powder cocaine had failed, a fairly large proportion (just under one-third--32 percent) ascribed their failure to police presence. (See Appendix Table 4-13.) There are similar exceptions for crack purchasing. In Houston, more than half (52 percent) the crack market participants said their transactions failed, and of these, 28 percent attributed the failure to police activity. In New York, attempts to buy any of these drugs ended in failure for relatively small proportions of arrestees, but even here the police role was notable. For crack cocaine, 9 percent of arrestees said the transactions failed, with police cited as the reason by 45 percent. For powder cocaine, the figures were 11 percent and 39 percent, and for marijuana, 12 percent and 41 percent. If police activity was not directly responsible for deterring drug transactions, it may have had an indirect effect on the availability of drugs, even if few arrestees cited it. In Oklahoma City, 59 percent of the crack market participants said they had experienced a failed transaction at least once during the past month, but only 2 percent attributed the failures to police activity. Here, 17 percent of the arrestees cited the reason for failure as lack of availability of dealers; 42 percent said the dealer had no crack to sell; 13 percent said the quality they wanted was not high enough; and 26 percent noted other reasons. In these cases, it may have been that police activity against dealers prevented them from being able to meet customers' needs. Transaction dynamics: frequency, volume, and price In this section on the dynamics of market transactions, the focus is on the sites where markets were very active for all three drugs studied. Level of market activity was measured by calculating the number of arrestees who said they had obtained drugs in the past 30 days either by cash or noncash transactions. In order to minimize bias that would be introduced if there were too few cases, 100 arrestees was set as the minimum number of unweighted cases for use in the analysis. (See "Analyzing Drug Transaction Dynamics" for the definition of an active market and a discussion of the questions asked of arrestees in order to elicit information about transaction dynamics.) Thus, the transaction dynamics analysis was limited to the four ADAM sites where this cutoff point was reached in the markets for all three drugs: Miami, Phoenix, Seattle, and Tucson. (See Appendix Table 4-1.) Before examining these dynamics it is essential to distinguish between the proportions of cash and noncash transactions, because the analyses differentiated between these two types of transactions. The proportions varied considerably by site as well as by drug.[19] (See Table 4-3.) In two of the active market sites, Phoenix and Tucson, marijuana transactions were conducted for the most part on a cash basis. In the two others, Miami and Seattle, cash and noncash transactions for this drug were more evenly divided. Except in Miami, the markets for crack and powder cocaine were about evenly divided between cash and noncash. Transaction frequency was defined as the number of times that transactions involving the same drug took place on the same day. Because Phoenix and Tucson are close geographically, the expectation might be that they were in this respect distinct from the other two sites. However, there was little variation among the four sites in the number of times per day arrestees said they obtained any of the three drugs by either cash or noncash means.[20] (See Table 4-4.) Interaction among the most recent transaction, the number of transactions per day, and transaction days per month were investigated to produce a measure of the average (mean) number of days a month in which a given drug was obtained. On this measure, distinctions emerged among the four sites, although there is one striking similarity. (See Table 4-5.) In all four sites there are cumulatively 25 to 100 percent more cash and noncash crack cocaine transaction days than powder cocaine and marijuana days. In looking at the interactions among these variables, it is evident that transactions in the crack cocaine market were two to three times higher than the highest rates for the other two drugs. On average, arrestees obtained crack almost twice a day every day. This could, of course, mean obtaining the drug many times during binge days and one or no times on other days; however, it is clear that the level of market activity for crack was higher. For marijuana and powder cocaine in Miami, the total number of transactions per month are similar (33 and 30, respectively), but in the other three active drug market sites, the numbers were very different for these two drugs, with differences close to a 2:1 ratio. (See Table 4-6.) Analysis of the number of buyers in the market and the frequency of their transactions revealed that a relatively small proportion of arrestees--8 to 19 percent--generated more than half of all drug transactions in all four sites. (See Table 4-7.) Market size was measured by the dollar value of cash transactions and reflected the 30-day drug market involvement of each site's arrestee population. It was calculated by multiplying the dollar value of the arrestee's most recent cash transaction by the number of transactions on the day of that transaction and then by the number of transaction days per 30 days. In all four sites, the market size of crack cocaine was by far the largest. (See Table 4-8.) This approach is a first step toward estimating the ADAM population's involvement in the drug markets of the catchment areas. It has some limitations. One is that the dollar value of noncash transactions needs to be estimated. Also, because ADAM studies arrestees only, the figures presented here reflect only data for that population. To use ADAM data to determine total market size, other approaches must be taken. For example, it may be possible to apply the method of estimating hardcore drug use to the ADAM data to obtain a figure closer to the size of the market. Refining the analysis As a result of the redesign of the ADAM program, it is possible, for the first time, to systematically collect data about drug markets on an ongoing basis at the local level. The approach used by ADAM was designed to produce a representative account of the nature of drug exchanges among arrestees. There are a variety of applications for these data. One example would be using the data to estimate success in drug sweeps. After conducting a major local sweep/crackdown of local drug dealers, a police department could review the ADAM data on total market size before and after the sweep. A reduction in the dollar value and total number of exchanges in the market after the sweep would be one possible indicator of success. The estimates presented here are for the part of the drug market in which ADAM arrestees participate. Presumably, there are people who participate in the drug markets analyzed here who did not get arrested and thus did not become part of the ADAM sample. For this reason the ADAM analyses will need to be supplemented and integrated with other methods to account for the entire drug market in the selected catchment areas. Information collected by ethnographers, including qualitative data on people who use drugs but never get arrested, might prove useful to understanding the size of the entire market.[21] The ADAM program is currently developing a modeling strategy that would permit drawing inferences from hardcore users' market participation and applying them to the broader population. Researchers could use this strategy, which involves modeling the rate at which hardcore market participants are arrested, to infer the size of the entire market. (The logic of this method is presented in detail in Chapter 9.) * Bruce G. Taylor, Ph.D., is Deputy Director of the ADAM program. Michael Costa is a Senior Analyst with Abt Associates Inc. Notes 1. See, for example, Johnson, B., et al., Taking Care of Business: The Economics of Crime by Heroin Abusers, Lexington, MA: D.C. Heath and Company,1985; Needle, R. and A. Mills, Drug Procurement Practices of the Out-of-Treatment Chronic Drug Abuser, Rockville, MD: National Institutes of Health, National Institute on Drug Abuse, 1994; Edmunds, M., M. Hough, and N. Urquia, Tracking Local Drug Markets, Home Office Research Study No. 80, London: Home Office, 1996. Ethnographic studies include Preble, E. and J. Casey, "Taking Care of Business: The Heroin User's Life on the Street," International Journal of the Addictions 4 (1969): 1-24; Curtis, R. and M. Svidorff, "The Social Organization of Street-Level Drug Markets and Its Impact on the Displacement Effect," in Crime Displacement: The Other Side of Prevention, ed. R.P. McNamara, East Rockaway, NY: Cummings and Hathaway, 1994; Curtis, R. et al., "Street-Level Drug Market Structure and HIV Risk," Social Networks 17 (1995): 219-228; and Williams, T., The Cocaine Kids, Reading, MA: Addison-Wesley, 1989. 2. Riley, J., Crack, Powder Cocaine and Heroin: Drug Purchases and Use Patterns in Six U.S. Cities, Washington, DC: U.S. Department of Justice, National Institute of Justice, and Executive Office of the President, Office of National Drug Control Policy, 1997, NCJ 167265. 3. Brownstein, H.H., "Drug Distribution and Sales as a Work System," in Encyclopedia of Criminology and Deviant Behavior, Volume 4: Self Destructive Behavior and Disvalued Identity, eds. C. Faupel and P.M. Roman, Philadelphia: Taylor and Francis, 2000: 224-227. 4. Brownstein, H.H., S.M. Crimmins, and B.J. Spunt, "A Conceptual Framework for Operationalizing the Relationship Between Violence and Drug Market Stability," Contemporary Drug Problems 27 (2000): 867-890; Golub, A. and B. Johnson, Crack's Decline: Some Surprises Across U.S. Cities, Research in Brief, Washington, DC: U.S. Department of Justice, National Institute of Justice, 1997, NCJ 165707; and Johnson, B.D., A. Hamid, and H. Sanbria, "Emerging Models of Crack Distribution," in Drugs, Crime, and Social Policy: Research, Issues, and Concerns, ed. T. Mieczkowski, Boston: Allyn and Bacon, 1992: 56-78. 5. The proportions of arrestees who use heroin and methamphetamine are much smaller. Separate papers will be written by ADAM staff to examine the market for these two drugs. 6. Unless otherwise indicated, averages are expressed as medians throughout this report. 7. In addition to multiple transaction types for obtaining drugs (cash, noncash, and a combination of the two), there are multiple markets, one for each drug. One person might participate in the markets for all three drugs (marijuana, crack cocaine, and methamphetamine) or, alternatively, in the market for only one drug. For purposes of analysis, arrestees are categorized as having engaged in only one type of transaction for each drug market in which they participated. For example, an arrestee who made only cash purchases for marijuana is classified in one group, an arrestee who made only noncash exchanges for marijuana is classified in a second group, and an arrestee who made both cash and noncash transactions for marijuana is classified in a third group. This classification scheme does not exclude individuals who participated in multiple drug markets by different transaction methods. For example, an arrestee might obtain marijuana by noncash means only, but pay cash for crack. This categorization should help law enforcement agencies approximate the percentage of offenders involved in the markets for the various types of drugs and the type of transactions in which they engage to obtain them. 8. Except in the final section of this chapter (on the dynamics of market transactions), 23 ADAM sites were selected for analysis of drug market participation. These were the sites in which at least 50 (unweighted) arrestees participated in the drug market for all three drugs analyzed. They are listed on Appendix Table 4-2. 9. The proportion who used "other" types of noncash transactions was higher than the proportion who bought on credit with cash paid later. 10. All three types of transactions are included in the "combination" category because at almost all sites, all combination transactions involved two separate transactions, one cash only and one noncash only. 11. Resource constraints of the ADAM program limited the analysis in this section to cash purchases only. 12. The finding that couriers were not often used does not preclude the possibility that ADAM did not measure them accurately. ADAM attempted to confirm media accounts of young men in the urban cores using beepers and cell phones who operate as couriers and insulate established drug dealers from direct involvement with users. For ADAM, this is a new area of investigation, and the program will continue to explore ways to measure the drug courier phenomenon. 13. Massing, M., "Crack's Destructive Sprint Across America," New York Times Magazine, October 1989: 38, 40-1, 58, 60, 62; and Witkin, G., "The Men Who Created Crack," U.S. News and World Report, August 1991: 44- 53. 14. Belenko, S., "The Impact of Drug Offenders on the Criminal Justice System," in Drugs, Crime and the Criminal Justice System, ed. R.A. Weisheit, Cincinnati: Anderson, 1990: 27-48; Brownstein, H.H., et al., "The Relationship of Drugs, Drug Trafficking, and Drug Traffickers to Homicide," Journal of Crime and Justice, 15 (1992): 25-44; Fagan, J. and K.L. Chin, "Violence as Regulation and Social Control in the Distribution of Crack," in Drugs and Violence: Causes, Correlates, and Consequences, NIDA Research Monograph No. 103, eds. M. de la Rosa, E.Y. Lambert, and B. Gropper, Rockville, MD: National Institute on Drug Abuse, 1990: 8-34; Goldstein, P.J., H.H. Brownstein, and P.J. Ryan, "Drug-Related Homicide in New York: 1984 and 1988," Crime and Delinquency 38 (1992): 459-476; and Goldstein, P.J., et al., "Crack and Homicide in New York City: A Conceptually Based Event Analysis," Contemporary Drug Problems 4 (Winter 1989): 651-87. 15. Compare Blumstein, A., Youth Violence, Guns, and the Illicit-Drug Industry, Washington, D.C.: U.S. Department of Justice, National Institute of Justice, 1995, NCJ 162687; Brownstein, H.H., The Rise and Fall of a Violent Crime Wave--Crack Cocaine and the Social Construction of a Crime Problem, Guilderland, NY: Harrow and Heston, 1996; Golub, A. and B. Johnson, The Rise of Marijuana as the Drug of Choice Among Youthful Adult Arrestees, Research in Brief, Washington, DC: U.S. Department of Justice, National Institute of Justice, 2001, NCJ 187490; Lattimore, P.K., et al., Homicide in Eight Cities: Trends, Context, and Policy Implications, Research Report, Washington, DC: U.S. Department of Justice, National Institute of Justice, 1997, NCJ 167262. 16. Because the question was," Did you buy it [name of drug] in the neighborhood where you live or outside your neighborhood?" the definition of "neighborhood" reflected the arrestees' perceptions. 17. Sviridoff, M. and S.T. Hillsman, "Assessing the Community Effects of Tactical Narcotics Teams," in Drugs and Crime: Evaluating Public Policy Initiatives, eds. D.L. MacKenzie and C.D. Uchida, Thousand Oaks, CA: Sage, 1994:114-128; and Office of National Drug Control Policy, National Drug Control Strategy: 2000 Annual Report, Washington, DC: Executive Office of the President, 2000, NCJ 180082. 18. Arrestees were asked, "Was there a time in the past 30 days when you tried to buy [name of drug] and had the cash, but you did not buy any?" Those who did not buy were asked why. 19. In this section, the category "cash transactions" includes cash-only and combination cash and noncash transactions. 20. These data were first adjusted to eliminate anomalous cases in which unusually large numbers would skew the means. This was done by placing caps on the number of events counted per day. On the basis of the distribution of the data, the number of transactions for marijuana were recoded, from much higher numbers, to 1, 2, 3, or 4 or more; crack was recoded to 1, 2, 3, 4, 5, 6, or 7 or more; powder cocaine was recoded to 1, 2, 3, or 4 or more. 21. Curtis, R., "Drug Markets on the Lower East Side of Manhattan, NYC," final report submitted to the National Institute of Justice, U.S. Department of Justice, 1999. ---------------------------- Drug Market Monitoring by the DEA Other than ADAM, the only other major program that monitors local drug markets is the Drug Enforcement Administration's (DEA's) price/purity tracking system. It has the following components: o The System to Retrieve Information from Drug Evidence (STRIDE) data system o The Domestic Monitoring Program (DMP) o The Heroin Signature Program (HSP). System components The STRIDE system contains data on the price and purity of outdoor drug purchases made by informants hired by the DEA. It is not a research program. STRIDE data are collected for operational purposes and are obtained by recording nonrandom drug acquisitions made in support of criminal investigations. In addition to Federal agencies, the Metropolitan Police Department of Washington, D.C., participates in this program. The DMP is a heroin purchase program that provides data on the purity, price, and origin of retail-level heroin available in major metropolitan areas of the country. The data come from ten $100 purchases made quarterly in 22 locations. The HSP uses laboratory analysis to determine the geographic source of heroin made from seizures at U.S. ports of entry and from a sample of other seizures and purchases by DEA and FBI agents. DEA data in research STRIDE, DMP, and HSP data are used by researchers. STRIDE data have been used to estimate the amount of pure drug purchased per dollar spent. However, the data cannot reveal what dollar expenditures are typical in retail drug markets because the distribution of purchases made by police, in STRIDE, is not the same as the distribution of purchases by other buyers. STRIDE also does not account for drug purchases made indoors. By contrast, ADAM makes it possible to estimate the distribution of dollar expenditures for illicit drugs by analyzing the responses made by arrestees to an array of questions about local drug markets. ---------------------------- Asking about Drug Market Participation When the ADAM redesign was under way, early testing of the new interview questions about drug market participation revealed that arrestees were often unable to accurately describe a "typical" exchange in which they obtained a drug. They either resorted to "war stories" of "best scores" or tried to describe an average transaction on the basis of a number of different transactions. The pilot data also indicated that among arrestees involved in the drug market, drug purchases were frequent. Many obtained drugs several times a week and some did so several times per day, employing a wide range of methods and types of exchanges. As with all events that take place frequently, separate episodes blend together and did so in the interviewees' memories. This made it difficult to create an accurate "average" transaction. The new interview question For these reasons, "typical" was not a cognitively feasible term for describing an arrestee's drug market transaction. Instead, arrestees were asked to describe the last (most recent) instance in which they obtained drugs in the past 30 days through "cash" and "noncash" transactions (e.g., by trading property or sex). In this way, the arrestee's attention focused on one real event--the last one in the 30-day period, and he was given the opportunity to describe it accurately. Overall, there is little reason to believe that the "last" transaction is necessarily different from the other transactions, and thus the approach should produce a representative account of the nature of drug exchanges among arrestees. Sources for the question redesign In designing the new drug market section of the ADAM survey, the ADAM team consulted with researchers and practitioners who had expertise in the area of drug markets. Additionally, focus groups were conducted among street-level drug marketers, drug buyers, and sellers who had recently been arrested. The focus groups brought to light information that proved essential to the development of the new drug market questions. For example, the ADAM team decided on the basis of the focus groups that it would be very difficult to collect valid data on direct involvement in selling drugs. People were understandably reluctant to discuss this type of illegal behavior. For that reason, the drug market section of the interview focused on buyers' views of market dynamics. ---------------------------- Analyzing Drug Transaction Dynamics For each drug studied--marijuana, crack, and powder--the questions were intended to yield information about the frequency of transactions and the amounts obtained. The focus was on the most recent transaction, with the data gathered including cost, number and types of units of drugs obtained (for example, one or two bags of powder cocaine), and amount kept for personal use. Once this information was obtained, the arrestees were asked the number of times on the day of transaction that they obtained the drug they named. They were then asked about the number of days they obtained that drug in the seven days before their arrest and, finally, the number of days they obtained that drug in the past 30 days. This line of questioning was pursued separately for marijuana, crack, and powder cocaine obtained through either cash or noncash exchanges. In order to compute a total, information about frequency and units was used to calculate the number of events in one day and in 30 days. For instance, if the arrestee said he obtained two units of a drug in the most recent transaction and completed three transactions that day, and then reported 15 transaction days in a month, the total would be 90 units in 45 transactions in a 30-day period. Selection of most recent transaction as the unit of analysis compels the respondent to choose a random transaction, rather than one of his or her choosing. In some instances, data on most recent events will reflect transactions that are inordinately large, small, or biased in some other way. The distribution of cases over time (in concert with weighting of the data) will minimize the effect of bias introduced by any one respondent's recollection of the most recent event. ---------------------------- V. Drug Use Among Adult Female Arrestees by Bruce G. Taylor, Phyllis J. Newton, and Henry H. Brownstein* Of the 14 million people arrested in the United States in 2000, almost 1.6 million were arrested for drug abuse violations.[1] Women constituted only about 20 percent of these arrestees and a slightly smaller percentage of drug offenders. Nonetheless, at 272,000, the number of women charged with drug offenses is not inconsequential.[2] A considerable amount of research was conducted in the last decades of the 20th century to understand the relationship between drugs and crime, but most of it focused on male drug users and male offenders.