The Accidental Experiment That Changed Men’s Lives

The Vietnam draft lotteries functioned as a randomized experiment—which has allowed social scientists to study its life-changing effects.

Draft Director Curtis W. Tarr spins a Plexiglas drum in 1972 as the fourth annual Selective Service lottery begins.
Draft Director Curtis W. Tarr spins a Plexiglas drum in 1972 as the fourth annual Selective Service lottery begins. (Charles W. Harrity / AP)

Festooned with mustard-yellow drapes and a dangling American flag, the room resembled a grange hall on bingo night. At center stage sat a wide vase containing oblong, plastic lotto balls, and over that vessel stood Representative Alexander Pirnie of New York. As his hand dug into the vase he averted his eyes, like a game-show contestant pulling prizes from a mystery bag. Almost as many U.S. television viewers as had seen the Apollo 11 moon landing a few months earlier were watching him now.

Inside each capsule was a small sheet, to be pulled out like the slip from a fortune cookie. But these small strips did not predict the future; they changed it. Each paper’s inscription scheduled the assignment of what scientists would call a “treatment condition”—an intervention that, from that day onward, would alter the life outcomes its subjects experienced, just as a pill randomly allocated in a pharmaceutical trial might alter a participant’s health. Pirnie would not have thought of his role in these terms, but on December 1, 1969, he was serving as a lab assistant in one of the most significant randomized experiments in history: the Vietnam Selective Service Lotteries.

“The lotteries” not only changed how the Selective Service chose men for the conflict in Vietnam, they also marked a turning point in the history of science. By assigning military induction via an arbitrary factor uncorrelated with personal traits, the lotteries amounted to an experiment.

Yet, unlike most academic experiments, its treatment condition utterly changed individuals’ lives. And, unlike previous draft lotteries, the Vietnam lotteries arrived at a Goldilocks moment in the history of human science. They began just when the systematic collection of data in durable formats had taken root, but before social and behavioral scientists became so enamored with field experiments that excessive efforts to study them degraded their “naturalness.”

Now, 50 years later, the Vietnam draft lotteries have become the drosophila of the social sciences: the model organism for researchers to discern how a life-changing intervention carries implications for the individuals who experienced it, versus those who escaped it by chance.


The first investigation to treat the Vietnam lotteries as an experiment focused on an enduring public concern—the challenges facing veterans upon returning to civilian life. After U.S. troops left Vietnam, stories of veterans suffering difficult returns to civilian life spread widely and were detailed in news reports and dramatized in films such as The Deer Hunter. In the most tragic instances, veterans’ suffering resulted in death. How common was this outcome?

Eleven years after the fall of Saigon, Norman Hearst, Thomas B. Newman, and Stephen B. Hulley used their knowledge of the Selective Service Lotteries to design a study that would answer that question. They could not simply examine the correlation between service in Vietnam and mortality, because serving in the military might correlate with other factors—such as a willingness to take risks—that would independently make individuals more likely to die. Hearst, Newman, and Hulley recognized this problem and knew the solution: a randomized experiment, which assigns treatment (here, to military service) by chance.

The draft lotteries worked in just this way. In each lottery, dates—representing the birthday of draft-eligible men—were randomly paired with the numbers 1 to 365 (or 366 for lotteries covering a leap year). In the first lottery, the succession of birthdates drawn from a vase determined the assigned lottery number—the first date drawn received lottery number 1; the second date, number 2; and so on. In subsequent lotteries, officials improved the randomization by simultaneously drawing numbers and birthdates from different receptacles. The number paired with each birthdate determined the order in which men were called for military induction.

This procedure made those with lower numbers more likely to face military service, not because of any personal attribute likely to be correlated with life outcomes, but because of a random draw of an innocuous attribute unrelated to much of anything—their birthdate. Indeed, that was its intention. The lottery aimed to replace a system that disproportionately forced some individuals into service with a system in which everyone had the same chance of induction. With induction assigned by chance, no correlation should exist between service and inductees’ personal attributes (social class, race, risk tolerance, and so forth).

