THE VALIDITY OF ADULT ARRESTEE SELF-REPORTS OF CRACK COCAINE USE^*

Abstract

Despite the many problems associated with crack use, little validated empirical evidence about the prevalence of crack cocaine exists. Researchers that track crack cocaine use have relied on self-reports to differentiate crack and powder cocaine. Prior research suggests that the accuracy of self-reports for the use of a variety of illicit substances is relatively low. To examine the validity of self-reports of crack use, this article employs a newly developed technology to detect specifically the presence of markers of crack cocaine in urine specimens. With a sample of 2327 arrestees from six cities that participate in the Arrestee Drug Abuse Monitoring (ADAM) Program, both face-to-face interview and urinalysis data were examined. Using a positive urinalysis result as the validity standard, we assessed the extent to which arrestees underreport crack cocaine use as compared to the use of marijuana, opiates, and methamphetamine. Logistic regression models were also developed to predict the factors that relate to underreporting. The results showed a considerable amount of underreporting for all the drug measures. In most cases, only about half the people who had a positive urinalysis test for drugs admitted using drugs. Overall, the least amount of underreporting occurred for the use of marijuana (63.6% told the “truth”), followed by methamphetamine (56.1% told the truth), crack (48.2% told the truth), and opiate (45.9% told the truth). Female crack users, as compared to male crack users, were more likely to admit using crack. Black arrestees were more likely to admit using crack than white or Hispanic arrestees. Arrestees with a history of prior drug treatment or a prior arrest, as compared to those without such histories, were more likely to admit using crack. The older the arrestee was, the more likely the arrestee would admit using crack. The more money an arrestee spent on drugs, the more likely the arrestee would admit using crack. Differences in underreporting were also observed across the six cities in this study. The implications of these findings for the monitoring of crack use are discussed.

• Arrestees
• Crack cocaine
• Drug testing
• Urinalysis
• Validity of self-report

Notes

^*We do not provide concordance statistics for powder cocaine because of measurement problems. The extremely low urine positive rate for powder cocaine (3.6%) may be due to the fact that the people who tested positive for crack cocaine were also using powder cocaine. We believe that the fact that our self-report results show a higher positive rate (5.6%) than our urinalysis results (3.6%) demonstrates a key problem with our powder cocaine measure. For all of our other drug measures (crack, marijuana, opiates, and methamphetamine), the urinalysis results show a higher positive rate than the self-report results. Another problem with our powder cocaine measure is revealed through the sensitivity statistic. Whereas the sensitivity statistic is around 50% for most of the other drugs (crack, marijuana, opiates, and methamphetamine), the sensitivity statistic for powder cocaine is only 15.7% (only 15.7% of the people who tested positive for powder cocaine, according to our testing system, admitted using powder cocaine).

^*One by-product of our poor measurement system for powder cocaine is that we ended up with too few cases to estimate such a model. There were only a small number of arrestees that tested positive for powder cocaine (n = 83), and the sample size was reduced further when we accounted for missing data on the independent variables (n = 69). When we attempted to estimate a logistic regression model for a powder cocaine – only model, it produced estimates that appeared to be very unstable. That is, the overall model fit statistics suggested a very good fit, but there was only one significant independent variable, and there were very high standard errors for the other independent variables.

^†Those arrestees who tested positive for crack cocaine (mean 34 years old) were older than those who tested negative for crack cocaine (mean 29 years old) (F = 87.5, p <. 001), and those arrestees who tested positive for marijuana (mean 27 years old) were younger than those who tested negative for marijuana (mean 33 years old) (F = 185.2, p <. 001).

^*Arrestees with a history of past drug treatment were more likely to test positive for crack cocaine (45.2%) than arrestees without a history of past drug treatment (23.7%) (χ² = 76.5, p <. 001). Arrestees with a history of past drug treatment (35%) were not statistically different from arrestees without a history of past drug treatment (33.1%) in their likelihood to test positive for marijuana (χ² = 0.59, p =. 47). Arrestees with a history of past drug treatment were more likely to test positive for opiates (15.9%) than arrestees without a history of past drug treatment (6.1%) (χ² = 39.8, p <. 001).

^†Those arrestees who tested positive for crack cocaine (mean $403) spent more money on drugs in the past 30 days than those who tested negative for crack cocaine (mean $110.8) (F = 14.5, p <. 001). Those arrestees who tested positive for opiates (mean $446.2) spent more money on drugs in the past 30 days than those who tested negative for opiates (mean $169.9) (F = 4.7, p <. 001). However, there was no difference between testing positive for marijuana and testing negative based on the amount of money spent on drugs in the past 30 days (F = 0.89, p =. 34).