[3] Earlier research on women's involvement in drugs and crime tended to focus on prostitution,[4] but other than this, knowledge of women offenders and drug use by women remains limited.[5] More recently, some attention has been paid to women's involvement in crime as it relates to participation in drug markets, but much of this research has been based on limited data.[6] With the redesigned ADAM, more information about women's involvement in drugs and crime will be forthcoming. In 2000, the new, expanded ADAM interview instrument was used with female as well as male arrestees. Urinalysis continues to be used to detect recent drug use and, as in the past, during the interview, the women, like the men, were also asked if they used drugs. In the new instrument, arrestees are now asked about their experience with treatment and participation in drug markets; and they are also asked a series of questions to assess whether they are at risk for dependence on drugs. Unlike the data for male arrestees, the data for women were not gathered with probability-based sampling. Rather, the convenience sampling of ADAM's forerunner, the Drug Use Forecasting (DUF) program, was used. Because that creates uncertainty about the representativeness of the data,[7] and because confidence levels cannot be established, caution should be used in interpreting the findings. Additionally, there were relatively few women arrestees, making the sample sizes in some analyses very small and limiting the number of sites analyzed to 29. (For a discussion of the size of the samples of adult female arrestees, see "Sample Size--Issues.") Extent of drug use as detected by urinalysis As in previous years, the levels of drug use detected by urinalysis were high. In all but three of the 29 sites where data on women arrestees were analyzed, more than half of them tested positive for recent use of at least one NIDA-5 drug (cocaine, opiates, marijuana, methamphetamine, or phencyclidine [PCP]).[8] In half the sites, 63 percent or more tested positive, with the rates ranging from 31 percent (Laredo, where 18 women arrestees tested positive) to 80 percent (Chicago, where 298 tested positive). (See Appendix Table 5- 2.) Of the ten drugs analyzed by ADAM,[9] four--cocaine (both crack and powder), marijuana, methamphetamine, and opiates--were the ones used by the highest percentages of women arrestees on average in the ADAM sites. Overall, cocaine (undistinguished here between crack and powder) was the drug most commonly used by adult female arrestees, followed by marijuana, opiates, methamphetamine, and PCP. (See Exhibit 5-1.) About one-third of adult female arrestees, on average,[10] had used cocaine. The proportion who tested positive for this drug was lowest in San Jose (8 percent, or 4 female arrestees), with Chicago at the top of the range (59 percent, or 222). Marijuana was the next most popular drug, with more than one in four adult female arrestees (27 percent, on average) testing positive for it in half the sites. The lowest rate of marijuana use was recorded in Laredo, where 17 percent, or 10 female arrestees, tested positive. The highest rate was in Oklahoma City, where 45 percent, or 135 females, tested positive. The West is the region where methamphetamine use among adult female arrestees was most prevalent in 2000. This was also the case for men. Confirmatory tests[11] indicated the proportion who tested positive for methamphetamine was highest in Honolulu (47 percent, or 34 arrestees), followed by San Jose (40 percent, or 20 arrestees), Salt Lake City and San Diego (29 percent, or 22 and 77 arrestees, respectively), Phoenix (24 percent, or 93 arrestees), Portland (24 percent, or 52 arrestees), and Las Vegas (21 percent, or 76 arrestees). In 8 of the 29 sites analyzed, those largely in the eastern part of the country (New York, Fort Lauderdale, Detroit, Philadelphia, Cleveland, Atlanta, Laredo, and Albany/New York Capital Area), there was no methamphetamine use among female arrestees. Few women arrestees tested positive for opiates. The average was 7 percent among the sites, with the range 1 percent (in Omaha, where one woman tested positive) to 40 percent (in Chicago, where 150 tested positive). In addition to Chicago, the sites with double-digit positive rates for opiates were Detroit, Portland, New York, Tucson, Albuquerque, and Philadelphia. No geographic pattern is evident. PCP is used by only a very small percentage of arrestees. In only two sites (Cleveland and Oklahoma City) did the proportion of women testing positive exceed 4 percent. In half the sites none tested positive, although this may have been a function of the small sample size. For the most part, the adult female arrestees who tested positive had used only one of the NIDA-5 drugs. In half the sites, 80 percent or more tested positive for only one. By contrast, the proportion testing positive for multiple drugs was relatively small. In more than half the sites, less than 20 percent of the arrestees had done so, with multiple drug use among the sites ranging from 10 percent (Albany and Houston, with 4 and 6 arrestees, respectively) to 41 percent (Chicago, with 154). As with recent drug use by male arrestees (discussed in Chapter 1), the findings for female arrestees need to be interpreted cautiously, because studies have consistently shown past year or past month[12] polydrug use the norm, with users substituting one drug for another when the drug of choice is scarce, or mixing drugs to counter or moderate the effects of one or the other. And again, for female arrestees, the small size of the sample may explain these anomalous findings. If there were major variations among the sites in drug use by female arrestees, there were also differences among the sites in the age, type of offense, and race of those who tested positive. (See Appendix Tables 5-3a through 5-3d, which present breakdowns by age; Appendix Tables 5-4a through 5-4f, which present breakdowns by offense; and Appendix Tables 5-5a and 5-5b, which present breakdowns by race.) Once again, because these types of analyses generated even fewer cases, the findings should be interpreted cautiously. Demographics and sociodemographics For most of the 29 sites where data on women were analyzed, the largest category of arrestees interviewed were in the oldest age range--36 years of age or older. In half the sites, 35 percent or more were 36 or over. (See Appendix Table 5-6.) In most sites, more than half the adult female arrestees were more than 31 years old. The average age ranged from 28 (Laredo) to 34 (Fort Lauderdale). (See Table 5-1.) In half the sites, 40 percent or more were white, with the proportion of blacks a very close second, at 37 percent on average. Hispanics constituted a much smaller proportion (4 percent, on average), as did "other" (also 4 percent). In some sites, a single racial/ethnic category predominated. Thus, in four sites, at least 60 percent of the women arrestees were white (Salt Lake City, 82 percent; Portland, 73 percent; and Des Moines and Fort Lauderdale, 64 percent); and in seven sites at least 60 percent were black (New Orleans, 85 percent; Chicago, 80 percent; Atlanta, 78 percent; Philadelphia, 73 percent; Cleveland, 70 percent; Detroit, 68 percent; and Houston, 62 percent). In sites in the West, relatively large percentages of the women arrestees identified themselves as Hispanic (Laredo, 71 percent; Albuquerque, 57 percent; San Jose, 37 percent; Denver, 28 percent; Tucson, 27 percent; Phoenix, 21 percent; Los Angeles, 21 percent; and San Diego, 20 percent). In most sites, a fairly high percentage of the women did not have a high school diploma. The proportion without a diploma was 29 percent or more in half the sites, with the range 21 percent (San Jose, where 11 women had none) to 47 percent (120 women, New Orleans). (See Appendix Table 5-7.) With respect to employment status, just under half the women, on average, said they were working (45 percent). The lowest percentage was in Honolulu, where 23 percent (20 women) said they were working; and the highest percentage was in Dallas, where 64 percent (43 women) were working. Also, some women were homeless, with at least 5 percent in half the sites saying that in the month before they were arrested they had no fixed address. In seven sites the proportion of women who were homeless surpassed 10 percent (Denver, Honolulu, Phoenix, Portland, San Diego, San Jose, and Tucson). Many women did not have health insurance at the time of their arrest (the average was 56 percent); many were single (average was 54 percent); and many had a history of arrest (average was 43 percent). Self-reported alcohol and drug use In addition to using urinalysis to detect drug use, ADAM also asks arrestees during the interview about their use of drugs. Of the two methods of detection, urinalysis is the more objective, but because most drugs do not stay in the body long, it can detect use only in the very recent past--depending on the particular drug, no more than a few days or a few weeks. Self-reports of drug use thus complement urinalysis, offering a more retrospective, though less objective, view. If an arrestee uses drugs, but has not done so recently, such use would not be detected by urinalysis, but only by the arrestee's self-reports. In the past, ADAM had asked arrestees about use in the week and the month before the arrest, but the redesigned program also asks about use in the year before the arrest, providing an even longer perspective. Questions about alcohol consumption have also been added. The resulting self-reported data are then used as the basis for analyzing a number of behaviors related to drug and alcohol use. Marijuana was the drug female arrestees were most likely to say they had used in the year before their arrest. In half the sites, 42 percent or more said they had used marijuana. (See Appendix Table 5-8.) It was also the drug they were most likely to say they used in the month before their arrest. In half the sites, one-third or more said they used it the past month. The next most frequently used drug was crack cocaine, which 27 percent of the women arrestees, on average, used in the past year and 23 percent on average used in the past month. Powder cocaine followed, with 15 percent on average saying they used it in the past year and 9 percent in the past month. Relatively few said they used methamphetamine: 6 percent in the past year and 3 percent in the past month. The new questions about alcohol use focus specifically on heavy drinking, whose link to various behavioral problems, including crime, has been documented by research.[13] Heavy drinking is defined here according to the National Household Survey on Drug Abuse (NHSDA) characterization of "binge" drinking.[14] Because the NHSDA covers the general population, it includes arrestees not currently incarcerated. However, the many arrestees who do not have fixed addresses and are missed by NHSDA are included in ADAM. In this way the new ADAM permits researchers to compare heavy alcohol use by arrestees with that of the general population.[15] By the NHSDA definition, one-third of the women arrestees, on average, engaged in binge drinking in the month before their arrest. The range was 17 percent (Houston, where 10 women were identified as engaging in binge drinking) to 60 percent (Anchorage, where the figure was 81). (See Appendix Table 5-9.) The average for these arrestees surpasses the figure for binge drinking by the general population, which was just over one-fifth in 2000.[16] Drug dependence and treatment needs Women arrestees' need for treatment was measured not simply by their own self-reports of heavy use of drugs and alcohol, but also by whether they were considered at risk for dependence. (The way risk for dependence was identified is described in Chapter 2.) On this measure, more than 20 percent of the women arrestees, on average, were found at risk for dependence on alcohol. The range was 5 percent (Houston, with 3 women) to 45 percent (Anchorage, with 58 women). (See Appendix Table 5-10.) The proportions at risk for drug dependence were considerably higher. On average, 42 percent of the women arrestees were deemed at risk for drug dependence, with the range 21 percent (Laredo, where 13 women were at risk) to 53 percent (Chicago, where the number was 254). Injection drug use is another measure of severity of drug involvement and consequent need for treatment. In half the sites, 9 percent or more of the women said they had injected drugs in the year before their arrest. The range among the sites was from a low of no women (Laredo) to a high of 25 percent (Portland, with 56 women saying they had injected drugs). That a relatively high percentage of women arrestees need treatment for alcohol or drug use is of great concern, particularly from a public health perspective. Perhaps of equal concern, very few who said they need treatment had health insurance to cover it, and very few said they had received treatment. In half the sites, 56 percent or more of the women arrestees did not have health insurance. The range among the sites was 38 percent (Portland, with 90 lacking coverage) to 73 percent (Laredo, with 45 lacking coverage). (See Appendix Table 5-7.) Only a very small percentage of women said they had been treated for drug or alcohol use on either an outpatient or inpatient basis in the year before their arrest. The average among the sites was 11 percent, and the range was 1 percent (Omaha, with one woman receiving treatment) to 23 percent (Portland, with 53 women receiving treatment). The proportions who received inpatient or outpatient treatment were about the same, averaging 7 percent and 6 percent, respectively. For mental health treatment, the proportion was lower, averaging 3 percent of women arrestees. (See Appendix Table 5-11.) There appears to be no particular pattern among the sites that might help explain the likelihood of arrestees receiving one type of treatment rather than another. In Salt Lake City, for example, the percentage of women who were treated on an outpatient basis was double the percentage treated on an inpatient basis (14 percent, for 11 women; and 7 percent, for 6 women, respectively). In Dallas the opposite was the case, with the proportion receiving inpatient treatment higher. In 10 of the 29 sites the proportion of women arrestees who received outpatient treatment was higher than the proportion who received inpatient treatment. The proportions of women treated, by type of drug used, were also calculated. (See Appendix Tables 5-12a, 5-12b, and 5-12c.) Again, caution is advised in interpreting these data because of the small size of the samples and the fact that they are not probability-based. Participation in drug markets In most sites, the percentage of adult female arrestees who participated in the marijuana market was higher than the percentage participating in the market for the other drugs studied: crack cocaine, powder cocaine, heroin, or methamphetamine. In half the sites, almost one-third participated in the marijuana market, obtaining it by either cash, noncash, or a combination of cash and noncash means. Crack cocaine was the next most prevalent, with 23 percent or more in half the sites saying they participated in the market for this drug. For powder cocaine the median was 10 percent, for heroin 5 percent, and for methamphetamine 3 percent. (See Exhibit 5-2.)[17] For marijuana market participation the range among the sites was 18 percent of the adult female arrestees (Laredo, for 11 women) to 52 percent (Denver, for 120 women). For crack cocaine, the range was 6 percent (Salt Lake City, for 5 women) to 43 percent (Chicago, for 207 women) and for powder it was 4 percent (Los Angeles, for 6 women) to 28 percent (Tucson, for 37 women). For heroin, the proportions ranged from none (Houston and Birmingham) to 34 percent (Chicago, for 164 women). For methamphetamine, the percentage of self-reported market participants range from none (Albany, Birmingham, Laredo, and Philadelphia) to 41 percent (Honolulu, for 34 female arrestees). (See Appendix Table 5-13.) Paying for drugs Marijuana and crack cocaine were the drugs for which the market was most active, as measured by proportions of women participating (and for which absolute numbers of market participants were large enough for meaningful analysis).[18] The findings reveal that cash was not the sole way of paying for drugs. Among the sites, about half the women, on average, obtained marijuana by noncash means.[19] (See Table 5-2.) In 10 of the sites, more than 40 percent used this type of transaction. The exception was New York, where just 20 percent obtained marijuana by means other than cash. The proportions who obtained crack cocaine without paying cash were relatively low: only 18 percent of the women arrestees used noncash means to obtain crack; in 6 sites fewer than one in five women did so. Of these noncash marijuana transactions, the vast majority involved receiving the drug as a gift.[20] In half the sites, at least 85 percent of women arrestees who obtained marijuana by noncash means received it as a gift. In 10 of the 11 sites (with Denver the exception), more than 75 percent of the women arrestees who obtained the drug this way received it as a gift. (See Appendix Table 5- 14a.) For noncash crack cocaine transactions, gift-giving was somewhat less prevalent than for marijuana, although it was the dominant method of transaction. In more than half the sites, just under 60 percent of the women who obtained crack by noncash means received it as a gift. In only one site (San Diego) was the proportion greater than 75. (See Appendix Table 5-14b.) The second most common type of noncash crack transaction (aside from the category "other") was obtaining the drug on credit and paying cash later. In half the sites, 11 percent or more of the women arrestees who obtained crack without paying cash did so this way. A fairly large proportion of the female arrestees said that their noncash transactions in the crack market involved trading sex for the drug. In half the sites, 10 percent or more said they did so, with the range 3 percent (in Atlanta, with 1 woman saying she did so), to 21 percent (in New York, with 7 saying they did so). These figures contrast dramatically with the data for adult males, none of whom said they traded sex for any drug. Although the samples of males and females are not entirely comparable because of the relatively small number of women arrestees and the nonprobability basis of the sample, some other research supports this finding.[21] In contrast to noncash transactions, transactions involving cash only were proportionately smaller in the marijuana market, where in half the sites at least 16 percent of the women paid cash for this drug. In only 3 sites did the proportions who paid cash exceed one-third. The proportions who paid cash for crack were far higher: at least 39 percent in half the sites, and the percentages exceeded those for marijuana in each of the 11 sites. Some women arrestees used cash at times and at other times used something else to obtain drugs. The proportion who obtained crack by such combined means was at least 41 percent in half the sites, about the same as paid cash for this drug. The range of these combination transactions was 24 percent (New York, with 30 women) to 55 percent (Cleveland, with 99 women). The proportions who combined cash and noncash to obtain marijuana were lower than for crack: About one in four women arrestees obtained marijuana this way. (See Table 5-2.) The range was 19 percent (Denver, 23 women; and San Diego, 20 women) to 35 percent (Oklahoma City, 44 women). Does outdoor or indoor purchasing predominate? The proportion of women arrestees who obtained crack cocaine outdoors, in open air markets, was 44 percent or more in half the 11 sites studied.[22] The range was wide--from a low of 11 percent (in Oklahoma City, where 4 women purchased crack outdoors) to a high of 81 percent (in Chicago, where 132 did so). Participation in open air markets for crack cocaine was particularly evident in Chicago, New Orleans, and New York. Outdoor purchasing of marijuana was less common, with 30 percent or more obtaining it this way in half the sites. With the exception of Oklahoma City, in each of the 11 sites the proportions who obtained marijuana outdoors were lower than for crack. The range was 9 percent (in Las Vegas, where 4 women bought this drug outdoors) to 64 percent (in New Orleans, where the number was 25). (See Table 5-3.) Community advocates contend that outsiders (people who do not live in the neighborhood) come into the neighborhood to buy drugs, thereby promoting instability. To buy marijuana, 44 percent or more of the women arrestees in half the sites said they went outside their own neighborhood.[23] To buy crack cocaine, 40 percent or more in half the sites did so. (See Table 5-4.) Why do some attempts to buy drugs fail? Over the years, a considerable amount of law enforcement resources have been devoted to making it more difficult for drug users to obtain illicit drugs.[24] The findings for 2000 indicate that in attempting to buy crack cocaine, the majority of adult female arrestees did not have much difficulty.