Recognizing the parallels between the draft lotteries and an experiment, Hearst, Newman, and Hulley began scanning the birthdates of men who died in California and Pennsylvania from 1974 to 1983. The team tallied the number of birthdates called for induction and compared that with the count of birthdates not called for induction. If the draft lotteries did affect death rates, the tallies would differ.

And that’s what they found. In an article published in The New England Journal of Medicine, the team reported a greater frequency of birthdates that had been called for induction among the death certificates. Specifically, in results still relevant to today’s veterans, the team reported that having a draft-selected birthdate increased mortality among draft-eligible men by about 4 percent, including a 13 percent increase in the rate of suicide and an 8 percent increase in the rate of motor-vehicle death.

The social sciences would never be the same—not just because of the results (which were later shown to have faded with time), but because of the method deployed. Hearst, Newman, and Hulley demonstrated that sometimes the experiments required to answer important questions in the social sciences don’t need to be designed; they need to be found.


After discovering the lotteries’ experiment-like qualities, the social sciences realized that studying the lotteries could answer many other challenging questions.

Consider, for instance, the puzzle of how life experiences interact with individuals’ genetic endowments. According to research by Lauren Schmitz and one of us (Conley), being drafted propelled men who were already genetically disposed toward smoking to start doing so. Normally, one cannot randomly assign smoking in a scientific study; each lottery effectively did so because of the greater access to cigarettes that it provided draftees.

More remarkable, the lotteries also shed light on seemingly tangential issues. In 1990, the MIT economist Joshua Angrist became the first to use the draft lottery as an experiment for studying social and economic experiences. In a now-classic study, Angrist found that white men suffered a 15 percent earnings penalty in the 1980s for being drafted, while black men experienced no such disparity. Far from being a story about better outcomes for black veterans, this result provided stark evidence of labor-market bias: If you faced discrimination limiting your civilian job opportunities, losing two years of civilian experience didn’t make a difference. The random nature of the lotteries ruled out health maladies, personal issues, and all other phenomena correlated with both leaving the labor market for multiple years and experiencing lower earnings. Thus, Angrist’s study produced a valid estimate of how exiting the labor market affects earnings.

Angrist’s study accelerated research using the Vietnam draft lotteries and helped launch what came to be called the “causal revolution” in social science—a three-decade, ongoing search for other “natural experiments” to sort out cause and effect. Yet the Vietnam draft lotteries remain the preeminent natural experiment, whose use has extended even beyond the Vietnam generation.

For instance, in 2018, a team of researchers including Matt McGue, William G. Iacono, and two of us—Dawes and Johnson—studied the effect of men’s lottery numbers and military service on the subsequent generation’s decision to enlist in the military. Long before, social scientists had noticed high correlations between the occupational choices of parents and their children. Such correlations could result from various factors—parents’ occupational choices could serve as an example to their kids or highlight a job’s advantages. Alternatively, the connection could be biological: The genetic inheritance of particular skill sets or attributes could make children more likely to end up in the same professions their parents were in. Studying the long-term consequences of the draft lotteries allowed researchers to sort out these possibilities.

Results indicated that the sons of draftees were themselves more likely to enlist in the military. The study’s key insight informed the broader issue of occupational inheritance: If children find themselves in the same job as one that was randomly assigned to their parents, then we know that it’s not the parent’s biological qualities that are responsible for the phenomenon. (Concurrent research by Sarena F. Goodman and Adam M. Isen came to a similar conclusion using a comprehensive data set of all U.S. federal tax returns.)

