In the courts, the harm reduction rationale has generally trumped the use reduction rationale. For example, in Vernonia, the Court held that the importance of deterring drug use among schoolchildren “can hardly be doubted.”But the Court also focused on the harm reduction benefits of use testing: “[I]t must not be lost sight of that this program is directed more narrowly to drug use by school athletes, where the risk of immediate physical harm to the drug user or those with whom he is playing his sport is particularly high.”The D.C. Circuit has ruled that random testing is an unreasonable invasion of employee privacy except for safety-sensitive positions.Based on its reading of three Supreme Court decisions,the Substance Abuse and Mental Health Services Administration has identified four classes of presumptive testing— employees who carry firearms, motor vehicle operators carrying passengers, aviation flight crew members and air traffic controllers, and railroad operating crews— “that are to be included in every plan if such positions exist in the agency.”The National Research Council took a comprehensive look at the evidence for a safety-promoting benefit of drug testing in the workplace. A potential objection to this causal chain model is that drug use might have an additional indirect association with accident risk through some common cause, such as poor self-control skills. The NRC committee noted that any observed link between drug use and accidents or work behavior could be spurious,grow cannabis due to common causation by a third variable. The committee offered this hypothesis: “[D]eviance may be a better explanation than impairment of the links between alcohol and other drug use and undesirable work behavior.

If so, confronting deviant behaviors and attitudes may be a more effective strategy than narrow antidrug programs for both preventing workplace decrements and treating poorly performing workers.”From a prediction standpoint, one might argue that drug tests can serve as a double proxy for both drug use and low self-control. But psychometrically, a better strategy would be to directly assess low self-control and psychomotor functioning, as illustrated in Figure 2. Psychologists and ergonomic specialists have developed a wide variety of valid psychomotor tests, and many are already in use in the military and other “mission-critical” organizations.The private sector has also begun to recognize the potential advantages of directly testing impaired psychomotor performance.There are a variety of psychometrically reliable and valid measures of impulsivity, sensation-seeking,and self-control.More controversially, there are paper-and-pencil “integrity tests” that allow corporations to assess drug and alcohol use, honesty, and other behavioral factors.Psychomotor testing and integrity testing need not replace drug testing; they can complement it. They may be less intrusive81 and, in the case of psychomotor testing, more diagnostic of accidents. Of course, psychomotor testing will pick up impaired performance due to factors other than drug use—alcohol consumption, age, lack of sleep, depression, and so on. Some of these other factors are preventable. Hence, a program of random psychomotor testing may well be an effective deterrent against drug use, but also against alcohol use, sleep deprivation, and other factors that impair safety. At the same time, not everyone who is using a psychoactive drug will show impaired performance on such tests. The NRC notes that “many employees who do work under the influence may be able to compensate for their impairment, and there is a substantial amount of variation across individuals as to how a specific drug at a given dose affects performance.

”As discussed below, this illustrates the tension between the “criminal deviance” and “safety regulation” framings of the problem. It also calls into question the relative importance of the stated motives for use testing: deterring drug use and preventing accidents. A preference for drug testing over psychomotor testing suggests that use testing is really about drug control rather than safety. This is also shown by the fact that drug testing is more common than alcohol testing,even though the link between alcohol and accidents is better established.Granted, it may be easier to consume alcohol without intoxication than cannabis or other drugs.And alcohol is far more prevalent, meaning far more positive test results—though from a safety perspective that is not much of an argument at all. Use testing may also have some unintended consequences. Theoretically, it could encourage users to substitute less detectable intoxicants. In 1995, 20% of worksites tested for illicit drugs but not for alcohol.The most commonly tested substances are the NIDA-5: marijuana, cocaine, PCP, opiates, and amphetamines. Thus, users might shift from the NIDA-5 to other illicit drugs like MDMA and barbiturates, and from illicit drugs to alcohol. I am unaware of studies examining such substitution effects,but these effects have been linked to other policies. There is some evidence that users substitute marijuana for hard drugs when marijuana is decriminalized,and that users substitute marijuana for alcohol when the legal drinking age is raised or beer prices increase.Because marijuana has the longest window of detectability in urinalysis, one might see a shift toward less readily detectable substances like MDMA, amphetamines, and barbiturates.