[25] In half the sites, 40 percent or less failed when they tried to buy this drug. (See Table 5-5.) The proportion who failed in attempting to buy marijuana was slightly lower (36 percent or less in half the sites). The reasons the arrestees' attempts to buy marijuana or crack failed were about the same for both drugs. For marijuana, the explanation noted by the highest percentage of women arrestees was that area dealers did not have the drug available to sell. In half the sites, 31 percent or more who tried and failed to buy marijuana said this was the reason. The reason noted by the second largest proportion (after the 31 percent who noted "other" reasons) was lack of dealers (24 percent or more women in half the sites). (See Appendix Table 5- 15a.) The reasons for crack cocaine transaction failures were similar. The explanation noted by the highest proportion of women arrestees was that no dealers were available (23 percent or more in half the sites said this), followed (after "other" reasons, cited by 24 percent) by lack of availability of the drug from dealers (cited by at least 26 percent of the women in half the sites). (See Appendix Table 5-15b.) As was the case with male arrestees, in nearly all 11 sites police activity was rarely identified by female arrestees as the reason a drug transaction failed. The proportions who said police activity was why marijuana transactions failed ranged from none (in Denver, Indianapolis, Las Vegas, and Phoenix) to 12 percent of arrestees (in New York and San Diego, where 5 and 2 women, respectively, said this was the reason). For crack, about 13 percent or less in half the sites noted police activity as the reason an attempted purchase failed-- twice the percentage who said police activity caused a marijuana transaction to fail. The range among the sites was none (in Indianapolis) to 29 percent (New York, where 16 women noted this reason). Comparison with adult male arrestees As noted above, these findings should be interpreted cautiously because the number of women arrestees is relatively small and because the samples were not drawn randomly nor were they probability-based. In many sites there are few women arrestees and overall, as in previous years, there were fewer women than men arrestees. Moreover, women selected for inclusion are likely to represent more serious offenses, as are women in general who are arrested. Thus, unlike the findings for men, which were based on probability sampling, the findings for women cannot be generalized to a larger population. In 2000, adult female arrestees tested positive for at least one of the NIDA-5 drugs almost as often as their male counterparts. On average, 63 percent of women tested positive, compared to 64 percent of the men. However, the drug for which female arrestees were most likely to test positive was cocaine; among male arrestees, marijuana was the most prevalent drug. For risk of drug dependence, the proportion of women was slightly higher than the proportion of men: in half the sites, 42 percent of the women were found at risk, compared to 37 percent of the men. The women were also more likely than the men to use alcohol heavily. In half the sites, one-third or more of the women drank heavily (had five or more drinks on the same occasion at least once in the month before they were arrested); 29 percent of the male arrestees did so. When it came to participation in drug markets, men outpaced women for marijuana, while the opposite was true for crack cocaine. In half the sites, just under one-third of the women (32 percent) participated in the market for marijuana, while 44 percent of the men did so. In the crack cocaine market, 23 percent of the women participated, in contrast to 15 percent of the men. Despite the caveats that must apply in interpreting the data, the findings offer some useful information about women arrestees, some of which confirm or are confirmed by previous research. They show that there are differences and similarities between male and female arrestees in their involvement in drugs and drug-related behavior. The value of the findings for women arrestees will increase when probability-based sampling is adopted for them. * Bruce G. Taylor, Ph.D., is Deputy Director of the ADAM program; Phyllis J. Newton, M.A., of Abt Associates Inc., was the contracting Project Manager; and Henry H. Brownstein, Ph.D., is Director of Drugs and Crime Research Division, National Institute of Justice, and Executive Director of the ADAM Program. Notes 1. Federal Bureau of Investigation, Crime in the United States, 2000: Uniform Crime Reports, Washington, DC: U.S. Government Printing Office, 2001: 216. 2. Greenfeld, L.A. and T.L. Snell, Women Offenders, Bureau of Justice Statistics Special Report, Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics, December 1999, NCJ 175688: 5. 3. For a review, see "Dynamics of the Drug-Crime Relationship," by H.R.White and D.M. Gorman, in The Nature of Crime and the Continuity of Change, Volume 1, Criminal Justice 2000, ed. G. LaFree, Washington, DC: U.S. Department of Justice, National Institute of Justice, 2000: 151-218, NCJ 182408. 4. For examples, see Prostitution and Drugs, by P.J. Goldstein, New York: Lexington Books, 1979; and "Drugs and Consensual Crimes: Drug Dealing and Prostitution," by D. Hunt, in Drugs and Crime, Volume 13, Crime and Justice: A Review of Research, eds. M. Tonry and J.Q. Wilson, Chicago: University of Chicago Press, 1990:159-202. 5. See Inciardi, J.A., D. Lockwood, and A.E. Pottieger, Women and Crack Cocaine, New York: MacMillan, 1993; and Rafter, N.H., ed., Encyclopedia of Women and Crime, Phoenix, AZ: Oryx Press, 2000. 6. For example, "Women Who Kill in Drug Market Situations," by H.H. Brownstein et al., in Justice Quarterly 12 (1995): 473-498; "Female Crack Sellers in New York City: Who They Are and What They Do," by E. Dunlap, B.D. Johnson, and L. Maher, in Women and Criminal Justice 8 (1997): 25-55; "Women and Drugs Revisited: Female Participation in the Cocaine Economy," by J. Fagan, in Journal of Drug Issues 24 (1994): 179-225; "Drugs and Consensual Crimes," by Hunt; Women and Crack Cocaine, by Inciardi, Lockwood, and Pottieger; "Women in the Street-Level Drug Economy: Continuity or Change?" by L. Maher and K. Daly, in Criminology 34 (1996): 465-491; and "Experiences of Women Who Sell Crack: Descriptive Data from the Detroit Crack Ethnography Project," by T. Mieczkowski, in Journal of Drug Issues 24 (1994): 227-248. 7. For a discussion of the difficulties in interpreting DUF results, see Methodology Guide for ADAM, by D. Hunt and W. Rhodes, Washington, DC: U.S. Department of Justice, National Institute of Justice, May 2001. It can be downloaded from the ADAM Web page (http://www.adam-nij.net) on the NIJ Web site (http://www.ojp.usdoj.gov/nij). 8. The ten drugs for which arrestees are tested in the ADAM program are cocaine, opiates, marijuana, methamphetamine, phencyclidine (PCP), methadone, benzodiazepines, methaqualone, propoxyphene, and barbiturates. The first five are the "NIDA-5," established as a standard panel of commonly used illegal drugs by the National Institute on Drug Abuse. 9. See note 8 for the ten drugs for which ADAM uses urinalysis. Compared to the percentages testing positive for cocaine, opiates, marijuana, and methamphetamine, the percentages testing positive for the other six drugs were low. 10. Unless otherwise indicated, averages represent medians. 11. Urinalysis can detect drugs in the amphetamine group, but only a confirmatory test indicates whether the drug is methamphetamine. The confirmation is also necessary because several cold and diet medications contain amphetamines, which would produce false positives. 12. "Month" and "30 days" are used interchangeably, as are "year" and "12 months." 13. For a discussion, see Chapter 3. 14. A single episode of binge drinking is defined by the NHSDA as consuming five or more drinks on the same occasion on one day in the past 30. Heavy drinking is defined more fully in Chapter 3, which presents the findings for adult males and includes a discussion of the behavioral problems associated with alcohol abuse. 15. The ability of the new ADAM to crosswalk data with NHSDA is discussed in Chapter 8. 16. Office of Applied Statistics, Summary of Findings from the 2000 National Household Survey on Drug Abuse, Rockville, MD: U.S. Department of Health and Human Services, Substance Abuse and Mental Health Services Administration, September 2001: 173. The figure represents people 12 years of age and older. 17. Because the sample of adult female arrestees is much smaller than that of adult males, and because probability-based sampling was not used, the findings are not presented in as much detail. 18. In very few sites were there even as few as 50 female arrestees who said they had participated in the markets for the various drugs. In only one site (Phoenix) were there more than 50 powder cocaine market participants; in only two (Chicago and New York) were there more than 50 heroin market participants; in only three (San Diego, Phoenix, and Las Vegas) were there more than 50 methamphetamine market participants. A cutoff point of 50 cases in the marijuana and cocaine markets was set for all analyses that follow in this chapter. A threshold of 50 is by no means definitive, but appears to be reasonable because it generated at least 10 cases per cell even when variables with as many as five categories were analyzed. The meaningfulness of a percentage is almost certainly lost when cell sizes start to fall below 10. The sites where 50 or more adult female arrestees said they had participated in the marijuana and crack cocaine markets were Atlanta, Chicago, Cleveland, Denver, Indianapolis, Las Vegas, New Orleans, New York, Oklahoma City, Phoenix, and San Diego. (See Appendix Table 5-13.) 19. For an explanation of the categories noncash and combination (cash and noncash) transactions, see Chapter 3. 20. Examples of gifts include getting or sharing a marijuana joint at a party or sharing crack with a partner. 21. The research revealing women's trading sex for drugs is based on anecdotal evidence or small samples. See Baskin, D.R. and I.B. Sommer, Casualties of Community Disorder: Women's Careers in Violent Crime, Boulder, CO: Westview, 1998; Inciardi, Lockwood, and Pottieger, Women and Crack Cocaine; and Maher and Daly, "Women in the Street-Level Drug Economy." 22. See Chapter 4 for a discussion of issues involved in purchasing drugs outdoors. 23. Because the question was," Did you buy it [name of drug] in the neighborhood where you live or outside your neighborhood?" the definition of "neighborhood" reflected the arrestees' perceptions. 24. See Office of National Drug Control Policy, The National Drug Control Strategy: 2001 Annual Report, Washington, DC: Executive Office of the President, 2001, NCJ 185400. 25. Arrestees were asked, "Was there a time in the past 30 days when you tried to buy [name of drug] and had the cash, but you did not buy any?" Those who did not buy were asked why. ---------------------------- Sample Size--Issues Of the 38 ADAM sites, data for adult female arrestees were available from 35. They are the sites where data were collected in at least one quarter of calendar year 2000.* (See Appendix Table 5-1.) About the same number of female arrestees were selected for the sample in each calendar quarter, with the average close to 50 in each site each quarter. Not all women selected for inclusion in the sample could be interviewed. For example, in Albuquerque, of the 164 women selected, only 112 were interviewed. The 52 not interviewed were either not available, not asked (for a variety of reasons), or declined. Operational issues at the sites made it impossible to report a true response rate; that is, the percentage of the selected sample for which all data were available. However, the vast majority of female arrestees who were asked agreed to be interviewed. On average, only 17 percent refused. The refusal rate ranged from a low of none (Charlotte-Metro Area) to 39 percent (Chicago). Of the women interviewed, most agreed to also give a urine specimen. In half the sites, 92 percent or more did so, with the range among the sites 63 percent (Detroit) to 100 percent (Charlotte, Dallas, and San Jose). At more than half the sites, interviews were conducted with 100 or more adult female arrestees, with the range one interview (Charlotte) to 510 (Chicago). But at some sites, there were too few to make reliable judgments about the distribution of the adult female data (for example, how many tested positive for a given drug). To avoid presenting findings that might be misleading because the numbers were small, the analyses were based on data from sites where at least 50 women were interviewed. A cutoff point of 50 generated about 10 cases per cell, even for questions with as many as five categories of responses. Of the ADAM sites, there were 29 in which 50 or more women were interviewed. (Although Des Moines had 49, it was included.) The sites where findings on women arrestees were not reported are Charlotte, Miami, Minneapolis, Sacramento, St. Louis, San Antonio, Seattle, Spokane, and Washington, DC. * In 19 sites data were collected in all four quarters, in 10 sites data were collected in three quarters, in 4 sites data were collected in two quarters, and in 2 sites data were collected in only one quarter. ---------------------------- VI. Drug Use Among Juvenile Detainees by Diana C. Noone* A large body of research has demonstrated that substance use by young people may lead to physical and social problems, including declining school grades, truancy, accidental injuries, risk of contracting HIV and other sexually transmitted diseases, alcohol-related traffic incidents, depression, family dysfunction, and suicide.[1] Young people's use of drugs, as well as alcohol and tobacco, is measured by at least two major surveys. One is the annual Monitoring the Future study.[2] Another is the National Household Survey on Drug Abuse.[3] These surveys look at young people in general, but there are few studies of drug use by young people in the juvenile justice system. ADAM and its predecessor, the Drug Use Forecasting (DUF) program are alone in annually measuring substance abuse by juvenile detainees. Probability-based sampling, introduced by ADAM for adult male arrestees, has not yet been introduced for juvenile detainees. Nor has the interview instrument used with juveniles been expanded to include more issues related to drug use and related behavior. And in relatively few sites do juvenile detainees participate in the ADAM program. (For a discussion of sample size and related issues, see "How Drug Use by Juvenile Detainees Is Measured." A list of the sites is in Table 6-1.) ADAM anticipates that once the sampling method is strengthened, the interview instrument expanded, and the number of sites increased, the findings on juvenile detainees will be a more useful source of information for local policymakers. Given their current limitations, they are presented as an informational tool only. Findings--juvenile males As revealed by urinalysis, the patterns of use of specific drugs by juvenile male detainees were similar in all nine sites where data were collected. Of the NIDA-5 drugs (marijuana, cocaine, opiates, methamphetamine, and PCP),[4] marijuana was the one most commonly used. In all nine sites, at least 41 percent tested positive for marijuana, with the range 42 percent (22 detainees, in Birmingham) to 55 percent (251 detainees, in Phoenix). (See Table 6-2.) The findings are consistent with those of recent years, when marijuana was also the drug most commonly detected among juveniles in the ADAM sample. They are also consistent with the findings of the National Household Survey on Drug Abuse, which revealed marijuana as the drug of choice among young people.[5] Cocaine (undistinguished between crack and powder) came in a distant second in 2000, except in two sites, San Diego and Portland, where (measured by absolute numbers rather than percentages) methamphetamine was more widely used than cocaine. In only three sites (Denver, Tucson, and Phoenix) did more than 10 percent of the juvenile male detainees test positive for cocaine. The rates of methamphetamine use ranged from none (Birmingham, Cleveland, San Antonio, and Tucson) to 8 percent (in San Diego). The percentages testing positive for the other two NIDA-5 drugs (opiates and PCP) were extremely low in all sites. Demographics and sociodemographics Juvenile detainees interviewed by ADAM range in age from about 12 to 18. In 2000, the largest proportion was between 15 and 17. Among those who tested positive for use of any drug, the largest group was age 17. In half the sites, 70 percent or more of the juvenile detainees[6] said they were still in school, with the range 55 percent (138 detainees, in Phoenix) to 93 percent (98 detainees, in San Antonio). On average, less than 5 percent of the juvenile male detainees lived either on the street, or in a shelter, a drug treatment facility, a halfway house, or in prison in the month before they were detained.[7] The vast majority (in half the sites, 93 percent or more) lived in houses or apartments, including public housing. Just over half the detainees (52 percent) lived in two-parent households, while 40 percent lived in single-parent households in the month before they were detained. Of those in single-parent households, 82 percent lived with their mothers and 10 percent with their fathers. Among those who tested positive for any drug, the breakdown by household type was similar to that among all the ADAM juvenile male detainees: 54 percent, on average, living in two-parent households and 40 percent in single-parent households. The proportions of drug-positive juvenile male detainees who lived in single-parent households with their mothers ranged from 42 percent (65 detainees, in Los Angeles) to 59 percent (54 detainees, in Denver). Types of offenses Depending on the law of a particular State and the offense allegedly committed, a youth may or may not be considered a juvenile and may or may not be transferred out of the juvenile system. These differences undoubtedly affect the kinds of offenses for which the juvenile detainees interviewed by ADAM, whether male or female, are charged. FBI data show that juvenile offending has been declining overall in recent years, registering a 15 percent drop between 1996 and 2000.[8] They also show that in three categories of offenses, juvenile detentions increased in about that same period: driving under the influence (36 percent increase), liquor law violations (31 percent increase), and curfew violations (9 percent increase).[9] The charge faced by the largest percentage of juvenile males who participated in ADAM in 2000 was an unspecified technical violation of a condition of release from the justice system (for example, violating a condition of probation or an order imposed by a drug court). In five of the nine sites (with Cleveland, Denver, Portland, and San Antonio the exceptions), at least 18 percent of the juvenile males had been detained on this charge. For juvenile detainees overall, the most common offense in 2000 was larceny-theft.[10] Because the FBI does not include technical violations in its count of juvenile offenses, it is not possible to determine how common this type of offense is among juveniles in general. But the fact that technical violations are the largest offense category among ADAM male juveniles suggests that many of these young people had previous contact with the juvenile justice system. The next largest category of offense was drug possession (although in Cleveland and San Antonio this was the category for which the percentage detained was the highest). Exceptions were Denver and Portland, where the next largest category was being detained on a warrant, or because of flight or escape. These types of charges also strongly suggest previous contact with the juvenile justice system. The vast majority of juvenile male detainees (81 percent) said they were not under the influence of drugs or alcohol when they were detained. The same was true even of the juvenile males who tested positive for any drug: In half the sites, 75 percent or more said they were not under the influence, with at least 66 percent in all nine sites saying this. Findings--juvenile females Because the samples of juvenile female detainees were small, totaling 423 in all, the findings should be viewed as illustrative only. For juvenile females as for juvenile males, urinalysis indicated marijuana as the leading drug among the five tested. The range among the sites was 17 percent (3 juvenile females, in Birmingham) to 58 percent (15 juvenile females, in Denver). (See Table 6-3.) In four of the eight sites for which data were collected on juvenile females, the second most commonly used drug (as measured by absolute numbers) was cocaine (Denver, Phoenix, San Antonio, and Tucson), yet in all eight sites the proportion testing positive for this substance was less than 20 percent. In Los Angeles, Portland, and San Diego methamphetamine was the second most commonly used drug, as measured by absolute numbers who tested positive. The pattern of relatively heavy use in the West mirrors that for adults. Demographics and sociodemographics In age, the juvenile females who tested positive for any NIDA-5 drug fell primarily into the category 15 to 17 years old. Of those testing positive, the largest proportion--just under one-fourth--was 17 years of age. In every site except Phoenix, at least half the juvenile females said they were still in school. Phoenix was at the low end of the range, with 38 percent (20 juvenile females) in school, and Tucson at the high end, with 91 percent (10 juvenile females). As with the juvenile male detainees, the vast majority of juvenile females (88 percent or more in half the sites) lived in houses and apartments, including public housing rather than in a shelter, treatment facility, or similar arrangement. Among those who tested positive for any drug, the percentage living primarily in houses or apartments was slightly lower (at least 84 percent in half the sites). The breakdown of traditional vs. single-parent households was about the same as for juvenile males. In the month before they were detained, just over half (51 percent) of the female juvenile detainees, on average, lived in a two-parent household, while just over one-third (36 percent) lived in a single-parent household. Among the sites, the proportions of juvenile female detainees testing positive for any drug who lived in single-parent households with their mothers ranged from 33 percent (1 juvenile female, in Birmingham) to 59 percent (13 juvenile females, in Denver). Types of offenses Of the juvenile female detainees who tested positive for any drug, few said they were under the influence of drugs or alcohol when they were detained. The range among the eight sites was 22 percent (4 juvenile females, in Los Angeles) to 42 percent (5, in Tucson). The same as for the males, the most frequent charge among the juvenile females was an unspecified technical violation of a condition of release from the justice system. Los Angeles was the site where the proportion facing unspecified technical violations was highest (47 percent, representing 22 juvenile females). * Diane C. Noone, Ph.D., J.D., is a Social Science Analyst with NIJ. Notes 1. Dembo, R.L., et al., "The Relationships Among Family Problems, Friends' Troubled Behavior, and High Risk Youths' Alcohol/Other Drug Use and Delinquent Behavior," The International Journal of Addictions 29 (1994): 419- 442; Fendrich, M., et al., "Substance Involvement Among Juvenile Murderers: Comparisons with Older Offenders Based on Interviews with Prison Inmates," The International Journal of Addictions, 30 (1995): 1363-1382; Huizinga, D., R. Koeber, and T.P. Thornberry, Urban Delinquency and Substance Abuse-- Initial Findings: Research Summary, Washington, DC: U.S. Department of Justice, Office of Juvenile Justice and Delinquency Prevention, 1994, NCJ 143454; Inciardi, J.A., R. Horowitz, and A.E. Pottieger, Street Kids, Street Drugs, Street Crime: An Examination of Drug Use and Serious Delinquency in Miami, Belmont, CA: Wadsworth, 1993; and Office of Juvenile Justice and Delinquency Prevention, Capacity Building for Juvenile Substance Abuse Treatment, Washington, DC: U.S. Department of Justice, 1997, NCJ 167251. 