Further Reading


Reason, December 1991, pp. 63-65.

Asleep at the Switch

Stanton Peele


On August 28, a New York City subway motorman failed to slow at a switch and derailed his train, killing five passengers and injuring another 200. It was the worst New York subway disaster in 63 years. The driver, Robert E. Ray, had a .21 blood alcohol level hours after the crash. According to police, Ray admitted he had been drinking all day before going to work on the night shift and that he was asleep at the throttle when the train hit a switch at many times normal speed. He has been indicted on five counts of second-degree murder.

The day after the accident, the New York Metropolitan Transportation Authority (MTA) announced a program of random drug and alcohol testing for its subway crews and bus drivers. The transit union, Transport Workers Local 100, accepted the program after having strenuously opposed such testing for years. If it hadn't, the union would have risked being seen as insensitive to the danger public transit riders face every day. Yet, notwithstanding the August accident, intoxicated transit drivers may not be a very present danger to New York subway riders. After all, the New York City transit system is immense (including 600 subway trains and 2,900 subway operators), and this accident stands out as unique in recent memory.

Nonetheless, the image of the intoxicated motorman fuels the juggernaut of drug and alcohol testing. Perhaps, New Yorkers fear, the rising drug tide has increased the risk of accidents caused by intoxicated subway operators. Early news reports mentioned that an empty crack vial had been found in the operator's compartment. But Ray tested negative for illegal drugs. And although alcohol abuse has always existed, surprise testing of drivers for drunkenness was not deemed necessary until now.

Is random drug and alcohol testing the best way to prevent horrible,if isolated, public-transit accidents? Actually, MTA policies already in place at the time of Ray's accident permitted some drug and alcohol testing. For example, transit workers could be tested for alcohol if their actions aroused the suspicion of a supervisor. Ray's behavior seemed to justify such testing. Indeed, Ray had given any number of indications that he should not have been driving a subway, that night or at any time:

  • Ray reported for work 15 minutes late, wearing tennis shoes instead of the regulation hard-soled shoes, for which the dispatcher failed to cite him.
  • The dispatcher is responsible for inspecting the fitness of train operators but apparently did not notice Ray was heavily intoxicated. (According to a driver, the dispatcher appeared angry with Ray but took no action.)
  • Ray's record during the previous year was marked with problems, including three suspensions (following two warnings) for lateness, absence, and running a red signal. These violations should have signaled the system to take action.
  • The night of the accident, Ray screeched to stops past station platforms and sped wildly. An off-duty transit worker riding the train said he and the conductor warned Ray about his erratic driving. But the conductor did not radio the transit command station about Ray's recklessness, as he was required to do.
  • Several drivers interviewed by The New York Times said it was common knowledge that Ray drank on the job and reported for work drunk.

In light of this information, will random testing safeguard New York City transit riders? Random tests carried out over such a large work force would not be likely to find a driver who showed up for work intoxicated unless he or she did so regularly. Moreover, definitive tests of drug and alcohol use yield results hours or days after testing, meaning the tests would not prevent operators from driving intoxicated on the occasions when they were tested.

Why are transit authorities in such a hurry to implement a method of monitoring driver intoxication that is less reliable and more expensive than reports from coworkers, examination of previous job performance, and scrutiny from supervisors? Despite the great concern for rider safety the testing program supposedly indicates, the system repeatedly ignored obvious signs that Ray was not performing adequately the night of the accident or in general. Drug testing seems primarily a public-relations effort aimed at alleviating anxiety.

But less dramatic reforms would be far more effective at actually reducing the risk of another accident. The MTA needs to make sure all transit workers understand that drunken subway operators will not be tolerated and that employees are required to report immediately that a motorman is visibly intoxicated or otherwise unfit to drive. Supervisors, dispatchers, and conductors should know that identifying and halting intoxicated or unsafe drivers are imperatives and that they may be fired for not doing so. Meanwhile, transit-authority regulations should be sufficiently rigorous that unreliable drivers are quickly weeded out. Such policies are already on the MTA's books, even though they were not enforced in Ray's case.

Randomly testing all workers for drugs and alcohol appears to be an easier alternative than implementing common-sense organizational procedures. Based on last summer's accident, the MTA and the public are ready to indiscriminately invade the privacy of thousands of drivers in the New York transit system who would never consider working while drunk or on drugs. The Fourth Amendment guarantees that Americans will not be subjected to capricious searches. The Supreme Court has ruled that urine testing is a search by constitutional standards, but that such searches may be justified in the interest of public safety.

In the current environment, it's easy to argue that the clear and present danger of on-the-job drug use outweighs constitutional guarantees. And the most frightening example of such danger is the intoxicated train motorman, pilot, or bus driver. As far as we can tell, however, transit crashes caused by intoxicated drivers and pilots are extremely rare events. According to the FAA, no fatal accidents involving a major commercial airline have ever been traced to a drunken pilot. In 1989, the National Transportation Safety Board ruled for the first time that drug use contributed to a commercial aviation accident—the 1988 crash of a Continental Express plane in Colorado, in which nine people died. But the board did not claim that drug use caused the accident. The levels of cocaine in the pilot's bloodstream were so low that it seemed unlikely that he was intoxicated at the time of the crash.

Thus, aside from constitutional issues, we must address the question of whether drug-testing programs for pilots or drivers address real dangers through means that are cost-effective, sensitive, accurate and timely enough to make a difference. The government's own research suggests that they don't. A 1986 study by the General Accounting Office found drug tests for federal workers costly and ineffective, with false positive rates of from 5 percent to 20 percent.

One alternative to drug testing is to have workers perform motor-skill and hand-eye coordination tasks on a computer before allowing, them to drive. Such "critical tracking tests" avoid the constitutional problems of drug testing and are directly relevant to the task of operating a vehicle. But these tests would be time-consuming, and they could be ignored. Why would MTA supervisors who failed to question an intoxicated and erratic driver be more likely to administer and evaluate critical tracking tests daily to thousands of drivers?

Before searching for high-tech alternatives, the MTA and other transit authorities might try to ensure that employees follow existing procedures for catching accident-prone, intoxicated, and malfeasant drivers like Ray. To its credit, the MTA followed its quick announcement of a random drug-testing program with a more reasoned, organization-wide response a week later. Along with drug testing, the program includes computerizing disciplinary records to flag problem drivers, using film of the Ray train wreck to underline to MTA personnel the importance of reporting intoxicated drivers, reviewing employee selection and training procedures, and installing stop signals before switches. Future aims include developing track sensors to automatically control train speed.

The question remains whether this overall program—which will cost $700 million in the next year—is actually the most cost-effective way to protect rider safety. Expensive components such as track sensors and drug testing must be balanced against the replacement of track and other deteriorating equipment that may ultimately be more central to the safety of the system. And in the final analysis, the overwhelming majority of transit employees around the country accept the responsibility not to drive public conveyances while intoxicated. Using a scatter-shot approach such as random drug testing to catch the few who don't may entail sacrifices that substantially outweigh any benefits.


Drug-testing is now a fact of life in America. "In the United States, screening the urine of employees or job applicants for the presence of drugs has become commonplace. A survey of 794 large- and medium-sized companies conducted...in January 1994 found that 87 percent of them now test applicants for drug use. In 1987, a similar survey found that 22 percent screened job applicants." This growth was due in good part to several highly visible transportation accidents, including the New York subway crash. (D.K. Broadwell, The evolution of workplace drug screening, Journal of Law, Medicine & Ethics, Fall 1994, pp. 240-246.)