In 1938, Dr. Arlen V. Bock, professor of hygiene and chief of Harvard’s student health services, launched a study of 268 Harvard sophomores (all male, of course), selected as the best and the brightest in the classes of 1939 through 1944. The study was meant to last for fifteen to twenty years, and answer the question of what defines the best health possible, something it was assumed this highly privileged group would exemplify. Financial support came from William T. Grant, owner of a chain of “dime” stores, who had a narrower aim—to find out what makes a good store manager.
Despite the original intention to end the study after two decades, and despite financial trouble after Grant pulled out in 1947, it continues to this day (still known as the Harvard Grant Study, although officially renamed the Harvard Study of Adult Development). The study’s aim has grown broader: to determine which early traits best predict a successful life. Most of the surviving men are in their nineties. It is thus one of the longest prospective studies of adult development ever conducted, and certainly the most exhaustively documented.
Over the course of the men’s lives, a team of investigators collected a vast amount of information about them. They performed thorough physical and psychiatric examinations, IQ tests, and various medical and lab tests, and conducted interviews with parents and, later, with wives and children, as well as repeated interviews with the men themselves. At least every two years, the investigators used lengthy questionnaires to delve into everything from the men’s daydreams to whether they liked their subordinates at work. So frequent and intimate was the contact between investigators and subjects that a bond formed between them, and except through death, there were very few dropouts.
In 1954, when it looked as though the study would end for lack of funding, the Tobacco Industry Research Committee stepped in and provided most of the money for about a decade, ostensibly because it wanted to learn about the “positive reasons” people smoke. (Whether it did is unclear.) After that, funding came from several sources, including the National Institute on Alcohol Abuse and Alcoholism, which required the investigators to pay particular attention to alcoholism. The study is now supported by the National Institute on Aging, for obvious reasons.
During the seventy-five years of the study, there have been only four directors: Clark Heath, MD (1938–1954), Charles McArthur, Ph.D. (1954–1972), George Vaillant, MD (1972–2004), and Robert Waldinger, MD (2005–present). Vaillant’s term was by far the longest, and it is he who has most put his stamp on the study. Now seventy-eight years old, with a string of papers and three books about the study to his credit, he intends his newest book, Triumphs of Experience: The Men of the Harvard Grant Study, to be his final summary of the overall findings.
So what has the study found? Too much, as it turns out. As Vaillant himself acknowledges, there was no hypothesis being tested, so a mountain of information was amassed to no specific end. The study “had been very ambitious in collecting data,” he writes, “but not always very thoughtful about what it would do with the results.” And he refers to “desperate efforts to discover publishable material.” Moreover, computers to organize and analyze the piles of written records were not available until 1969, so until then, the sheer volume would probably resist organization, even if there had been a clear hypothesis. But he compares the study with the Lewis and Clark Expedition and Darwin’s passage on the Beagle, in that “the Grant Study was not a clearly focused experiment but a voyage of discovery (or, as some have less charitably suggested, a fishing trip).”
With no definition of what constitutes a good life and so many possible predictors, it has been difficult to draw unambiguous conclusions, except for a few major ones, which I’ll get to later. Given the mass of data, you could more or less pull out what you were looking for, and that was partly a matter of which theories were in vogue at any given time. During the early years, it was thought that success was largely determined by physical constitution, so one of the criteria used to select men for the study in the first place was a mesomorphic (muscular) body build, which was thought to be characteristic of men destined for success. By the time Vaillant, a psychiatrist, took over the directorship, the emphasis had shifted to the quality of personal relationships, particularly in childhood. Now, with a new director and new technology, there is renewed attention to physical findings—not muscles, but brains, as revealed by imaging studies.
In 2009, in an effort to impose some order, Vaillant attempted to define what constitutes a good life in men aged sixty to eighty years by devising what he called a “decathlon of flourishing.” It consists of ten variables, including being listed in Who’s Who in America, earning income in the study’s top quartile, and being in a good marriage. Using these variables, each man could be given a score ranging from 0 to 10. Vaillant then looked back at the data collected in earlier years to see what factors best correlated with a high score in late life.