2. Monitoring the Future Study, Ann Arbor, MI: Institute for Social Research, University of Michigan, 2000. This study, which annually ----measures young people's use of alcohol, tobacco, and other drugs, is sponsored in part by the National Institute on Drug Abuse. 3. The National Household Survey on Drug Abuse estimates illicit drug use among people age 12 and older. See Packer, L., et al., Summary of Findings from the 1999 National Household Survey on Drug Abuse, Rockville, MD: U.S. Department of Health and Human Services, Substance Abuse and Mental Health Services Administration, 2000. 4. The "NIDA-5" is a list of drugs established by the National Institute on Drug Abuse as a standard panel of commonly used illegal drugs. 5. Packer, L., et al., Summary of Findings from the 1999 National Household Survey on Drug Abuse. The NHSDA figure reflects drug use by people ages 12 through 25. 6. Unless indicated otherwise, averages are expressed as medians. 7. "Month" and "30 days" are used interchangeably, as are "year" and "12 months." 8. The decline is measured in numbers of arrests of young people under age 18. Federal Bureau of Investigation, Crime in the United States 2000, Washington, DC: U.S. Department of Justice, 2001: 222. 9. Cited in Office of Juvenile Justice and Delinquency Prevention, Annual Report 2000 (Washington, DC, 2001), NCJ 188419. 10. Federal Bureau of Investigation, Crime in the United States 2000: 222. ---------------------------- How Drug Use by Juvenile Detainees Is Measured As with adult arrestees, drug use and related behavior among juvenile detainees are measured by means of a questionnaire and urinalysis. And as with adult arrestees, participation is both voluntary and anonymous. The ADAM data and resultant findings for juveniles have certain limitations arising from the sampling procedure and the interview scheduling. Because of these limitations, the data and findings should be interpreted cautiously. Obtaining the data Juvenile arrestees are interviewed after the study is explained to them and their consent is obtained. The survey instrument used in the interview consists of 28 questions that explore a variety of issues, including demographics, living arrangements, and educational status. The same questionnaire has been used for adults and juveniles since the program began as DUF in 1988. The expanded ADAM questionnaire, used with adults, both men and women, has not yet been adopted for juvenile detainees. Following the interview, juvenile detainees are asked if they are willing to provide a urine sample for testing. All urine specimens are screened for up to10 drugs: cocaine, marijuana, opiates, amphetamine, PCP (phencyclidine), barbiturates, benzodiazepines, methadone, methaqualone, and propoxyphene.* If amphetamine is detected, a confirmation test is conducted to determine if it is methamphetamine. In this analysis, only use of the "NIDA-5" drugs is examined. These drugs--marijuana, cocaine, opiates, methamphetamine, and PCP--were established by the National Institute on Drug Abuse as a standard panel of commonly used illegal drugs. Sample sizes and limitations In contrast to the samples of adult male arrestees, the samples of juveniles are not probability-based. They are drawn as convenience samples, and for this reason it is not possible to place confidence limits around the data. This means in turn it is not possible to estimate the statistical significance of the findings. The schedule for interviewing juveniles may introduce a certain amount of bias into the sample. Juveniles included in the ADAM sample are only those who are available during the times when the interviewers are working--primarily weekends and evenings. An additional bias may be introduced when facilities release rather than hold the juveniles detained on less serious charges. These released juveniles are then not available to be interviewed, possibly increasing the representation in ADAM of juveniles detained (and held) for more serious offenses. The findings suggest that many detainees interviewed by ADAM had previously been involved in the juvenile justice system. The interviews with the male juveniles were conducted in nine detention centers (in Birmingham, Denver, Cleveland, Los Angeles, Phoenix, Portland, San Antonio, San Diego, and Tucson) and with female juvenile detainees in eight detention centers (in Birmingham, Denver, Los Angeles, Phoenix, Portland, San Antonio, San Diego, and Tucson). For juvenile males, the sample sizes ranged from 421 (Phoenix) to 53 (Birmingham). For females the range was 114 (Phoenix) to 18 (Birmingham). In all, 2,106 juvenile males were interviewed and gave a urine sample. For juvenile females the total was 423. Because the interviews are conducted in few sites, this further limits the ability to generalize about the findings; that is, they should not be interpreted as representing all youthful detainees nationwide. * The program uses the EMIT (Enzyme Multiplied Immunoassay Testing) system to screen for drugs in the urine. ---------------------------- Part II: 2000 Findings, by Site--Adult Male Arrestees ---------------------------- Part III: Applying The New ADAM Method VII. Implementing the New ADAM Study Design at the Local Level by Phyllis J. Newton and Margaret E. Townsend* When the National Institute of Justice (NIJ) decided to strengthen the Drug Use Forecasting (DUF) program, it envisioned a way to measure drug use and drug-related behavior among arrestees that could withstand methodological scrutiny and would be an even better tool for local policymakers than in the past. That meant developing a statistically sound method of data collection, improving the way the local sites collected data, enhancing the survey instrument (questionnaire), and increasing the number of sites. The premise underlying the change was straightforward: to build an infrastructure that ensured standard data collection protocols; an unbiased, probability-based sample of arrestees; and a data management system that generated standardized data for use by the sites. The theoretical ideals underpinning the new program, ADAM, have now been applied in the practical world of the jail environment in 35 sites nationwide, and the program has had one full year of experience administering a new collection instrument and probability-based sampling in all the sites. That application of research in a real-world setting raised several questions, which are explored here. Among the questions are whether it is possible to develop data collection protocols applicable in all jurisdictions, whether the ADAM program can ensure adherence to these standardized protocols, whether methodologically sound sampling strategies can be imported into local jails and still retain their scientific rigor, what adaptations the methodology can tolerate before it no longer meets ADAM's standards, and whether probability-based sampling can work in a jail environment. Explaining the full the transition from DUF to ADAM requires discussing why standardization is important; the reasons for probability-based sampling; how the new, county-level and facility (jail)-level sampling designs were implemented; and what challenges are posed by the jail environment. Once the transition is complete and all sites operate with probability-based sampling for female arrestees as well as male arrestees, ADAM should have even greater potential for generating information that will assist local officials at the sites in making policy decisions affecting these at-risk populations. From DUF to ADAM Until ADAM was established, researchers had never before attempted to use the setting of the jail as the focus of ongoing, standardized data collection and application of rigorous sampling procedures. NIJ did so, creating a program at multiple sites nationwide that included the following components: o Data collection procedures common to all sites o Probability-based sampling that would allow the sites to place confidence intervals around their findings on drug abuse and related behavior o Enhanced data collection capabilities, with questions about drug treatment and drug markets o The ability to compare ADAM data to those from national surveys of drug use.[1] The importance of standardization, or common data collection procedures, cannot be overstated. Regardless of the sophistication of the sampling process, the findings are not reliable unless procedures are the same and carried out the same way in all the sites. All interviewers must administer the survey questionnaire (instrument) in the same way in all the sites. No step in the data collection process can be omitted. The population from which the data are drawn must be identifiable and the same type in all sites. Sampling must always be conducted the same way. The data collection instrument, or questionnaire, was expanded from its initial focus on drug use, and now covers treatment and the dynamics of drug markets. And with the greater methodological sophistication of probability-based sampling, the interview also needed to include information required for weighting cases so that they are generalizable to a larger population--the county. Thus, ADAM needed a mechanism for estimating such information as the number of arrestees in the county who used drugs and the number who needed treatment. And because ADAM data are a measure of drug use among a limited spectrum of Americans, NIJ decided to build into the new collection instrument the ability to compare findings to those of other, national surveys of drug use. Achieving standardization Creating standardized data collection procedures involved first making sure the "catchment areas," or regions from which arrestees are drawn at the sites, are defined the same way in all the sites. Because a representative mix of the types of offenses committed is needed for the sample, standardization also meant resolving varying ways in which the sites define and deal with crime.[2] Redefining the catchment areas Arrestees who participate in ADAM are selected from people brought to booking facilities--generally a jail. In DUF, the sites collected data from arrestees booked at one or two jails, but they did not necessarily reflect all arrestees in the community. For example, in Philadelphia data were collected in one facility, but because there are seven, the data could not represent drug use among all people arrested citywide. In Los Angeles, data were collected at two booking facilities out of nearly 100 in the county. At the other extreme, in New Orleans Parish (equivalent to a county), the sheriff's department operates the sole holding facility, so it is possible to make assertions about the generalizability of the data to the county's arrestee population. In fast-growing cities in the West, the number of facilities grew with the population, but data continued to be collected in only one. DUF began operations in 1988. Phoenix, whose population increased 40 percent in the past 10 years,[3] continued to collect data at one facility. In sprawling communities in Texas, California, and elsewhere in Arizona, as metropolitan areas grew and encompassed more and more localities, it became less clear which locality or localities the arrestees at the central jail represented. ADAM defined the common catchment area as the county. County lines generally served as a reasonably common demarcation, though there were exceptions. Atlanta, for example, extends across two counties, but because the site felt the city should be covered as a single entity, ADAM included both Fulton and DeKalb counties in the definition of this catchment area. Because the city and county of Philadelphia are coterminous, the city limits define this catchment area. New York City consists of five boroughs, making it necessary to attempt data collection in all of them.[4] Not all counties are the same. Sites in the West manage catchment areas that are considerably larger, geographically, than in the East, with local law enforcement practices, such as deployment of officers, sometimes contingent on the number of miles to cover or amount of time it takes to bring arrestees to holding facilities. The catchment areas also vary in the number of local, county, and State law enforcement officials having arrest authority, with the result that procedures might differ by facility. There were also variations in the number of booking facilities capable of holding arrestees. The ADAM program aimed to ensure that each booked arrestee in a catchment area had some probability of being selected for participation, and these two variables affected that probability. In all sites, the first step for ADAM was to find out how many booking facilities there were in the defined area and where arresting officials took arrestees to be booked. Variations in "arrest" and other terms Every State and local jurisdiction in the country has its own laws and system of justice, and while there are commonalities, the system of legal requirements in each reflects local conditions. For example, what one jurisdiction calls "breaking and entering" another may call "burglary." What one community may call an arrest another may call a citation; for some communities booking and arrest are synonymous. Cite and release in one community may mean a street release/field release and in another may mean coming to the station/facility for booking and release. In order to understand which arrestees constituted the sample in each site, ADAM needed common definitions so that segments of the arrestee population were neither over nor underrepresented. The goals of ADAM dictated that all booked arrestees--from low-level misdemeanants to serious felons--have some probability of being selected for inclusion in the sample interviewed. If arresting authorities in one community bring all arrestees to booking facilities, the result for ADAM's purposes would be a reasonable mix of types of offenses among the arrestees. However, if another community cites and then releases most arrestees who commit only misdemeanor and city ordinance offenses, the arrestee population that remained to be interviewed would overrepresent those booked for more serious offenses. Even "misdemeanor" and "felony" are not defined the same by all jurisdictions. In some, such a label refers solely to the potential length of a jail sentence, while in others it refers more generally to the seriousness of the offense. Adopting probability-based sampling The ADAM procedure was also redesigned to account for the variations in the structure and size of local criminal justice systems and processes. That involved designing county-level sampling models and strategies for sampling arrestees in each selected facility. The goal of sampling in ADAM is to be able to estimate with a known probability, the likelihood that an arrestee will be selected for the ADAM interview and to be able to use that information to weight the sample data.[5] To avoid biasing the sample in favor of the types of arrest charges and types of arrestees represented at the facility during the times when interviewers are not collecting data, the sampling plan must reflect all days of the week and all times of the day. Further, the plans must represent all of the facilities in the county, whether large or small, or urban, suburban, or rural, in order to represent all types of offenders and offenses. The "stock" and the "flow" For each adult male arrestee booked in a 24-hour period,[6] ADAM had to determine the probability of his being selected for the sample. The approach had to be one that could be readily implemented in 35 sites in which there were in all more than 100 booking facilities. The easiest way--sampling and interviewing around the clock--would not be cost-effective. Another option would be to choose blocks of time randomly during the day and sample arrestees booked during those time periods. This was also too costly. The sampling plan adopted covered the full 24-hour period by splitting the day into two parts: a "flow" and a "stock" period. In the flow period were the arrestees booked during each daily ADAM data collection shift--the 8-hour period when the flow of arrestees was highest. In the stock period were arrestees booked during the 16 hours when data collection did not occur. Using systematic selection procedures, interviewers sampled from both groups, with the result an arrestee sample that represented each 24-hour period of the one- to two-week ADAM data collection period. The four sampling models Variations among the catchment areas required developing a county-level sampling model that was flexible enough to be applied to the specific counties/sites. This in turn required that no matter the number of booking facilities in a county, each county would have a sampling plan that generated estimates that could be extrapolated to the entire arrestee population of the county. For counties with only a single booking facility, this was easily accomplished, but for those with multiple facilities, ADAM developed a procedure that would not bias the sample against certain facilities or certain types of offenses. To accommodate the variations, four sampling models were developed. The simplest plan was the "single jail" design, in which collection took place at one jail only. Only slightly more complex was the "stratified" design, used for counties/sites with six or fewer facilities. This design called for data collection at all jails, with target numbers of arrestees selected proportionate to the number of arrestees processed at each jail. For counties with more than six jails, a "stratified cluster" sample design was developed. In this type of design, every jail was assigned to one of a small number of strata, with one or two jails sampled from each stratum. This design generated estimates for all jails, even though only some jails were included in the sample. In a few sites, this model needed to be further refined into a "feeder" design. It was applied in counties where a large number of jails quickly transported arrestees to a central holding facility, which for ADAM reduced the probability of interviewing arrestees in the outlying jails to virtually zero. In counties where the feeder design was used, interviews took place at the central holding facility (which represented all jails in the county) as well as at the "feeder" jails, so that arrestees who were not transported to the central facility could be sampled. To determine which model was appropriate where, the ADAM sites were asked to identify every booking facility in the county and furnish information about the number of adult males booked at each in the past year. These two pieces of information were used to select facilities and facility sample target numbers, proportionate to size. In hindsight, it would have been helpful to have worked with each site to document the movement of arrestees through the various booking facilities in the county before developing county-level sampling plans. Had that been done, it would have been possible to adjust sampling plans to accommodate the county-specific variations, and thus implementing the plan would have been easier. However, that would have delayed implementation and would not have guaranteed immunity from other idiosyncracies. Weighting the data to ensure representativeness In order to demonstrate the extent to which numbers in a sample represent a larger population, they are weighted. In ADAM, this requires identifying the probability of each adult male arrestee's being included in the sample, based on information about that larger population. The ADAM data required a nontraditional approach to weighting. The main reason is that there is no way to know, before sampling takes place, who will be arrested and thus no way to assign a probability of any given arrestee's being included in the sampling frame (the list of cases of interest). The issue was resolved by using post-sampling stratification to identify the probability of inclusion in the sample of like groups of arrestees. This required obtaining information about the total population arrested during each ADAM data collection period in each site (the "census"). In ADAM, all sites provided, as the census data, information about all adult males booked into the facility during the one- to two-week ADAM data collection period. These data include each arrestee's date of birth, race/ethnicity, date and time of booking, and description of the arrest charge and its severity. They are used to match the sample of arrestees interviewed with the larger population and to separate the population into strata based on these characteristics. The number of arrestees in each stratum was then counted, and the probability of their being included in the sample was determined in comparison to that of everyone else in that stratum. Essential to this approach was ensuring that bias could be minimized; that is, that for everyone in each stratum the probability of selection was the same. Using "census" data Several assumptions were used in weighting the data: o Arrestees charged with more serious crimes spend more time in the jail facility than those arrested on less serious charges. o Arrestees booked at the same time of day are processed similarly; that is, they all spend approximately the same amount of time in the jail before arraignment and/or transfer to another holding facility. o The stock and flow model (described above) may mean more serious offenders will be over-represented in the stock population, while the flow sample should represent the full range of charges. Thus, an assumption was made that for all people arrested on like charges and booked at about the same time (stock or flow), the probability of selection for the sample is similar. o Because the number of interviewers is static, the day of the week affects the probability of an arrestee's being selected. For arrestees booked on days when many people enter the system, the probability of their being selected is lower than for those booked on slow days. In stratifying the population, time of booking is used to separate stock and flow, and charge severity and day of the week are used to further separate the population into strata having approximately equal probabilities of selection. For example, all felony "stock" booked on Friday and Saturday may be in one stratum, while misdemeanants booked Monday through Wednesday are in another. The importance of census data to the weighting process makes accuracy and comprehensiveness of those data essential. Some sites find it more difficult than others to meet these requirements. Some facilities either do not store booking information electronically, or do not allow the jail database to be queried at the user end. For these sites, census data must be collected manually. Thus, on each day of ADAM data collection, in addition to orchestrating the sample selection