A related concern is that use testing will drive users away from testing organizations—workplaces, schools, sports teams, and the military. This might make those organizations safer, but it displaces the harm to other settings where use might even escalate. A similar argument is suggested by the “labeling theory” tradition in criminology.Labeling theory predicts that legal controls can actually enhance the likelihood of future offenses if the stigma associated with criminal sanctioning alienates the individual from conventional society. Alienation encourages contact with criminally involved referent groups, and weakens the reputational costs that may restrain deviance— thus creating a self-fulfilling prophecy. Several lines of evidence support this prediction,indoor cannabis grow system but the results are not conclusive. Neither Mehay and Pacula nor Bachman found any evidence linking pastdrug use to self-selection into the military.On the other hand, using the 1994 NHSDA survey, John Hoffmann and Cindy Larison of the National Opinion Research Center found that those using marijuana or cocaine at least weekly were more likely to work for companies that had no testing program.And the NHSDA 1994/1997 workplace analysis suggested that current users were more likely than nonusers to say they might avoid working for an employer who conducts pre-employment screening , random drug testing , or “for cause” testing.Even in its heyday, safety testing of illicit drugs was extremely rare. Between roughly 1972 and 1984, safety testing was done by numerous independent local laboratories run by universities, nonprofits, and health centers.These laboratories tested anonymous samples dropped off at street locations or submitted through the mail. The samples involved a wide variety of alleged substances, including cannabis products, amphetamines, barbiturates, opiates, and various psychedelic drugs. Although they were scattered around the country, relatively few communities had such a center, and the utilization of national services was sparse. For example, PharmChem’s national testing program—the largest such program in the 1970s—analyzed a total of 10,778 samples alleged to be cocaine between 1973 and 1983.97 In 1982, PharmChem’s busiest year of cocaine testing, they received 1385 samples.But there were at least 3 million U.S. cocaine users in 1973, and at least 12 million in 1983.Under the most optimistic assumption that each sample came from a different user, only 0.012% of all users participated in their testing. Even if PharmChem accounted for only 1% of the national market for street testing—almost certainly far below their actual share—that would still imply that only 1% of all users had samples tested that year.

The picture is similar for the late 1990s and early 2000s. DanceSafe is the major source of samples for the EcstasyData.org testing operation.100 EcstasyData.org tested only 1521 samples alleged to be MDMA between 1996 and 2006.To put this in perspective, in 2001 an estimated 3.2 million Americans used MDMA at least once102— 1.7 million of them for the first time.103 In that year, DanceSafe tested only 332 samples, which accounted for at most 0.01% of users, and this time their operation nearly cornered the market. At present, use testing is far more common than safety testing. While safety testing may have an important impact on the lives of those who submit samples, they account for only a negligible fraction of users. Thus, any aggregate impact of safety testing must be due to the diffusion of this information and its use by rave organizers and harm-reduction activists. The low prevalence of safety testing is not difficult to explain. The legal risk to participants is the most obvious factor,but there are others. Volunteering a useable sample means giving away a valuable commodity. And the test results, once publicized, are a public good, and hence subject to free riding by nonparticipants. Another consideration is the high cost. Most of the 1970s testing programs appear to have collapsed due to loss of funding rather than legal intervention, and few users can or will pay the high cost of screening.At an aggregate level, test results would seem to be less accurate for safety testing than for use testing, at least for random use testing. Because safety testing is voluntary, there is no coercion or threat to civil liberty, but the samples are also unlikely to be statistically representative. The direction of any selection bias is hard to identify. Safety testing volunteers may be more cautious, wealthier, or better educated, but their samples may disproportionately represent the results of suspicious transactions and dealers. Still, tests of drugs seized by law enforcement agents often show high levels of impurity, despite a very different set of sampling biases.The remarkably low and variable purity rates in the safety testing data have implications for the interpretation of use testing data, as well other sources of drug indicator data. Typically, use testing targets a specific set of illicit drugs and does not attempt to detect or identify adulterants. Because the samples are not volunteered as drugs or labeled with street names, use testing samples may test negative even when the source was using street drugs. As noted above, such false negatives will occur due to the presence of nontested street drugs, or because the critical sampling periods of the target drugs have passed.But the safety testing data suggest that false negatives will also occur because tested individuals who were trying to use a NIDA-5 drug unwittingly used something else. On the other hand, the Drug Abuse Warning Network annual series,which records emergency room “drug mentions,” may overstate the link between the mentioned drugs and acute health crises, because someone who mentions a drug may have actually consumed something entirely different. To date, I have not been able to locate any empirical study of the effects of safety testing on levels of drug use. This is hardly surprising; safety testing has always been rare and research on safety testing is even rarer.Moreover, safety testing is not intended to influence the prevalence of drug use per se; it is intended to prevent harmful consequences and make users more cautious about their behavior. Still, there are good reasons to consider the question. From a hawkish perspective, one may reasonably ask whether safety testing encourages drug use, either wittingly or unwittingly. But it is possible that consumers infer tacit messages from DanceSafe and related organizations. Psycholinguistic theory and research suggest that people readily draw additional inferences that are pragmatically implied by an actor’s conduct, regardless of whether those inferences were endorsed, or even intended, by the actor.The very way that test results are framed implies that safety testing treats drug use in a less stigmatizing way than use testing does. In safety testing, a positive test is pure, and a negative test denotes failure and contamination. In use testing, it is the positive test that connotes failure; the user is the contaminant. Second, safety testing may encourage use by changing perceptions of risk. At the margin, a harm reduction mechanism might change a person’s assessment of the expected value of taking drugs. If an intervention reduces harm, then at the margin it should increase the attractiveness of the activity for most people. In my earlier treatment of this topic, I reviewed evidence of this mechanism, much of it appearing under the labels “compensatory behavior,” “offsetting behavior,” or “risk homeostasis.”For example, there is strong evidence that people drive faster when they have seat belts and airbags, both in econometric analyses and in controlled experiments on driving test tracks.