What underlies the broad acceptance of a practice that conservative Supreme Court Justice Antonin Scalia has called a "needless indignity"? One factor may be the alarming statistics cited by testing advocates to demonstrate the high costs of drug abuse. Examination of some of these claims suggests they do not always accurately reflect the research on which they are based. In fact, some of the data could be used to "prove" that drug use has negligible or even beneficial effects. Consider these examples.

Last year President George Bush declared that "drug abuse among American workers costs businesses anywhere from $60 billion to $100 billion a year in lost productivity, absenteeism, drug-related accidents, medical claims and theft." Variants of this statistic abound in discussions about drug abuse and are commonly repeated without qualification by the media. Yet all such claims derive from a single study, one that "was based upon assumptions which need additional validation," according to an assessment last year by NIDA, the chief federal agency sponsoring research on substance abuse.

The study grew out of a survey of some 3,700 households by the Research Triangle Institute (RTI) in 1982. The RTI group found that the average reported income of households with at least one person who admitted to having ever used marijuana daily (20 days or more in a 30-day period) was 28 percent lower than the average reported income of otherwise similar households. The RTI researchers defined that difference in income as "loss due to marijuana use"; the total loss, when extrapolated to the general population, came to $26 billion. The researchers then added on the estimated costs of drug-related crime, health problems and accidents to arrive at a grand total of $47 billion for "costs to society of drug abuse." This figure - "adjusted" to account for inflation and population increase - represents the basis of Bush's statement, according to Henrick J. Harwood, who headed the RTI study and is now in the White House drug-policy office.

The RTI survey included questions on current drug use (at least once within the past month). Yet according to Harwood there was no significant difference between the income of households with current users of any illegal drug - including marijuana, cocaine and heroin - and the income of otherwise similar households. Does this mean that current use of even hard drugs - as opposed to perhaps a single marijuana binge in the distant past - does not lead to any "loss"? "You would be on safe ground saying that," Harwood replies.

Officials of the U.S. Chamber of Commerce have testified before Congress and at national conferences on drug abuse that employees who use drugs are "3.6 times more likely to injure themselves or another person in a workplace accident ... [and] five times more likely to file a workers' compensation claim." The pharmaceutical giant Hoffmann-La Roche, which is leading an anti-drug campaign among businesses (and has a big share of the drug-testing market), also promulgates this claim in "educational" literature.

In fact, the study on which the claim is based has "nothing to do with [illegal] drug users," according to a 1988 article in the University of Kansas Law Review by John P. Morgan of the City University of New York Medical School. Morgan, an authority on drug testing, has traced the Chamber of Commerce claim to an informal study by the Firestone Tire and Rubber Company of employees undergoing treatment for alcoholism.

In an interview with SCIENTIFIC AMERICAN, J. Michael Walsh, who heads NIDA's applied research division and is a strong supporter of workplace testing, singled out two studies that he said showed drug users are more likely to cause accidents, miss work and use health benefits. The studies were done at two utilities: the Utah Power and Light Company and the Georgia Power Company. The 12 workers in Utah and the 116 in Georgia who served as the primary research subjects were tested "for cause": they had either been involved in accidents, exhibited other "problem" behavior (commonly, high absenteeism) or submitted to treatment for alcoholism or drug abuse. Critics point out that it should not be terribly surprising if these subjects exhibited the cited traits at a higher-than-average rate.

What may be surprising is that, according to a report published by NIDA last year, Utah Power and Light actually "spent $215 per employee per year less on the drug abusers in health insurance benefits than on the control group." Those who tested positive at Georgia Power had a higher promotion rate than the company average. Moreover, Georgia workers testing positive only for marijuana (about 35 percent of all positives) exhibited absenteeism some 30 percent lower than average. Nationwide, Morgan says, marijuana accounts for up to 90 percent of all positive findings, both because it is by far the most widely used illegal drug and because it persists in urine for up to a month (compared with two days for most other drugs).

Perhaps the study most publicized of late by testing proponents involves employees of the U.S. Postal Service. The service tested 4,396 new hirees in 1987 and 1988 and - keeping the test results confidential - tracked the performance of positives (9 percent of the total) and negatives. By last September, the service reported, 15.4 percent of the positives and 10.5 percent of the negatives had been fired; the positives had also taken an average of six more sick days a year.

This study may be distorted by more subtle biases - related to race, age or gender - than those displayed by the utility studies, according to Theodore H. Rosen, a psychologist and a consultant on drug testing. Indeed, Jacques L. Normand, who headed the study, acknowledges that minority postal workers tested positive at a much higher rate than nonminority workers and that previous studies have shown minorities to have higher absenteeism.

Morgan points out, moreover, that the Postal Service study (like all those cited above) has not been published in a peer-reviewed journal. In fact, he says, only one study comparing the work of drug-test positives and negatives has passed peer review. Last year, in the Journal of General Internal Medicine, David C. Parish of the Mercer University School of Medicine in Georgia reported on a study of 180 hospital employees, 22 of whom had tested positive after being hired. Parish examined supervisor evaluations and other indexes and found "no difference between drug-positive and drug-negative employees" at the end of one year. He noted, however, that 11 of the negatives had been fired during that period and none of the positives.

Proponents of testing often imply that drug use among workers is growing. A Hoffman-La Roche brochure, for example, quotes Walsh pronouncing that "the problem of drug abuse has become so widespread in America that every company must assume that its employees will eventually be faced with a substance abuse problem." Yet in 1989 NIDA reported that illegal drug use has been decreasing for 10 years and that the decline has accelerated over the last five years. From 1985 to 1988 the number of current users (at least once in the last month) of marijuana and cocaine dropped by 33 and 50 percent, respectively.

To be sure, a subset of this group of current users is increasing: NIDA estimated that from 1985 to 1988 the number of people using cocaine at least once a week rose from 647,000 to 862,000 and daily users increased from 246,000 to 292,000. NIDA found that addiction to cocaine (including "crack") is particularly severe among the unemployed - who are beyond the reach of workplace testing.