and the interviewing, the ADAM site staff keeps a running census of all adult male arrestees booked into the jail. They then submit these data to the ADAM Data Center at the end of the one- or two-week collection period. Although this is typically the procedure in small or "low-flow" facilities, some jails where case flow is very high are not automated and require this manual approach. For sites with only one facility or for those that use a stratified sampling design (because they have six or fewer facilities), all facilities are sampled, because there are so few. However, sites that use stratified cluster design (because they have more than six facilities) and feeder designs must submit census data for all booking facilities in the county. For a few sites, where the county maintains a countywide criminal database, this is not an arduous task. For other sites, collecting census data from all facilities in the county is difficult and the cost prohibitive. When it is not possible for them to collect census data, annual booking statistics for each facility or countywide booking data are used to develop annualized countywide estimates. Resolving census problems Three problems that can adversely affect the weighting process often arise after the census data are transmitted to the Data Center. The data may include ineligibles, duplicates, and inconsistently recorded booking times. Ineligibles and duplicates. Not everyone booked into jail is eligible to participate in ADAM. (Those ineligible include Federal holds, extradition holds, and court holds, all of which involve people detained in a local facility before trial). ADAM has to ensure that ineligibles are identified and removed from the sample so that they do not inflate the number of arrestees in the county. For example, an offender released on bond who arrives at court for arraignment may be remanded into custody and booked into the jail for holding pending transfer to another facility. This person would not be eligible for ADAM, but in most facilities would appear in the booking system. Some facilities track the type of booking and can include it as a variable in submitting their census data. In other cases, eligibility for the ADAM sample can be deciphered only by reviewing the charge and the information from the arresting agency and making educated guesses. It is the sites that confirm that ineligible populations are excluded from the census submission (and that there are no missing cases). Duplicate cases also inflate the arrestee population. Cases are duplicated when more charges are added to a previous booking of an arrestee or an arrestee's use of an alias is discovered. The sites will need to learn to recognize the potential for duplicates and work with the ADAM Data Center to merge or delete duplicate records. Inconsistent booking times. There can be variation in and confusion about the definition of "booking time" at the local level. This is a variable essential to the weighting process because in order to assign cases to the correct stratum, the time of the day when arrestees are selected for data collection must match the time the census data indicate as booking time. However, in some cases, data submitted to the ADAM Data Center indicate booking times different from the times when the ADAM sample is selected. For example, many sites use the jail's intake log to identify stock and flow and select their sample by using intake time as a proxy for booking time. In the census data, the booking time indicated often reflects the time at which booking data were entered into the facility's computer system, rather than the intake time. This poses a problem for weighting because in some cases data are not entered into the booking system for several hours after intake, which means some stock might be weighted as flow and some flow as stock. In such situations, the sites will work with the Data Center to find a solution to the discrepancy. The county-level sampling plan design In adopting probability-based sampling for ADAM, there are two stages in planning the sampling at the sites: the broad, county-level stage that determines a site's general sampling design, and a more specific, facility-level stage that specifies the actual mechanics of drawing the sample. To determine which of the four sampling strategies/models is to be used in a site, it was first necessary to identify the number of booking facilities in each site's catchment area. Once this number was known, each site was "labeled" with the type of county-level sampling plan to be used in that catchment area (single-jail design, stratified design, stratified cluster design, or feeder design). Implementing the county-level sampling design involved setting target numbers of arrestees to be included in the sample; understanding who, on the basis of various characteristics, the population of arrestees in county facilities represent; and ensuring fiscal accountability. Each ADAM site needed its own unique sampling plan, based on the number of people arrested and booked in the county and on the procedure each arresting agency follows after citing or arresting suspected offenders. And although the plans had to be unique, adherence to standardization required that variations among the sites did not compromise the overall sampling principles. Setting target numbers for the sample The basis of the sample size was the total number of bookings in the county and the number of facilities in the county, because this permits calculating a sample equivalent to the variance that results from sampling proportionate to size. In order to perform this calculation, the sites needed to provide information about the total number of arrestees booked in each facility in the catchment areas, or at least the total number of people booked in the county. However, this information was not always available to the site staff and, in sites where booking data were not available, the FBI's Uniform Crime Reports (UCR) were used to document the number of arrests in each county. UCR data theoretically include all arrests nationwide, but they have well-known limitations that derive primarily from differences in reporting. Other limitations in the UCR data meant they could be used to identify arrest numbers in only some ADAM counties. First, the UCR includes arrests that do not result in bookings. This is principally the case for minor crimes in which the individual receives only a citation. Second, the UCR excludes some arrests that do result in bookings. Warrants and revocations are examples. Third, some arrests are double-counted in UCR data. Despite these limitations, UCR data were used in some sites to develop county-level sampling plans. At the same time, ADAM worked with the sites to identify alternate sources of data to validate the preliminary target numbers. On the basis of these numbers, ADAM set initial targets for each site in number of adult male arrestees who should be interviewed, although this was done with the understanding that more accurate data, provided later, might lead to changes in the sample sizes. Defining the population to be sampled Success in implementing the ADAM sampling designs requires understanding the movement of arrestees from one facility to another and the length of time they spend at each facility. Arresting authorities identified which facilities were those where arrestees were booked. They also provided in-depth information about booking and arrest procedures, including potential points of release in the field (that is, release where the arrest took place) and release from local booking facilities. ADAM needed to know about the extent of law enforcement's discretion in arrest and release decisions in each jurisdiction and have some understanding of the sites' transfer and hold procedures. It is important to know, for example, whether local booking facilities have holding capabilities and, if not, how soon and to what facility arrestees are transferred. In general, law enforcement agencies in all counties can exercise some discretion in whether to release on citation people who violate city ordinances. They can do so either in the field or from local booking facilities. Whether an arrestee is processed in the field or is booked and released has implications for ADAM sampling, because it affects the size of the arrestee population available to participate in the program. Arrestees released on a field citation are not available to be interviewed. This makes it necessary to understand the categories of arrestees who have no probability of being selected and what proportion of the arrestee population is processed this way. Just as important as understanding release on field citations is understanding the procedures used to book and release arrestees from the local booking/holding facilities or stations. This includes knowing how much time the procedures take and where book and release occurs. Because the population of arrestees who are booked and released generally consists largely of misdemeanants, it must be included in the ADAM sample. Without it the representativeness of the sample would be called into question. Both options--field citations or book and release--are typically available for processing the misdemeanor population but not the felony population. Systematically excluding arrestees who are released would heavily bias the sample against people arrested for minor offenses and would thus potentially inflate the extent of drug use among ADAM respondents. Ensuring fiscal accountability Cost was a consideration in redesigning ADAM. At some ADAM sites, especially those where there are several facilities, expanding the catchment area to the entire county significantly increased the cost of data collection. Cost considerations necessitated a series of trade-offs between maintaining expenditures at a reasonable level and retaining the overall goals of the program. The first trade-off involved setting the target numbers of interviewees. All DUF sites tried to obtain 250 interviews per quarter, a target goal that often took several weeks to achieve in some sites. Any increases or decreases in this target number under ADAM would have cost implications, favorable or unfavorable. The best justification of the cost of probability-based sampling was that it produced, at each site, a sample size large enough to ensure a reasonable level of confidence in using the data (in other words, a sample large enough that the level of variance would be acceptable). Logic dictated that the Los Angeles County sample would be larger than that of Webb County (the primary city is Laredo, Texas), for example, but the question was: how much larger? The decision to base sample size on total number of bookings in the county and the number of facilities in the county (based on a sample equivalent to the variance resulting from sampling proportionate to size) met two needs. It generated a sufficient number of cases from which to estimate the number of people in a county who have certain characteristics and to distribute the cases equitably among the sites. Some smaller sites, understanding that their target numbers needed to be smaller than those of larger sites, were at the same time concerned that it might be difficult to make credible statements about drug use and related behavior on the basis of the numbers. A target of 75 cases, for example, might be large enough for sampling purposes, but local officials might be reluctant to use it to shape policy. Here the trade-off was meeting local needs while developing standard errors (such as those for confidence intervals) that were reasonable in relation to those of other sites. Cost considerations also affected the number of interviewers working in each jail. In order to contain costs, it became necessary for NIJ to set the number of interviewers per shift per facility. In doing so, NIJ also created a built-in mechanism for ensuring that sites adhered to their own specific sampling plans. The plans emphasized selecting the sample of arrestees to be interviewed rather than the total number of interviews completed. However, staff at the sites continued to focus on completing the prescribed number of interviews. The conceptual shift from quantity to quality was difficult for many sites. Practical issues in implementing the county-level design Conceptually, the new approach to creating county-level sampling designs was relatively straightforward: identify the number of booking facilities, establish the general design, find out how many arrestees move through each facility each day, and set sample targets for the county and each facility within the county. In practical terms, the new approach was not so straightforward. Except in a few sites, the UCR data were not useful in identifying the number of arrestees booked at each facility or the number of booking facilities in a county. The sites found it difficult to obtain information about the flow of cases through each facility and to gain access to the facilities selected. Identifying facilities Although seemingly a straightforward task, finding out how many facilities there were in a catchment area was not easy. The Uniform Crime Reports do not contain this information. Counts by county law enforcement authorities varied with the definition of a booking facility. Further complicating the task of counting facilities was that arrestees are often booked many times, first in the local jail and again in the county facility. Thus, although county facility staff might be correct in their assessment that all arrestees in the county are booked in the county facility, arrestees may be booked in local facilities as well. These approaches failing, the one that succeeded was to contact the source; that is, the arresting agency authorities in each county. Each ADAM site thus systematically contacted all arresting agencies in its catchment area to determine where arrestees are booked. Gaining access In the past, jurisdictions wishing to participate in the DUF or ADAM program submitted to NIJ letters of agreement from local jail facilities that ensured access to those facilities to conduct ADAM interviews. These letters generally applied only to the primary county facility but did not guarantee the site would be permitted to collect data in all facilities selected in the sampling plan. To gain access to all facilities selected, local ADAM staff contacted facility administrators, explaining the program and the reasons for including their facility. In general, jail administrators' initial reaction was to question the program or deny access because it would delay the booking process, interfere with operations, and raise security concerns. Often, NIJ or NIJ's ADAM contractor intervened. When jail administrators understood that the program had been in existence for more than 10 years and that numerous facilities nationwide permitted data collection without adverse effects, they were more amenable to working out an arrangement. In a very few cases, access was still denied, and higher law enforcement officials or other city/county officials were contacted. In rare circumstances, despite all efforts, the request was denied. When access is denied When access is denied, replacing the site or a specific facility is easily accomplished within the parameters of the sampling design. Increasingly, however, replacements are required repeatedly where there are several small facilities in the stratum. One cause may be lack of security at the facility. ADAM requires that to ensure safety and security, law enforcement officers observe the interviewers and arrestees during the interview process. Often, small facilities could not participate because of a shortage of officers. When a department is short-staffed and the jail or booking area is not sufficiently staffed to provide adequate security for civilians (ADAM interviewers), the program cannot continue at that facility. Even if an overtime incentive is offered, there may not be enough officers to work the additional hours. The other reason repeated replacements may be necessary is low case flow, which can make the cost of data collection prohibitive. Interviewers may be at the facility for several hours or even an entire shift without conducting a single interview, because no arrestees have been booked. In Bexar County (San Antonio), Texas, for example, there are approximately 25 booking facilities, but most book fewer than one arrestee per day. Again, the issue was resolved by trade-offs. The sampling plan was adjusted to eliminate facilities that produced fewer than three cases a day. In making all such trade-offs, the balance is between cost and risk to the integrity of the overall sampling plan. "Specialty" facilities Some jurisdictions have facilities where people arrested for only certain types of crimes are booked. For example, there may be a facility dedicated to booking people arrested on domestic violence charges. If ADAM interviews do not take place at these facilities, people charged with these types of offenses will not be included in the arrestee populations for sampling purposes. On the other hand, if ADAM uses its resources to interview at these facilities, only those specific types of arrestees will be interviewed. The challenge is to reach these arrestees before they are brought to the specialty facility, but this is not feasible in all jurisdictions. For now, the sites having specialty facilities must adapt by limiting the representative nature of their sample through adding written caveats to their findings; for example, by making it clear in their reports that their sample does not include domestic violence cases. Defining the arrestee population Counts of arrestees, essential to sampling, were difficult to ascertain. Arresting agencies maintain numbers for operational reasons, not for research purposes. ADAM had to use the operational numbers as the basis for constructing a mechanism to create a representative sample. One reason it is difficult to count arrestees is that their movement throughout a county is considerable. They are often transferred throughout the criminal justice system, sometimes very quickly. Many factors contribute to this movement, including geographic imperatives, municipal requirements, and overlapping jurisdictions. Additionally, the criminal justice process has several stages, with each one at a different location. The possibility of double-counting when arrestees move is inevitable because a given facility may not be aware that the arrestee has already been counted for ADAM's purposes. These factors affect all counties, but become more problematic as the number of law enforcement agencies and booking facilities in the county increases. Movement also means an arrestee who should be counted may not be. In most facilities, an arrestee's booking sheet follows him or her to each succeeding stage of the criminal justice process. Therefore, if an arrestee has left the intake area or been transferred to another facility, no record of demographic data and offense characteristics may be available. The booking log may indicate that the arrestee was booked into the facility, but if that arrestee (with his records) has gone to court or another facility, it would be difficult to include him in the sampling plan. In some ADAM sites, the individual idiosyncracies either in the facility populations, the movement of arrestees within and among facilities, and the booking processes can only be documented. These sites must satisfy themselves that their sample has limitations and must make it clear they exist. In some sites, such documentation will lead to adaptations that enable their samples to be consistent with the overall sampling plans. The facility-level sampling plan design Before data collection began at the facilities, a facility-level sampling plan was established for each one. The plans had several steps: setting the targeted number of interviews, determining what time of day the interviews take place, and identifying the number of interviewers needed on each shift. Setting a targeted number of interviews The number of interviews to be conducted quarterly was identified for each site. This was the site's target number. Then, to identify the number of interviews to be conducted at each facility in the site, the target was simply divided by the number of sampled facilities, proportionate to size. For sites having stratified and stratified cluster designs (those with, respectively, 6 or fewer and more than 6 facilities), the county sample target was divided among the strata and/or facilities, proportionate to the number of bookings each contributed to the whole. A site with a city jail and two suburban jails, for example, might have a site sample target of 168 completed interviews. Annual booking statistics there might indicate that 50 percent of the county's arrestee population is booked at the city jail and the other 50 percent at the two suburban facilities (25 percent at each). For the city jail the sample target would be 84 completed interviews and for each suburban jail the target would be 42. Once the total sample per facility was determined, the number of days required for data collection was set. Collection must take place every day of the week in order to account for variations, by day of the week, in the type of crimes for which arrestees are charged. The length of time data are collected is based on the average number of bookings per day. Using the example cited above, and assuming the daily flow in the city jail is significant, a 7-day collection period, with a daily sample target of 12 completed interviews, will meet the targeted 84 interviews. (ADAM assumes each week in a given calendar a quarter is generally like any other, so the sites do not collect data on holidays or during days when special local events take place, such as Independence Day, and Mardi Gras in New Orleans.) What time of day should data be collected? When the optimal time of the day for interviewing is determined, that becomes the data collection shift. In the probability-based design, an 8-hour shift represents a 24-hour period, and all arrestees booked during that period have a known probability of being selected for the sample. The "stock and flow" design of the sampling plan (described above) increases the likelihood that the sites will sample and interview arrestees charged with lesser offenses, whose numbers are typically larger in a county arrestee population. Because of the less serious nature of their charges, these arrestees will be released more quickly than the more serious offenders and therefore the window of availability for interviewing them is smaller. The data collected from the "flow" cases are weighted to represent lower-level offenders booked during the "stock" period who were released before collection began. The success of this process relies on a site's ability to maximize