5.13/The Lived Worlds/Drugs

What the Numbers Counted

The Mention

The system that produced the mentions was the Drug Abuse Warning Network, running since the early 1970s under the Drug Enforcement Administration and the National Institute on Drug Abuse. It drew on a standing sample of hospital emergency departments and medical examiners’ offices, and clerks reading the records after the fact tallied every drug named in every eligible file. The figure that reached Washington and the press as the national count of drug-related emergencies was, at its base, a count of notations — of what an admitting nurse or a resident had thought worth writing down, gathered by a second reader who had never seen the patient.

Through the crack years the network’s tallies were read as the pulse of the epidemic. A rise in cocaine mentions in a quarter was reported as a rise in cocaine harm. It could as easily register a change in how many hospitals were in the sample that quarter, or a change in what a triage note now routinely recorded, or the same visit counted three times because three drugs appeared on one chart. The instrument’s own designers had documented, before the decade began, that its basic unit failed to match the toxicology four times in five, and the tallies were published and cited anyway, because they were the only national numbers available on a schedule a policymaker or an editor could use.

The Sample of the Arrested

The instrument built to escape the mention’s unreliability replaced the clerk’s notation with the body’s chemistry. Beginning in 1987, the National Institute of Justice’s Drug Use Forecasting program collected urine from people newly booked into local jails, quarterly, at roughly two dozen sites.² Each specimen was screened for ten illicit drugs, and a screen did not depend on what anyone chose to write down — it read what was in the blood over the preceding days.² Against the standard the mention had set, this was hard data, and it was received that way: the arrestee figures were the closest the government came to a direct biological measurement of drug use in a city.

The specimen also settled an argument the program itself had provoked. Each arrestee gave a short interview before giving the urine, and was asked what he had used; the two records did not match. The urine routinely came back positive at rates well above what the same men had just admitted, and the gap was the point of collecting it at all — a person newly in custody had every reason to understate a felony, and only a measure that did not rely on his candor could catch what the interview lost. The program trusted the specimen over the interview by design.² That design carried a lesson the rest of the era’s counting did not absorb: wherever a drug number rested on a person’s own account of using an illegal drug, it rested on the least reliable instrument available, and it ran low for the same reason the surveillance tallies ran high.

The measurement had one population, and it was chosen for it by the police. The specimen came only from a body already in a booking cell, and arrest fell where it could be made — on the hand-to-hand street trade of particular blocks, not on the powder folded into a bathroom stall in an office tower. The most rigorous drug-use instrument the government possessed therefore carried the arrest pattern inside its results: when arrestees’ urine came back overwhelmingly positive for the street drug, the finding described, in the exact language of urinalysis, the chemical composition of the city’s holding cells. The raw arrest counts read as drug-use measures had the same defect at a coarser grain; the urine screen simply lent it the authority of biology. Revamping the program under a new name later in the decade sharpened the sampling. It did not change where the sample came from.

The Household That Answered the Door

The one instrument that reached past the cell and the gurney went to the front door. The National Household Survey on Drug Abuse drew a national sample of homes and asked the people in them what they had used. Its 1991 round found that 37 percent of Americans aged twelve and over — some 75.4 million people — had used marijuana, cocaine, or another illicit drug at least once.³ It was the largest count the era produced, and it came from the one instrument that spoke to ordinary people at their own kitchen tables rather than tallying them at a booking desk or a medical examiner’s slab.

The survey had two limits built into it. By definition it covered only the civilian population living outside institutions, which meant it excluded, by design, everyone in a prison, a hospital, a nursing home, or a drug-treatment center, and everyone homeless who was not in a shelter — the very populations in which the heaviest use gathered.³ And at the doors it did reach, it rested on a stranger telling a government interviewer the truth about a felony. The destitute user was not at the sampled address to answer; the professional who used at a dinner party was at the address and said no. The instrument with the widest reach returned the largest number and still missed the two groups that used most — the user whose life had collapsed out of any fixed address, and the user whose life was arranged so that no official ever asked.

The Record Someone Had to Move Into

Against the instruments, which produced rates and never a recognizable person, stood a kind of record that produced no rate at all. Terry Williams spent close to five years in the mid-1980s watching a teenage cocaine-selling crew, then, from 1987, followed as a nonparticipating observer the daily traffic of a working crack house in a faded townhouse.⁴ Philippe Bourgois moved into East Harlem and lived there roughly six years, from the mid-1980s into the early 1990s, until street-level dealers trusted him enough to let him record them.⁵ What the two produced was the fullest account of drug life the period made in the city — its money, its roles, the reasons the people in it gave for what they did — and it arrived as a pair of books about a few blocks and a few dozen named people, not thousands.

That record could not scale and could not count. It required a single person to spend years in a single place, and it yielded depth with no denominator: a life, richly, and no multiplier to turn one life into a figure for the city. The era’s knowledge of drugs divided in two. The numbers had reach and no context. The ethnographies had context and no reach — people fully drawn, standing for no one but themselves. The quantitative side distrusted the ethnographic one in turn: a study of a few dozen people on a few blocks was filed as anecdote, vivid but unrepresentative, beside a survey of tens of thousands, as though reach and accuracy were the same property — which in the era’s drug figures they were not. Nothing in the period’s apparatus joined the reach of the numbers to the understanding the ethnographies had.

The Interest in the High Number

The direction the numbers took had been named before the crack decade produced any of them. In 1971 Max Singer took a widely repeated estimate of how many heroin addicts New York held and how much they stole to feed the habit, and showed the figure could not survive its own arithmetic: at the addict counts and daily habits the estimate assumed, the stolen property implied ran to several times what the city recorded as stolen in a year.⁶ The number was impossible, and it was in constant use. Once in print, a figure like it gained standing by repetition rather than by test: a later report cited an earlier one, the earlier one a source that had itself borrowed the figure, and the trail, followed to its head, arrived not at a measurement but at someone’s first confident guess. In 1984, as the powder that fed the crack trade was already glutting the market, Peter Reuter generalized the case. Drug statistics, he argued, were “mythical numbers” — produced less by measurement than by a standing demand that the government appear to know a magnitude it had no means to know, and sustained by the plain fact that every institution handling a drug number had an interest in its being large rather than exact.⁷

The interests were not hidden. An enforcement agency budgeted against the size of the threat it named; a treatment system was funded against the size of the need it documented; a newsroom ran a bigger front page off a bigger figure. No institution in the chain was served by a smaller number, and several were served by a larger one. The count of drug use the decade relied on was therefore a political quantity as much as a measurement, and no institution that kept it had any interest in a smaller one.