In political science, researchers studied the lotteries to understand how exposure to public policy influences civic life. Tiffany C. Davenport found that parents whose sons received lottery numbers likely to be called for induction turned out to vote at a higher rate than parents whose sons did not receive such low lottery numbers—an effect that was most pronounced in towns with a war casualty. Jason M. Lindo and Charles Stoecker, two economists interested in antisocial behavior, used the lotteries to show how exposure to violence (as during military service) increases one’s own violent behavior. The business researcher Douglas H. Frank found that draft status influenced individuals’ ascent up the corporate ladder. Across disciplines, the lotteries became a tool to understand puzzles in the social sciences, and resulted in comparable research designs in contexts across the globe.


The Vietnam draft lotteries took place at the cusp of the Information Age, and this timing could account for the wave of research focused on them, as contrasted with previous draft lotteries. The U.S. government had conducted similar drawings for mobilizations during World War I and World War II, but the infrastructure to track “subjects” consisted, at best, of paper forms slotted into cardboard folders. Merging those records, if preserved, with information about the outcomes of men eligible for the earlier draft lotteries would be prohibitively costly. Those lotteries, therefore, are hard to label “experiments”—their consequences can’t be studied. The Vietnam lotteries could, however, because electronic records and databases were appearing for the first time in the 1970s.

But, still, why so much attention to a dated draft lottery when today massive experiments occur regularly? Every minute, tech giants conduct so-called A/B randomizations on their users, despite public aversion to the practice. Are you more likely to click on a banner ad when it’s red or blue? Does that depend on your age? Online content providers know these answers, and occasionally such experimentation even occurs in the name of science. (Think of the much-debated Facebook study on the contagion of sadness in 2014.) These internet experiments do touch the lives of a greater number of people than did the draft lotteries. However, they do so trivially—mainly to affect minor decisions of anonymous web surfers. The Vietnam-era lotteries radically altered the lives of a half decade of male birth cohorts, not to mention their families and friends.

Even compared with other large-scale randomized experiments, the lotteries appear special. Around the same time as the Vietnam draft lotteries, researchers rolled out the Negative Income Tax (NIT) experiment, giving several thousand families randomly assigned, unconditional cash transfers for more than 14 years. The study found that, among households receiving subsidies, unemployment spells lasted longer and marriages broke up at a higher rate.

The NIT doomed Nixon’s ambitious Family Assistance Plan, and it remains impressive, but, like other large experiments, it differed from the lottery experiment in a fundamental way: Experimenters and the subjects consciously recognized it as a study. Few people (if anyone) at the time of the draft lotteries would have realized that they doubled as a science experiment.

Today, such an oversight would not likely occur: A bevy of social scientists would be waiting with bated breath to study an event like the lottery. Speculative blog posts would have alerted lottery “subjects” to possible outcomes from their treatment status—inducing, perhaps, some to react accordingly and bias the experiment. The lotteries’ unintentional and initially unrecognized doubling as an experiment makes it unique.

On the 50th anniversary of the inaugural draw, the Vietnam lotteries deserve recognition as an unprecedented experiment, and, more important, the individuals affected by them deserve to understand their role in that experiment. Customarily, psychologists and social scientists provide a debriefing to experiment participants, with the aim of explaining the reasons behind the ambiguous tasks they performed. In field experiments, the debriefing fulfills a moral obligation, that people should “know what experiment they are in,” as the psychologist Donald Campbell put it. The generation of men who were eligible for the Vietnam lotteries likely did not know they were in an experiment; today, they should. We should stop and consider, for a moment, the vast knowledge gained from the sacrifices and hardships of those who experienced the lotteries: an utterly singular, accidental experiment.

Tim Johnson is the Grace and Elmer Goudy Associate Professor of Public Management & Policy Analysis at Willamette University's Atkinson Graduate School of Management. He directs Willamette's Center for Governance and Public Policy Research.
Dalton Conley is the Henry Putnam University Professor of Sociology at Princeton University and co-author of The Genome Factor.
Christopher T. Dawes is an Associate Professor of Politics in the New York University Wilf Family Department of Politics.