data collection from arrestees charged with all types of offenses by collecting during the busiest 8-hour period of the day. The "flow" period, which begins the moment the data collection team enters the facility, represents the period of the day when the number of bookings is highest. After an interview is completed, the interviewer then selects the arrestee whose booking time was closest to the time of that interview. This procedure ensures the interviewer works throughout the shift, regardless of the number of interviews completed. For "stock" (which comprises arrestees who were booked and whose numbers accumulated during the time when data were not collected at the site), interviewers work with the facility to develop a list of all arrestees booking during the stock period, organizing it chronologically by the time each arrestee was brought to the facility. Arrestees to be interviewed are selected at intervals determined by the stock sample target. Number of interviewers Once the interview time (shift) is identified, the sample targets for stock and flow are calculated. The basis is the number of daily bookings estimated during each of these periods. For example, if 50 percent of the daily bookings occur during the flow timeframe, a site with a daily sample target of 12 completed interviews would have flow and stock sample targets of 6. These targets are the basis for determining the number of interviewers. In calculating this number, the assumption is that one interviewer can complete approximately one interview per hour. In most cases, one interviewer is assigned to stock and another to flow. The difference is that the flow interviewer works the entire 8-hour shift, regardless of whether the target is met or surpassed, and the stock interviewer works to meet the daily stock quota and then ends the shift. This means the number of resources or interviewers in a given facility will be constant, regardless of whether, on a given day, the flow of arrestees is high or low. Additionally, because an interviewer is always needed throughout the flow shift, the probability is high that arrestees booked in low-flow facilities will be interviewed. The interviewer can usually interview all of the remaining stock and each flow as they are booked. The requirement that the number of interviewers be kept constant made implementation easier. Predicting daily flow activity in a jail and modifying the number of interviewers accordingly would have been difficult. Having the number of interviewers remain constant also is important in weighting the data because otherwise it would affect assumptions about the probability of selection in a way that could not be predicted. Practical issues in implementing the facility-level design In implementing the facility-level design, there are a number of factors essential to the success of the sampling. These include gaining access to booking data, identifying arrestees ineligible for the sample, determining when and where arrestees are to be interviewed, tracking arrestees' whereabouts as they are processed, ensuring the interview space is secure and, in general, adapting to the jail environment. Access to the booking data A site's ability to implement its sampling plan is directly related to the facilities' ability to provide the necessary materials with which to select stock and to establish an efficient process for identifying and interviewing flow. Access to booking and census information is essential if the facility is to participate in ADAM. Stock samples are drawn from a list of all arrestees booked during the stock period (the period after the data collection shift ended the previous evening). Obtaining this list can be particularly difficult for facilities that do not have an automated booking system. It often requires consulting one or more handwritten logs. For example, the interviewer would use the intake log, which identifies everyone booked during the stock period, and the inmate log, which identifies stock arrestees who are still in the jail. ADAM site staff merge these two lists to select the stock sample and the replacements for released arrestees. It is typically easier to obtain stock samples in jails that have automated booking systems. In the many facilities where booking staff do not have the authority to query the system, ADAM staff rely on command-level staff or department programmers to generate a report that can be used to create a stock list. However, many of these departmental reports either exclude arrestees released from the facility or do not cover the full stock period, which often begins in the middle of the night. In many sites, the stock selection process does not end at this stage. This is because the information used to identify and select stock often does not include such items as the arrest charge, the specific location of the arrestee in the jail, and other basic variables. Without this information, it is not possible to know whether an arrestee is available or eligible to participate in the ADAM program. Often, negotiations with facility staff are necessary to develop a procedure for obtaining information in a timely manner. Because many facilities purge the booking information after an arrestee has been released, site staff often work with facility staff to manually look up charge and location information for each person before beginning the selection process. In other cases, site staff return later in their shift and refer to the booking slips and/or the jail management system to fill in the information needed for the facesheet of the questionnaire for selected and replacement stock. This process is particularly time-consuming in jails where the booking information physically accompanies the arrestee as he is transferred to various locations in the jail for various purposes (for example, intake, fingerprinting, booking, classification, housing). Flow selection is often easier because it requires only access to arrestees as they are booked into the facility. In low-flow facilities, where arrestees are booked sporadically, this is often done by observation, but in most facilities it must be more systematized. In some cases, sites use an intake list or medical screening list to identify arrestees as they are booked or they use the booking slips, as they are generated, to select flow arrestees for interviewing. The major concern in flow selection is to ensure that certain types of cases that might be overlooked (for example, arrestees whose booking sheets/cards take a relatively long time to generate) are not. The method used to develop the stock list should also drive the flow selection process, because the time of day when stock and flow are selected must be the same for both groups. In most ADAM sites, this is the time when the arrestee comes into the facility (the intake/booking time) or the time when arrest and booking information is entered into the computer. Systematically recording time for all arrestees, no matter what charge or disposition, is essential. Using different times can affect the sample. Thus, for example, in some sites arrestees charged with minor offenses might not be screened for medical problems. If a site uses medical screening time as the time for selecting stock and flow, lower-level arrestees may be systematically excluded from the sample. Identifying eligibles and ineligibles When drawing the sample, the ineligibles must be excluded. In some facilities, it is particularly difficult to identify them. ADAM focuses on people arrested for crimes committed in the local jurisdiction--the county. Thus, arrestees held for crimes committed in other jurisdictions, including other counties, as well as for Federal crimes (for example, arrestees taken into custody by the INS, DEA, FBI, or the U.S. Marshal's Service), or arrestees remanded into custody by the court, are not of interest to ADAM. In general, these ineligibles are not represented in local arrest statistics or do not reflect an arrest on a new charge. Gaining access to the arrestees To implement the facility-level plan successfully, sites must understand the booking process and the restrictions it imposes on ADAM's ability to access and interview arrestees. In some facilities, booking may take only 20 minutes, and in others it may take as long as 8 or 10 hours--decreasing or increasing the possibility of the person's being interviewed. The amount of time can be determined by the number of arrestees brought in at once: more arrestees means more time is needed. In some facilities, local ADAM staff are permitted to interview arrestees before the booking process is completed. When this happens, the ADAM arrestees are more likely to represent a broader range of offenses because those who may be released on bond or on their own recognizance (that is, those charged with minor offenses) can be interviewed. In some facilities, the intake areas are chaotic, making it unreasonable to expect ADAM staff to conduct interviews in them. However, to the extent that ADAM interviews are conducted after the original intake, the likelihood increases that arrestees charged with minor offenses have been released and thus not represented in the sample. For stock and flow collection, ADAM interviewers need access to arrestees who are at various stages in the booking process, in order to sample a broad range of arrestees. In the past, many sites conducted DUF interviews at times and in locations convenient to facility staff, typically in the housing areas. Frequently, the result was a sample consisting only of arrestees in the general population of the jail or housing unit of the jail. Although acceptable for stock interviewing, this approach does not work for flow collection because arrestees may be released before they reach the housing cells. Other sites experienced the opposite problem: obtaining interviews from arrestees not in the intake area. DUF interviews had been conducted in intake, because facility staff were particularly concerned for the security of civilians in the housing area, or they were reluctant to move housed arrestees back to the intake area for interviewing. In almost all cases, site staff have negotiated a reasonable solution, usually involving selecting an interview location for stock and another for flow, or bringing one population into another area to accommodate the interviewers. In addition to negotiating with ADAM staff to determine when and where interviews take place, facility officials also determine the broad categories of arrestees who may be interviewed. In almost all cases, arrestees in medical units and psychological units do not participate in ADAM. Some arrestees are deemed too inebriated or are verbally abusive or violent and cannot be interviewed. Local site staff work with facility staff to expand the interview population as much as possible and to ensure that certain populations are not unintentionally eliminated. To eliminate unnecessary bias, sites complete "face- sheets" (forms containing information) on unavailable selected arrestees whose behavior initially prevents their being interviewed, and attempt to interview them later in the shift. Adjusting to arrestees' movements in the system Arrestees may be released or transferred before they can be interviewed. Because the stock is selected as many as 16 hours after arrestees arrive in the facility, some may have been released before data collection begins. As a result, sites may find it difficult to meet their quota of stock interviews. When this happens, the sites must find out whether the arrestees are being released into the community or transferred to another jail; whether certain groups of arrestees are missed when those released are not interviewed; whether arrestees are released, either after arraignment or through some other process; and whether there is a better time or location for conducting stock interviews. Depending on the answers, collection procedures may change so that stock are interviewed before a possible transfer or selected stock are interviewed at the new location. If large numbers of arrestees are released after arraignment, procedures may be changed to interview before arraignment. One change would be to split the collection shift into two periods of the day. It is more difficult to address the situation if arrestees are released on a preset bond and therefore not arraigned. Information about arrestees' movements is also important to ensure comparability from site to site. In one county, for example, close to half the arrestee population, regardless of arresting agency, receives citations in the field. The UCR data include these cases, which means they were also included in the statistics used to develop the ADAM sample targets. Because this made the site's sample targets too high, they had to be adjusted to account for the unique nature of the arrest process there, which generated a large proportion of arrestees who were not available for ADAM interviewing. It may or may not be possible to solve the problem of arrestees released from stock. But if the release problem arises during the flow period, it must be resolved, because it means the flow selection process is not being implemented as designed. Ultimately, it may not be feasible to interview all these cases, because the facility strives to process them as quickly as possible and will not allow interviewing to interfere. In most instances, the problem can be solved by negotiating access to different intake lists or holding areas in the facility. Security Security plays a major role in the success of sampling at the site level. Interviewers need a safe and private environment in which to conduct the interviews, but they also must meet any conditions the facility sets. In some sites, this may mean there is a glass partition between interviewer and arrestee; in others it may mean there are cell bars. Security may also affect the site's ability to implement the sampling plan if there are limitations on the number of hours permitted for stock interviewing or if interview times must accommodate meal schedules or lights out hours. These conditions can usually be met, but they require planning on the part of site staff. Security may play a major role in selecting arrestees to be interviewed because many jails do not allow civilians access to the information management systems needed to select stock or flow interviews or to complete facesheets (information forms). In these situations, it is essential that the facility officer understands what information is needed and does not unwittingly bias the sample in an attempt to assist in the research. This might happen, for example, if the officer excludes certain arrestees he or she deems unruly. Adapting to the jail environment Even if all processes specific to the facility are carried out in the sampling plan, the dynamic nature of the jail environment may adversely affect a site's ability to meet its daily target of interviews. For example, delays occasioned by lockdowns, meals, counts of arrestees, and fights among arrestees may in turn delay interviewing. In the case of regularly scheduled events, the data collection shift can be changed to limit the effects of the delays. But even regularly scheduled events are not always predictable. For example, the amount of time needed to count arrestees may vary significantly from one night to the next, lasting as long as two hours in some cases. Conducting research in a jail environment is different from and often more challenging than conducting the same research in prisons. Jail administrators act as the gatekeeper of the local criminal justice system. They are responsible for intake, medical screening, classification, release, feeding, and moving to and from court and otherwise transporting people who have only recently been arrested and are often unruly, intoxicated, or violent. To protect the balance of power and order in the jail, local ADAM staff will invest whatever effort is necessary to build relationships with facility staff, understand the rules and restrictions, and adopt procedures that adhere to those rules without transgressing the ADAM protocols. The result: more reliable and useful data Standardized data collection protocols and probability-based sampling will increase the reliability of ADAM's data and subsequent findings. With the adoption of probability-based sampling and the expansion of the questionnaire, ADAM will be more useful than in the past as a platform on which researchers can conduct studies of various aspects of drug use and related behavior. The expanded questionnaire enables local communities to validly estimate the prevalence of a variety of measures, including the proportions of arrestees who test positive for drug use by urinalysis and the proportions who need treatment. And for the first time, the questionnaire enables the ADAM sites to develop prevalence estimates of drug use in the nonarrestee population. * Phyllis J. Newton, M.A., of Abt Associates Inc., was the contracting Project Manager for the ADAM program; Margaret E. Townsend, M.A., is Field Operations Manager with Abt. Notes 1. These national collection programs are the National Household Survey on Drug Abuse (NHSDA), the DEA's System to Retrieve Information from Drug Evidence (STRIDE), and the Treatment Episode Data Set (TEDS). Other surveys used by ADAM are those conducted by the Center for Substance Abuse Treatment and the U.S. Census Bureau. 2. Variations in the booking process are discussed in the section on implementing facility-level sampling plans. 3. U.S. Census Bureau, 2000 data. 4. Data were collected during the first quarter of 2000 from all five boroughs. Because of unresolved sampling issues and cost constraints, data were collected in the second, third, and fourth quarters only in Manhattan. 5. A more in-depth discussion of the sampling and weighting method is in Methodology Guide for ADAM, by Dana Hunt and William Rhodes, Washington, D.C.: U.S. Department of Justice, National Institute of Justice, May 2001. The Guide can be downloaded from the ADAM Web page (http://www.adam-nij.net) on the NIJ Web site (http://www.ojp.usdoj.gov/nij). 6. Cost constraints and other practical limitations led ADAM to adopt probability-based sampling for adult male arrestees only. --------------------------- Setting and Ensuring Performance Standards To promote standardization of ADAM data collection at the sites, NIJ established basic criteria for data collection procedures. At each site, the agency also implemented training in the procedures. The contractor that administered the ADAM program (Abt Associates) developed the training materials that were used at all sites. It ensured the procedures for data collection were followed and observed data collection. Editing instructions for the questionnaires were provided to all sites as a tool for ensuring quality data. Common performance standards were established for all sites. They included requirements for meeting data collection targets (number of arrestees), for training, for minimum error rates in conducting interviews, and for adherence to a specified fiscal standard. A sampling plan to be used by all sites was developed and implemented. When data collected from the interviews at the sites were received at the ADAM Data Center, they were all subjected to the same data management procedures by means of automated editing and entry programs. By establishing these quality controls, NIJ anticipated that comparison of findings from site to site would be more defensible than in the past and that ADAM data would be that much more useful in informing local and national discussions of substance abuse. ---------------------------- VIII. "Calendaring" in ADAM: Examining Annual Patterns of Drug Use and Related Behavior by Dana E. Hunt, Sarah Kuck, and Patrick Johnston* Among the unique features of the ADAM survey instrument (questionnaire) fielded in 2000 is a technique called "calendaring." It is designed to examine drug use and related behavior over the period of an entire year and month to month within a year. To measure drug use only in the recent past (30 days) does not convey the complexity of the behavior. To measure drug use over a year addresses the complexity issue, but events that took place as long ago as a year may be difficult to remember. Calendaring helps solve the problem of recall by prompting the arrestee, during the interview, with questions about events that took place at about the same time during each period of time. The prompts include questions that can be used to "crosswalk" ADAM data with data in other surveys. Calendaring promotes recall Research designs often relinquish any attempt at long-term recall, instead asking people about events that took place in a single, relatively brief period of time (the past month, for example). Drug use, though a chronic behavior, changes over time. Thus, the recent past is not likely to be representative or typical, either of use or treatment history. In ADAM, it may be that at the time the arrestee was interviewed, he or she was in a period of escalated drug use (one that resulted in the arrest). Moreover, because drug use is a socially stigmatized behavior, people may be reluctant to respond to questions about recent use. Alternatively, research designs may focus on drug use over a longer period, presumed to be more typical of use patterns. The added advantage is that people more readily admit to drug use in the more distant past than to use in the past few days or weeks.[1] In this approach, arrestees might be asked to describe their annual consumption of drugs. But when frequency of drug use is very high or if it changes, recall can be compromised. It would not be difficult for a teetotaler to remember drinking no alcohol, but someone who has several drinks on certain days and fewer on others would find it difficult to come up with an accurate count of overall consumption. Asking people about an even longer period of time introduces significant chances of error and promotes guessing or mental averaging. The calendar approach to collecting data was developed to deal with recall and related issues. It was used in early research on drug careers[2] as well as in other fields in which researchers are interested in recall of events over an extended period.[3] Calendaring has been shown to increase the accuracy of recall over even longer periods than 12 months (the period used in ADAM).