Logistic regression analysis was applied to social indicators from 1990 census data in order to estimate the number of drug users among arrestees in the 185 largest cities in the United States in 1990.
The estimates overall confirm that drug use by arrestees was extensive in 1990, requiring policy strategies for intervention. In the 185 cities included in the estimate, approximately 925,000 arrestees used cocaine, 317,000 used opiates, 213,000 used amphetamines, 389,000 injected drugs, and 1,296,000 used an illicit drug. The estimation method proved to be cost-efficient. The two sources of data on drug users involved with the criminal justice system were the Drug Use Forecasting program and the Uniform Crime Reports. Several social indicators of environmental characteristics of communities were available from the 1990 U.S. Census. The Drug Use Forecasting program employs quarterly interviews conducted at local jails with adult men and women within 3 days after their arrests. Urine specimens are collected and tested for 10 drugs. The Federal Bureau of Investigation's Uniform Crime Division maintains a national data system of arrests reported by jurisdictions throughout the Nation, Reported arrests accounted for 98.2 percent of the weighted total arrests. The comprehensiveness of Uniform Crime Reports provides the basis for the estimation of the total arrestee population. Five city-level indicators were extracted from 1990 U.S. census data: overall population size, poverty, unemployment, high school graduates, and youth between the ages of 16 and 24. Separate logistic regressions were conducted for each of the dependent variables: the urine positive rates for cocaine, opiates, amphetamines, injection drug use, and any drug use involving each of the eight subgroups by the cross combination of gender and offense type. Predictors were five social indicators from census data hypothesized to be linked with drug-use levels in different cities. 3 tables and 15 references
Downloads
Similar Publications
- Evaluating the Robustness and Ruggedness of a Statistical Model for Comparison of Mass Spectral Data for Seized Drug Identification
- Evaluation and Classification of Fentanyl Related Compounds using EC-SERS and Machine Learning
- Investigative Decision-making in Public Corruption Cases: Factors Influencing Case Outcomes