It is a brave attempt, but plagued with problems. Many possible correlations were found not to be significant, but because of the small number of men, a real effect might not have met the usual threshold for statistical significance. Vaillant concludes that since the quality of personal relationships (for example, a “warm childhood”) was the strongest predictor of a high score, the “most important influence by far on a flourishing life is love.” That may be true, but it seems to me somewhat circular, since the “decathlon” includes several measures of success in personal relationships (and few measures of, say, feelings of usefulness to society at large, or enjoyment of solitude).
There is nothing ambiguous or circular about some of the findings, however. The men in the Grant Study were extraordinarily long-lived. Fully 30 percent of them survived into their nineties, compared with only 3 to 5 percent of men in the general population of that generation. We have long known that being affluent and well-educated is the best possible ensurer of good health, but the size of the effect in the Grant Study is startling. So strong is the correlation with education that the Grant Study men who went to graduate school lived even longer than those who didn’t.
One of Vaillant’s major themes is that adult development continues long after adolescence. In the Grant Study, the factors associated with flourishing changed with age, and those that were important in youth or middle age were not necessarily important in old age. In fact, many were so inconsistent that I have doubts about how much importance we can attach to them (a consequence of what is known as “data-dredging”). In general, however, these men seemed more content with their lives as they aged (Vaillant does not provide explicit data on this, but he refers to other studies that do), and they reported happier marriages—whether it was a long first marriage or a recent second. Where they landed at about age seventy seemed to matter most. (Vaillant, who candidly admits that his own development over the years greatly influenced his interpretation of the study’s findings, reports, “As I write this book I am recently remarried once again.”)
Another of Vaillant’s themes is the devastating effects of alcoholism. According to him, alcoholism was the cause, not the consequence, of unhappiness in these men. Most of the sixty-two divorces were associated with alcoholism, either in the men or their wives, as were professional setbacks and early death. Vaillant is absolutely certain that alcoholism is, in his words, the horse, not the cart. The men did not drown their sorrows in alcohol, he believes, but inherited a vulnerability to alcohol, which then caused their sorrows. (In my view, he is too accepting of the genetic explanation.) “In all the world literature,” he says, “there is no other study of lifetime alcohol abuse as long and as thorough as this one,” and he believes that this aspect of the study is “perhaps its greatest contribution.”
The main strength of the Grant Study—namely, its long life—is also one of its main weaknesses. These men were dinosaurs, in the sense that the world they inhabited for most of their lives is gone forever. For example, Vaillant found that a “warm childhood” was predictive of success, but that almost surely meant something different in the 1920s than it does today. Parents were stricter then; unlike today’s doting parents, they probably did not end every directive with “OK?,” as in “It’s time for bed, OK?,” and the household probably revolved around the husband, not the children. Marriages were different, too. Although Vaillant does not say, it is likely that nearly all the Grant Study men had housewives—something Harvard men are unlikely to acquire now. Thus, many of the findings are so dated that even when they provide unambiguous conclusions, they are unlikely to apply to people today.
In an appendix to his new book, Vaillant describes two other longitudinal studies of adult development—the Glueck Study of inner-city men and the Terman Study of gifted women. Both offer sharp contrasts to the Grant Study. Sheldon and Eleanor Glueck, well-known criminologists, began in the 1940s to study five hundred white male teenagers in Boston, mainly the sons of Irish and Italian immigrants, who had been sentenced to reformatories for juvenile delinquency. The Gluecks chose for comparison a control group of five hundred similar boys who had no history of delinquency. The average IQ of both groups was 95, as compared with 135 for the college men in the Grant Study, and most had not finished high school.
The Gluecks discontinued their work in 1962 when the boys were in their early thirties, but in 1970, Vaillant was able to include the control group of Glueck men (those who had not been delinquents) as a parallel group in the Grant Study. Since then, the nondelinquent inner-city men have been followed in the same manner as the college men. Their outcomes are much worse. On average, they die about ten years younger than the men in the Grant Study, and their health is worse. But twenty-five of them went on to complete high school and graduate from college, and the health and life expectancy of this subgroup has equaled those of the men in the Grant Study, even though their average IQ was 30 points lower and they attended less prestigious colleges. In Vaillant’s view, “parity of education alone was enough to produce parity in physical health.” But these boys must have been special in some other way that led them to seek higher education despite their disadvantaged backgrounds. Elsewhere, Vaillant points to conscientiousness as a quality associated with success; perhaps that is what these Glueck men had.