[4] It promotes recall by dividing the year into units conceptually manageable by the arrestee and then "anchoring" his memory around interconnected, real-life events that occur in these units of time. ADAM's calendar information consists of data covering events as they occur in one-month units of time in a 12-month period. For each unit of time, data on drug use, drug treatment, mental health treatment, number of arrests, and residency status of the arrestees are obtained during the interview. Each one-month "box" in effect contains sets of interrelated life events and behaviors. These events vary in duration (number of months) and intensity (level of drug use, for example) and may be correlated with or conditioned on each other. Thus, for example, the likelihood of using drugs at Time 1 is related to the likelihood of using drugs at Time 2. The relationship of one event to another can also be contemporaneous (occurs in the same month) or delayed (occurs in subsequent months). The ADAM interviewer records the behavior or events throughout the time period of interest (one year, month by month). The connections among the behaviors and events are presented visually (in a calendar) and mentally through a series of interrelated questions about life events. The life events themselves serve as cues to recall of other events. For example, the arrestee is asked where he was living, whether he was in treatment, whether he was arrested, and the approximate extent of his drug use. These questions are asked for each of the 12 months of the past year. As each type of behavior is recalled, it becomes a further anchor for recalling the next set of events. The result is a grid of events/behaviors, related to each and occurring over a period of time. Calendaring increases accuracy With calendaring, information becomes more accurate because it reflects aided recall of patterns unfolding over relatively long periods. For example, to calculate the amount of cocaine consumed or purchased by arrestees in an area in a year, ADAM data on total annual consumption of cocaine among users involved in the criminal justice system could be examined. If data from a single point in time were used to extrapolate to the entire year, that estimate would likely be biased. An arrestee might say he used cocaine 15 days in the past 30. In the previous months, however, he might have used it only one or two days in some months and not at all in others. Using past 30-day patterns as typical of a year would seriously inflate calculations of annual consumption of cocaine. Measuring "typical" drug use That the commonly used 30-day recall period may not substitute for typical or modal behavior is evident from self-reported drug use by adult male arrestees in the ADAM sample. Data from 2000 for selected sites--New York, Phoenix, and Las Vegas[5]--show that past-30-day crack use patterns were not in all cases the same as longer-term patterns. Crack users among adult male arrestees in New York were far more consistent in their long-term use than were their Phoenix or Las Vegas counterparts. Ninety-five percent of crack users in New York said they used the drug in the 30 days before their arrest and a similarly high proportion--82 percent--reported using it at least once in each of the previous 12 months (only 9 percent reported using crack in fewer than six months of the previous year). (See Table 8-1.) In Phoenix and Las Vegas, by contrast, the proportions who used crack throughout the year were far different from the proportions who used it in the past month. The 81 percent of crack-using arrestees in Phoenix who said they used the drug in the past 30 days dropped to 41 percent for use in all 12 previous months (41 percent had used crack half the year or less and 16 percent in only one month of the year). In Las Vegas, where the proportion who used crack in the recent past 30 days was also relatively high (78 percent), the figure for use in all 12 months of the past year dropped to 44 percent, and to one-third for use in fewer than six months of the past year. New York's 30-day figures could reasonably be used as the basis of an assessment of crack use at that site, because the level of use was consistent over time. In the other two sites, the picture was different. What difference do these differences make? If past-30-days data (when the proportions who used crack were high) were used to assess the amount of crack consumed in these sites, the result might be serious overestimation. While it might be safe to assume that the most recent month's use of crack in New York was like any other month's use, it would not be so for Phoenix and Las Vegas. At those two sites, crack use appeared more variable or episodic. A user might, for example, use crack one month, skip a month, return to use, and so forth. Calculating repeated arrests Calendaring also increases accuracy in calculating the rate at which people in the ADAM sample are arrested.[6] DUF and ADAM have always reported number of arrests rather than number of arrestees.[7] When number of arrestees rather than number of arrests are reported, this conceals repeat offending. In fact, law enforcement has long noted that repeat offenders are overrepresented in arrestee statistics. With calendaring, the new ADAM survey instrument makes it possible to examine the number of repeat arrests of a single arrestee in the sample. Many variables affect rates of arrest. ADAM measures some of them, including age and previous arrest history. Data from selected ADAM sites also show that the number of times someone in the ADAM sample has been arrested varies with type of drug used.[8] (See Exhibit 8-1.) In the cities where methamphetamine use was detected among adult male arrestees--San Diego, Oklahoma City, and Sacramento--the rate at which users were arrested was higher than the rate for users of any other drug. Oklahoma City's methamphetamine users were arrested almost three times more often than heroin users; in Sacramento, methamphetamine users were arrested more than twice as often as heroin users. In sites where no methamphetamine use was found among arrestees (New York, New Orleans, Atlanta), cocaine users were arrested most often, though the difference between users is not as dramatic as for methamphetamine. ADAM data support the law enforcement observation that many arrestees come back to the criminal justice system again and again. Nevertheless, about half the adult males in the sample in these selected sites indicated that their current arrest was their only arrest in the past 12 months. (See Table 8-2.) In New York, for example, almost half the adult male arrestees said the current arrest was their only arrest in the past year. By contrast, slightly more than half the arrestees in New York were arrested between one and five times in the past year and 1 percent said they were arrested 10 or more times in the past 12 months. How ADAM data can be used with other measures of drug use Calendaring also offers the opportunity to use ADAM data in conjunction with other measures of drug use and related behavior to find out whether arrestees are covered in these counts. Among these are the National Household Survey on Drug Abuse (NHSDA), the Treatment Episode Data Set (TEDS), and the State Treatment Needs Assessment Program (STNAP).[9] In developing the new survey instrument, the ADAM program made certain that the variables were defined in a way that would enable the resultant data to be "crosswalked" with (directly compared to) data in these and other relevant surveys and data sets. Crosswalking makes it possible, for example, to identify people not counted by NHSDA but included in ADAM. The NHSDA examines drug use by all people in the general population age 12 and older who are members of a household. It excludes people who are homeless, living in a temporary shelter, confined in jail, or in like circumstances. In ADAM, the questions about residence, whose answers are recorded on the month-by-month "calendar," were designed to match those in NHSDA. Comparing NHSDA and ADAM data reveals that in most of the selected sites examined, large percentages of arrestees who used drugs in the year before their arrest would not have been included in the Household Survey. They were excluded because they were transient or lived in unstable housing at least some part of that time.[10] (See Table 8-3.) The range among these sites was 4 percent (Birmingham) to 32 percent (Honolulu). While the NHSDA is the country's premier survey of drug use in the general population, it may miss some people who are the heaviest consumers of illegal drugs. Crosswalking ADAM with TEDS reveals a similar pattern. The Treatment Episode Data Set contains information about extent of drug and alcohol treatment. ADAM mirrors the TEDS definition of an "episode" of treatment, with questions about treatment received by drug users month by month in the past year.[11] In the sites selected for examination, the proportions of adult male arrestees who used drugs in the year before their arrest and did not receive any inpatient or outpatient treatment were high.[12] In New York, 81 percent of them did not participate in treatment; in New Orleans and Birmingham, the figures were 90 percent or higher. Thus, the TEDS data do not reflect large proportions of drug users in the arrestee population. The State Treatment Needs Assessment Program (STNAP) is a CSAT program that collects information about drug abuse and uses it to estimate the need for substance abuse services. ADAM crosswalked STNAP data to identify the proportions of drug-using arrestees who might be excluded from STNAP. Because the STNAP survey is conducted by phone, anyone who does not have a phone cannot be contacted. Having a phone is a proxy for inclusion in STNAP, so the redesigned ADAM survey instrument included a question about the number of noncommercial phone lines where the arrestee was living. By this measure, the proportions of arrestees[13] who were not counted in the assessment of treatment needs were high in the selected sites examined. The way in which these other national surveys of drug use are designed tends to exclude certain subpopulations. Analysis of the 2000 ADAM data revealed the extent to which this was so. ADAM provides the "missing" data on some of these drug users. Because DUF and ADAM have over the years consistently shown drug use among this at-risk population to be high, the ADAM findings are an important basis for State and Federal assessment of need and resource allocation. Using calendaring to examine patterns of drug use Much of the research literature on substance abuse describes it as a chronic and relapsing condition that involves cycles of moderate use, abuse, and abstinence. By permitting analysis of drug use and related behavior over time, calendaring affords insights into patterns of use, whether consistent or episodic. Setting these patterns against the backdrop of concurrent events in the lives of the arrestees can illuminate the dynamics of substance abuse. (For other ways in which calendaring is being explored by ADAM, see "Further Potential of ADAM Calendaring.") Patterns of heroin and cocaine use Data on self-reported past-year use of heroin and cocaine by adult male arrestees in New York reveal different levels or intensities of use. The percentages of arrestees who used these drugs most heavily were higher than the percentages who used them less frequently. Less than 20 percent of both heroin and cocaine users were involved with these drugs 1 to 7 days in each month of the past year, and similarly small proportions were involved 8 to 14 days per month. By contrast, fully 55 percent of heroin users and more than 40 percent of cocaine users were involved 15 to 30 days in each month of the year. (See Exhibit 8-2.) Information from three heroin users among the adult male arrestees in New York reveals one of many patterns, in this case escalating use in the months before arrest. (See Exhibit 8-3.) "User A" began the year using heroin fairly heavily, became abstinent, and escalated to a higher level of use in the months before his arrest. "User B" also escalated to a higher level before he was arrested, although his use over the 12-month period was more consistent--there was no period of abstinence. "User C" was a heavy user who nonetheless managed to abstain for three months but returned to heavy use in the two months before being arrested. Calendaring can document events in the lives of the arrestees that might affect their drug use. These include other arrests, jail confinements, and treatment experience. For "User B" (from Exhibit 8-3), time in jail affected the level of heroin use. The period in which this user reported reduced drug use corresponds to his arrest and jail time in month 7 of the 12-month period. (See Exhibit 8-4.) It is worth noting that this user said he was being treated for drug use (on an outpatient basis) during the entire 12-month period. * Dana E. Hunt, Ph.D., of Abt Associates Inc., is Principal Scientist with Abt and was the contracting Project Director of the ADAM program; Sarah Kuck is a Research Assistant with Abt; and Patrick Johnston, M.S., is an Associate with Abt. Notes 1. Anglin M., Y. Hser, and C. Chou, "Reliability and Validity of Retrospective Behavioral Self-Report by Narcotics Addicts," Evaluation Review 17 (1993): 91-108. 2. McGlothlin, W., M. Anglin, and B. Wilson, "Narcotic Addiction and Crime," Criminology 16 (1978): 292-315; and Nurco, D., I. Cisin, and M. Balter, "Addict Careers: A New Typology," The International Journal of the Addictions 16 (1981):1305-1325. 3. Freedman, D.A., et al., "The Life History Calendar: A Technique for Collecting Retrospective Data," Sociological Methodology 18 (1988): 37-68; and Furstenberg, F., et al., Adolescent Mothers in Later Life, New York: Cambridge University Press, 1987. 4. Anglin et al., "Reliability and Validity of Retrospective Behavioral Self-Report"; Freedman et al., "Life History Calendar"; and Collins, L., et al., "Agreement Between Retrospective Accounts of Substance Abuse and Earlier Reported Substance Abuse," Applied Psychological Measurement 9 (1985): 301-309. 5. A subset of ADAM sites was selected for ease and simplicity of presentation. No other selection criteria were used. 6. See Chapter 9 for an example of how essential rates of arrest are in estimating hardcore drug use. 7. In DUF and subsequently ADAM, number of arrestees is used as a proxy for number of arrests. 8. In the ADAM program, a user of one drug may also be using another; that is, these categories are not mutually exclusive. Research has shown that drug users may use a particular substance in preference to others, but they may also use other drugs. 9. The National Household Survey on Drug Abuse is conducted by the Substance Abuse and Mental Health Services Administration (SAMHSA), U.S. Department of Health and Human Services; the State Treatment Needs Assessment Program is administered by the Center for Substance Abuse Treatment (CSAT) of SAMHSA; SAMHSA also maintains the Treatment Episode Data Set (TEDS). 10. For NHSDA purposes, to be considered a member of a household a person need not own or rent the residence in which he or she is living, and the residence can be one of a variety of types (for example, a trailer, apartment, or house). However, to be included in NHSDA, a person cannot be a transient member of a household. If, for example, the survey respondent is living briefly (less than three months) with a girlfriend or a relative, he or she would not be considered a member of that household. 11. TEDS data come from reports of drug and alcohol treatment as measured by intake in treatment programs. An "episode" of treatment is measured as entry in an outpatient program and/or an overnight stay in an inpatient program. In the ADAM questionnaire, arrestees who say they used drugs are asked, for example, how many nights they stayed in inpatient treatment and how many times they entered outpatient treatment. 12. These drug users include people who may not need treatment. 13. The question was asked of all arrestees, not just those who said they used drugs in the year before their arrest. ---------------------------- Further Potential of ADAM Calendaring The ADAM program is exploring the potential for using calendaring to build predictive models. How would calendaring be different from simple regression analysis in studying the effect of one variable--such as treatment--on another? The difference lies in the nature of the data available to use in developing the model. A simple regression model might look at the impact of drug treatment on subsequent drug abuse, examining data on level of use before and after treatment and amount of treatment received. In "real life," however, the effects of treatment are conditioned on many variables, including the patterns of use the person brings to treatment, not simply the level of use at entry. Calendaring in ADAM makes available more dynamic information about these patterns. Calendaring in ADAM does have limitations that need to be taken into account in developing a predictive model. First, the ADAM data cover only a 12-month period, which is a relatively small window into the number of fluctuations or events in the career of the drug user. Second, the 12-month period is not likely to be "typical" in that in ADAM it always terminates with an arrest. In addition, the data are "left-censored"; that is, projections or predictions would be made about behavior with no information about them before the time frame studied. Clinicians often say that users have to "hit bottom" in drug use and other life crises before their treatment experience will be successful. One way to use a predictive model based on calendaring would be to try to measure "hit bottom." It may mean a series of experiences before treatment--escalating arrests, increasing drug use, increasing transience. These variables (number of arrests, level of drug use, residence status) are all measured in the ADAM calendar in each month before the event of interest (in this example, entry into treatment). The more traditional approach is to summarize these "events" without regard to either when they happened or to their interrelationship. By using the calendar data, it is possible to build models that can account for previous experiences and concurrent events and activities. For a given arrestee, the level of heroin use in June might be directly related to whether he was in treatment or in jail in March, April, or May, as well as conditioned on the amount of heroin he was consuming in those months. Level of use might also be correlated with arrests in those months. The effects of some interrelated events (being jailed, for example) might be immediate in reducing the amount of heroin consumed, while the effects of others (treatment, for example) might lag or be delayed by a month or two. ---------------------------- IX. Estimating Hardcore Drug Use in the Community by William Rhodes and Ryan Kling* Scientists and nonscientists alike seek to estimate prevalence--from the number of stars in the sky to the number of angels on the head of a pin. When policy analysts estimate prevalence, they focus on more prosaic topics. They seek to find out how many times a condition is occurring or an event is taking place because their clients--policymakers--need the information as the basis of reasoned decisions. Researchers have used data from ADAM and DUF (the Drug Use Forecasting program, ADAM's predecessor) to understand the prevalence of drug use and related behaviors among arrestees. The redesigned ADAM program now makes it possible to provide additional prevalence estimates, including the number of hardcore drug users in a county that has an ADAM program. How prevalence is estimated[1] "Hardcore" can be defined in any appropriate way. For example, a hardcore user might be someone who uses illicit drugs more often than some threshold number; alternatively, a hardcore user might be someone who is seen as needing treatment. To explain the estimation technique used here, a large rectangle represents the number of hardcore users in any ADAM county--the object of the estimation exercise. (See Exhibit 9-1.) Household surveys offer one way to estimate this number. However, these surveys would exclude a large number of hardcore drug users, either because they do not live in a household (as defined by the survey), because they are typically not at home when interviewers call, or because they lie about their drug use. Instead, inferences must be drawn about the large rectangle from information provided by the small square, which represents hardcore users who are arrested and booked. An initial assumption is that hardcore drug users all have the same probability of being arrested and booked. (They are said to be "homogenous.") The square is smaller than the rectangle to illustrate the fact that not all hardcore users are arrested and booked during a specific time period (one year, for example). The composition of the rectangle is inferred from information about the composition of the square. The ADAM data do not enumerate hardcore users booked into jail, because they are a sample. The sample is depicted as a triangle. Because the sample is probability based, the triangle can be weighted to estimate the square. An additional problem is that some hardcore users will deny their drug use or the level of their drug use, so admitted hardcore users in the sample underrepresent the actual number in the sample. The latter is represented with a cross. The problem stemming from underreporting could be overcome by estimating how frequently hardcore drug users are truthful about using drugs at the hardcore level. If the "truthfulness rate" could be calculated, the triangle could then be estimated on the basis of the cross and, if the sampling weights were known, it would then be possible to estimate the square from the triangle. Creating a model of the arrest process For the sake of simplicity, the explanation disregards the two complications of sampling and denial of drug use, and assumes the availability of information about all hardcore drug users arrested and booked in a given year (the square).