Abstract

Studies of treatment samples have long been the primary source of generalizations about drug users, especially for drugs with very low prevalence rates in the general population, such as heroin and crack. Sample selection bias is briefly discussed, and a 1988–1990 study of 699 cocaine users in Miami is described. The drug patterns of the 387 adult crack users interviewed in that study are compared by sample type—residential treatment versus street, controlling for gender. Some similarities between sample types were found, but differences were more numerous. Notably, street respondents started cocaine at a younger age; had used crack regularly for a longer period of time; were more likely to have used pills, heroin, and freebase cocaine; were much more likely to be using crack (but only crack) on a daily basis: and were more likely to obtain crack by being paid in it, especially for drug dealing. Treatment respondents were more likely to use multiple forms of cocaine, to me use cocaine in a binge pattern and with high per-day dosages, and to pay for cocaine with cash they got from a job.