[2] In practice, of course, neither sampling nor denial can be ignored during estimation. Estimating the composition of the rectangle from the composition of the square requires a mathematical model of the arrest process through which hardcore drug users move from the rectangle to the square. (See Exhibit 9-2.) Conceptually, modeling treats the police as if they were samplers, conducting a survey. The goal of the modeling is to determine how hardcore users in the booking population (the square) should be weighted to represent hardcore users in the county population (the rectangle). Data about previous arrests of hardcore drug users (in the square) provide a basis for estimating the rate at which hardcore drug users get arrested and booked. The inverse of that estimated arrest rate provides the means to weight the square to estimate the rectangle. Exhibit 9-3 is a simple illustration. The illustration assumes that the rectangle comprises 1,000 hardcore drug users (called "H"). By assumption they are identical with respect to the probability and rate of arrest; that is, each is arrested 0.3 times per year on average (the arrest rate is "R"). Thus, the 1,000 hardcore drug users generate about H x R = 300 arrests per year (the number of arrests is "A"). Continuing this illustration, the ADAM data would indicate that the square comprises 300 arrests of people who self-report hardcore drug use, and for them the interviews would reveal that the average arrest rate is 0.3 per year. Because H x R = A, then A/R = H. Thus, 300/0.3 = 1,000--the number of hardcore drug users in the county. This algebraic calculation illustrates the fundamentals of the estimation. Identifying and counting the 300 hardcore drug users requires a tabulation of the ADAM data (though keeping in mind the two complications, noted above, introduced by sampling and underreporting). The estimate of 0.3 arrests per year is the result of analysis of data about arrest histories obtained during the ADAM interviews. For the calculation to be correct, not all hardcore drug users need to be arrested. In fact, a hardcore drug user could elude arrest through the entire length of his or her drug use career and still be represented by the "sample" of people booked into jail. The estimation procedure makes no assumption that the booking population enumerates all hardcore drug users. Introducing measured heterogeneity. Abandoning the unrealistic assumption that hardcore drug users are homogenous in rate of arrest, the estimation methodology remains conceptually the same. This type of heterogeneity is called "measured heterogeneity" because a number of variables could explain the differences in arrest rates of hardcore users. Operationally the methodology is more complex, however. To illustrate the consequences of measured heterogeneity, the rectangle is divided into two parts (Exhibit 9-4), with the top half representing hardcore users of cocaine and the bottom half representing hardcore users of heroin. (There could, of course, be many types of hardcore drug users, but for illustration purposes, only two types are assumed.) The assumption is that each subgroup is homogenous; that is, all cocaine users are alike and all heroin users are alike. But hardcore heroin users are different from hardcore cocaine users; the former are arrested on average 0.4 times per year, while the latter are arrested on average 0.2 times per year. In this respect, arrest rates are heterogeneous but that heterogeneity can be explained by an observed factor-- type of drug use. Given these assumptions, the booking population (the square) is not representative of the county population (the rectangle). In the booking population, there are two hardcore cocaine users for every hardcore heroin user. By contrast, in the county, there is one hardcore cocaine user for every hardcore heroin user. Thus, the equation used previously, H = A/R, will not produce an accurate estimate because A is the total number of hardcore drug users (A = 300 in this illustration), while R is the average arrest rate in the booking population = (2/3)0.4 + (1/3)0.2 = 0.333 in this illustration). The previous calculation would produce an estimate of H = 300/0.333 = 900 hardcore drug users in the county population. Obviously, this is not correct. Accommodating measured heterogeneity. The correct approach accommodates measured heterogeneity in the entire population of hardcore drug users. (See Exhibit 9-5.) The type of drug (cocaine or heroin) is the variable differentiating the two groups of hardcore users. ADAM identifies hardcore cocaine and heroin use; hence, heterogeneity is explained by measured variables. In this example, the booking population consists of 200 hardcore cocaine users, whose average arrest rate is 0.4 per year. This means there are about 200/0.4 = 500 in the county. The booking population also includes 100 hardcore heroin users, whose arrest rate is 0.2. There are 100/0.2 = 500 of them in the county. Thus, there are 500 + 500 = 1,000 hardcore drug users in the county. Other than type of drug (in this example, heroin and cocaine), there are many measurable variables that should be taken into account when modeling the arrest process. The ADAM interview instrument was designed to obtain information about other major covariates (such as whether arrestees are employed). When the number of covariates expands beyond a single variable, however, dividing the booking population into mutually exclusive homogenous groups becomes impractical or even infeasible, and estimation then requires the use of regression models. Nevertheless, increased complexity of the model does not alter the basic logic of the approach. Introducing unmeasured heterogeneity. Not all factors that explain the arrest process are measurable, or at least not all factors are identified in ADAM. These too must be taken into account in calculating the number of hardcore drug users. To illustrate unmeasured heterogeneity, the rectangle is again redesigned. The upper part still represents hardcore cocaine users and the lower part hardcore heroin users (as in Exhibit 9-4). These two groups are further divided, with the divisions depicted as triangles. Hardcore cocaine users in the upper triangle represent those who have a serious mental illness. This dually diagnosed group has an average annual arrest rate of 0.5. Hardcore cocaine users in the lower triangle represent hardcore cocaine users who do not have a serious mental illness. Their arrest rate is 0.3. Hardcore heroin users are similarly classified. (See Exhibit 9-6.) The booking population (the square) still overrepresents hardcore cocaine users, but now it also overrepresents hardcore users who are mentally ill. ADAM does not include data about mental illness, so the modeling cannot account for this condition in the same way that it accounts for the cocaine-heroin distinction. Although the approach is conceptually the same, taking unmeasured heterogeneity into account requires mathematical modeling that defies a simple illustration. A technical report provides details.[3] Preliminary estimates of number of hardcore drug users Because the mathematical and statistical logic of the estimation technique is complex, the estimates of number of hardcore users are best presented visually, in the form of graphics (Exhibits 9-7 and 9-8). (The algebraic terms used are presented in "Key to Algebraic Terms.") Arrest rates and user rates As noted, "hardcore drug user" could be defined differently depending on policy objectives. The methodology is invariant with respect to definition. For illustration, hardcore use is defined here as the use of cocaine, heroin, or methamphetamine on ten or more days during the month before the ADAM interview. Exhibit 9-7 shows estimates of the arrest rate for the average hardcore drug user in each ADAM site. The estimate varies from site to site. Confidence intervals are appreciable, but most of them overlap a value of 0.75 arrests per year. On the basis of these data, it might be said that, in most sites, there are about 750 arrests and bookings per year for every 1,000 hardcore drug users. For some sites, the confidence intervals are extremely wide. This is largely because sample sizes are small, since several sites did not have data for all four calendar quarters. The presumption is that the confidence intervals will narrow as the ADAM samples grow. Exhibit 9-8 shows estimates of the number of hardcore drug users in the county per hardcore drug user in the booking population. The average of the 35 ADAM sites appears to be about 1,600 hardcore drug users in the county for every 1,000 hardcore drug users who appear among arrestees. Number of hardcore users in the community Table 9-1 shows what might be considered the ultimate estimate--the number of hardcore drug users in the county. Caution is advised in interpreting these estimates. One reason is the use of "stratified cluster sampling" in some sites. That is, when counties have many jails, ADAM interviewers cannot go to each of them, so jails themselves must be sampled. (These sites are marked on the table as "S/C," for stratified cluster sampling.) At the time of this study, ADAM had not yet made provisions for adjusting its estimates to account for nonsampled jails. Thus, in sites that use stratified cluster sampling, the estimates are too low. In one site, New York, the number of hardcore drug users is underrepresented because only in Manhattan were data collected in all four calendar quarters; in only one quarter were all five boroughs in the study. To adjust for the limitation, in the sites that use stratified cluster sampling (the S/C sites) the estimates should be doubled. That is only a crude adjustment. In Cleveland, Des Moines, and Minneapolis, doubling might produce too high a number. By contrast, the Detroit figures probably should be multiplied by a factor of five. For New York, the estimate might be roughly two and one-half times as great as indicated here. Another reason for caution in interpreting the estimates is that they are of adult males only. Because women constitute about 20 percent of the arrestee population,[5] the estimates should be inflated by about 1.25. However, precise adjustments would require taking into account the number of bookings of hardcore female drug users and the rate at which they are arrested and booked. When the ADAM redesign extends to women arrestees, the adjustment can be more informed. The estimates also exclude juvenile detainees, but this is a minor problem because there are good estimates of juvenile hardcore drug users in the annual Monitoring the Future (MTF) survey. Estimates based on the highly selective sample of hardcore drug users among juvenile detainees would probably not be much improvement over estimates based on MTF. One other limitation is that the methodology used to adjust for underreporting hardcore drug use is provisional. (Recall the step of drawing an inference about the triangle using information provided by the cross.) The ADAM project has not yet developed adjustments for underreporting drug use extending beyond the two or three days before the interview. Such adjustments would be welcome adjuncts to the hardcore user estimation methodology. Overcoming limitations in the estimation method Even beyond the limitations, the estimates should be considered provisional. ADAM is in its infancy. Improvements will lead to advances in estimation based on the ADAM data. There is no reason to assume that the method is immutable. Researchers will undoubtedly find ways to improve it. But even with the current limitations, the method of estimating the number of hardcore drug users at a given ADAM site can generate credible figures. And there is no reason that better estimation methods could not be applied retrospectively. What about people whose risk of arrest is low? The hardcore user estimation methodology does not require that all hardcore drug users be arrested. In fact, as noted above, many hardcore users may elude arrest throughout their entire drug-use careers and yet still be represented by the estimates. If the police are viewed as samplers, they no more need to arrest everybody than a sampler conducting a conventional survey needs to interview everybody for the resulting sample to represent the population. There may be subsets of the population, however, whose risk of arrest is so small that they would not, practically speaking, be represented in the police sample. Celebrities might be an example of one such group. They would either avoid arrest entirely or else the probability of their being arrested would be so small that the resulting prevalence estimate would be too imprecise to be useful. How is this potential "residual" to be handled? Arguably, a subset so immune to arrest is small or otherwise of marginal interest to policymakers. It is not a group that would make heavy demands on the criminal justice system, the publicly financed treatment system, or the public health system. Perhaps from the standpoint of public policy, it is sufficient to estimate the prevalence of hardcore drug users who run an appreciable risk of arrest. Avoiding undercounting If the above argument is not convincing, then undercounting could be corrected by extending the hardcore user methodology. The extension is best explained by way of example. In the example, the current calculation method produces an estimate of 80,000 hardcore drug users in a county in a given year. ADAM data, in this example, indicate that those 80,000 hardcore drug users generate about 20,000 drug treatment admissions per year. A final assumption in the example is that data from local treatment programs show that hardcore users actually account for a higher number of treatment admissions--25,000 per year. This means the estimates of hardcore drug use based on ADAM data understate hardcore drug use by 25,000/20,000 = 1.25. The ADAM-based estimates would be recalculated, with the result 1.25 x 80,000 = 100,000 hardcore drug users in a county. The "missing" hardcore drug users are "found" by looking at a second set of data. This would seem to be a way to avoid a gross undercount of hardcore drug use. Other applications The estimation method has other applications. Variations could be used, for example, to calculate the proportion of the general population with specific infectious diseases who come into contact with the criminal justice system, and to estimate the price of illicit drugs and the amount of money users spend purchasing drugs. The approach would seem to be useful for analyzing illicit drug markets and for estimating the number of career criminals in a population, among other applications. There are applications for which the approach is not suited. It would not be useful for estimating the prevalence of general drug use (that is, hardcore and occasional use combined) in a county. The reason is that for many occasional users of drugs the risk of arrest is negligible, so the numbers would be very imprecise. But for any population that runs an appreciable risk of arrest, this approach to estimation would seem to provide tolerably good estimates of counts of what are otherwise hard-to-study populations. * William Rhodes, Ph.D., is a Principal Scientist and Fellow, and Ryan Kling is a Senior Analyst, at Associates Inc. Notes 1. Developing prevalence estimates in the ADAM population requires both mathematical modeling and the application of statistical sampling theory. To make this chapter accessible to nonstatisticians, a heuristic explanation rather than a mathematically rigorous justification is presented. The latter is available in a separate publication, Rhodes, W., and R. Kling, "Estimating the Prevalence of Hardcore Drug Use Using ADAM Data," final report submitted to the National Institute of Justice by Abt Associates Inc., Winter 2002 (NIJ grant 88-IJ-CX- C001). 2. These last two complications are discussed in the section of this chapter on preliminary estimates. 3. Rhodes, W., and R. Kling, "Estimating the Prevalence of Hardcore Drug Use Using ADAM Data," final report submitted to the National Institute of Justice by Abt Associates Inc., Winter 2002 (NIJ grant 88-IJ-CX-C001). 4. Hardcore drug users in the community are unique individuals. Hardcore users in the booking population are not necessarily unique; that is, for example, 300 arrests could represent 200 arrestees. If 100 hardcore users are arrested once, 100 are arrested twice, and 100 are arrested three times, then the estimation is 300 hardcore users in the booking population but 100 + 200 + 300 = 600 arrests. [Note 4 appears in the footnote of exhibit 9-8, which is not presented in this ASCII plain-text file.] 5. Crime in the United States, 2000: Uniform Crime Reports, Washington, DC: U.S. Department of Justice, Federal Bureau of Investigation, 2001: 233. ---------------------------- Key to Algebraic Terms Used in the Estimation Let Ai represent the estimated rate at which hardcore users in the county, who are represented by the ith hardcore user in the sample of booked individuals, are arrested. (See model discussed in the text.) Let Wi represent the ADAM sampling weight associated with the ith hardcore drug user in the sample. This is derived from the ADAM sampling design. Let P represent the proportion of hardcore drug users who answer truthfully about whether they are hardcore drug users. The assumption is they either tell the truth or deny drug use altogether. Then, the ith hardcore drug user in the ADAM sample for a specified site represents Wi/[A P] hardcore drug users in the county population. In other words, letting H be the represent the estimate of the number of hardcore drug users in the county, then: H = [sigma] Wi/[AiP] The sum above is over the hardcore drug users in the sample. The average arrest rate for hardcore drug users in the community is depicted as A without a subscript: A = [sigma] [Wi/P]/H The above equation is the observed number of arrests divided by the estimated number of hardcore drug users in the community. (The arrest rates of hardcore drug users in all ADAM sites are presented in Exhibit 9-7.) Finally, let G represent the average number of hardcore drug users in the county per hardcore drug users in the arrestee population (see Exhibit 9-8): G = [sigma] Wi(1/Ai)/Wi. ---------------------------- What Do the Estimation Techniques Mean for the ADAM Sites? Most nonstatisticians, and perhaps even many statisticians, would consider the model/technique for estimating hardcore drug use sophisticated. Most statisticians would likely agree that applying it is beyond the ability of anyone not trained in statistics. That raises the question of how the approach can be helpful to the ADAM sites. Although the estimation procedure requires some familiarity with statistics, computing code containing a program to calculate these estimates will be available from NIJ for anyone who requests it. Some ADAM sites may want to use the code to make their own calculations. Researchers may want access to the code to improve it. Most sites are likely not to choose to use the code to replicate or extend the analysis of hardcore drug use. Moreover, it is unnecessary because ADAM will do it for them. As ADAM data accumulate over the years, the model should be reassessed and improvements made following the most recent assessment. Once the improvements reach a certain point, the only remaining challenge would be to develop weights suitable for estimating the number of hardcore users. After they are developed, the ADAM sites could estimate the number of hardcore drug users, along with confidence intervals, using software designed for statistical calculations. The degree of difficulty would be no greater than what is currently required for the sites to produce the more routine weighted estimates of drug use. ---------------------------- Appendix--ADAM Staff: National Institute of Justice and Abt Associates Inc. National Institute of Justice Henry Brownstein Executive Director, Drugs and Crime Division, NIJ Director, ADAM Brownsth@ojp.usdoj.gov Bruce Taylor Deputy Director, ADAM Taylorb@ojp.usdoj.gov Christine Crossland Program and Policy Analyst Crosslan@ojp.usdoj.gov Karen Cyrus Program and Policy Specialist Cyrusk@ojp.usdoj.gov Nora Fitzgerald Social Science Analyst Fitzgera@ojp.usdoj.gov Natalie Lu Drug Testing Technology Specialist Lun@ojp.usdoj.gov Diana Noone Social Science Analyst Nooned@ojp.usdoj.gov Abt Associates Inc. Dana E. Hunt Project Director Phyllis J. Newton Project Manager William Rhodes Technical Adviser Ryan King Senior Analyst Michael Costa Senior Analyst Patrick Johnston Associate Margaret E. Townsend Field Operations Manager Carol Witcher ADAM Data Center Manager Jessica Davis Site Liaison Douglas Fuller Site Liaison Sara Wolf Site Liaison Laura Gandolfo Programmer Sylvia Young Data Center Supervisor Hiwotte Amare Associate Programmer Sara Kuck Research Assistant Jackie Moore Administrative Assistant ---------------------------- About the National Institute of Justice NIJ is the research, development, and evaluation agency of the U.S. Department of Justice. The Institute provides objective, independent, nonpartisan, evidence-based knowledge and tools to enhance the administration of justice and public safety. NIJ's principal authorities are derived from the Omnibus Crime Control and Safe Streets Act of 1968, as amended (see 42 U.S.C. sections 3721-3723). The NIJ Director is appointed by the President and confirmed by the Senate. The Director establishes the Institute's objectives, guided by the priorities of the Office of Justice Programs, the U.S. Department of Justice, and the needs of the field. The Institute actively solicits the views of criminal justice and other professionals and researchers to inform its search for the knowledge and tools to guide policy and practice. Strategic Goals NIJ has seven strategic goals grouped into three categories: Creating relevant knowledge and tools 1. Partner with State and local practitioners and policymakers to identify social science research and technology needs. 2. Create scientific, relevant, and reliable knowledge--with a particular emphasis on terrorism, violent crime, drugs and crime, cost-effectiveness, and community-based efforts--to enhance the administration of justice and public safety. 3. Develop affordable and effective tools and technologies to enhance the administration of justice and public safety. Dissemination 4. Disseminate relevant knowledge and information to practitioners and policymakers in an understandable, timely, and concise manner. 5. Act as an honest broker to identify the information, tools, and technologies that respond to the needs of stakeholders. Agency management 6. Practice fairness and openness in the research and development process. 7. Ensure professionalism, excellence, accountability, cost-effectiveness, and integrity in the management and conduct of NIJ activities and programs. Program Areas In addressing these strategic challenges, the Institute is involved in the following program areas: crime control and prevention, including policing; drugs and crime; justice systems and offender behavior, including corrections; violence and victimization; communications and information technologies; critical incident response; investigative and forensic sciences, including DNA; less-than-lethal technologies; officer protection; education and training technologies; testing and standards; technology assistance to law enforcement and corrections agencies; field testing of promising programs; and international crime control. In addition to sponsoring research and development and technology assistance, NIJ evaluates programs, policies, and technologies. NIJ communicates its research and evaluation findings through conferences and print and electronic media. To find out more about the National Institute of Justice, please contact: National Criminal Justice Reference Service P.O. Box 6000 Rockville, MD 20849-6000 800-851-3420 e-mail: askncjrs@ncjrs.org