The longest study of human happiness ever run has followed the same men for more than 80 years — through war, careers, marriage and old age — and its director says the clearest predictor of who stayed healthy and content was not money or fame but the quality of their relationships

Since 1938, a group of researchers at Harvard has followed the same set of men through their entire adult lives. The project now known as the Harvard Study of Adult Development follows two original male cohorts: 268 Harvard College men from the Grant Study and 456 Boston men from the Glueck Study, together usually described as 724 participants. The researchers tracked their work, their marriages, their war service, their health, and eventually their deaths, using questionnaires every two years, with periodic health records and longer interviews across the decades. Its current director, the psychiatrist Robert Waldinger, summarises the central result in five words. Good relationships keep us happier and healthier. Not money. Not fame. The quality of a person’s relationships.

We are writers, not clinicians, and what follows is a reading of the research rather than health advice. We say that early because the finding is appealing, and appealing findings are the ones most worth slowing down for. This one is probably, in its broad shape, right. That is exactly why it is worth being precise about what it does and does not establish.

It is also worth saying plainly that this is one study. A long one, an unusually careful one, but a single longitudinal design following a particular set of people through a particular century. The specific claims that come out of it belong to this dataset before they belong to everyone.

What the study actually measured

The headline number that gets quoted most often is this: in the Harvard data, how satisfied a man was with his relationships at age 50 predicted his physical health at age 80 better than his cholesterol did. Waldinger has put it that the people most satisfied in their relationships in midlife were the healthiest three decades later, a result the study reports as holding across both cohorts, the Harvard men and the Boston men alike.

Notice the scope built into that sentence. It is a prediction, within this sample, of one outcome, health at 80, from one measure, relationship satisfaction at 50, across one span of time. That is a real and interesting finding. It is not the same statement as “relationships matter more than cholesterol for your health,” which drops the dataset, the timeframe, and the specific outcome, and turns a result into a slogan.

The study measured a lot besides. Waldinger has been careful in interviews to say the project is really about wellbeing across the big domains of a life: mental health, physical health, work, and relationships. The relationship result stands out partly because it was a surprise to the researchers, who did not set out in 1938 expecting warmth and connection to be the thread that ran through who aged well.

Why the finding is more than one study’s quirk

If the Harvard study sat alone, the careful reading would be to file it as suggestive and wait. It does not sit alone. The link between social connection and physical health turns up across a wide body of research that has nothing to do with Harvard sophomores.

The clearest example is the work of the psychologist Julianne Holt-Lunstad. Her 2010 meta-analysis in the journal PLoS Medicine pooled data from 148 separate studies and found that people with stronger social relationships had substantially better odds of surviving a given follow-up period, an effect she and her colleagues placed in the same rough territory as well-established health risks. Later work by the same researcher extended the pattern to social isolation and loneliness specifically.

This is the part that justifies taking the Harvard finding seriously rather than treating it as a charming story about one cohort. The direction of the result, that connection tracks health and longevity, is one of the better-replicated patterns in this corner of research. The Harvard study did not discover it alone, but its eighty years of follow-up give the pattern a depth that shorter studies cannot.

The easy misreadings

The trouble starts when a robust correlation gets handled as a simple cause you can operate. The study is observational. It watched lives unfold; it did not assign anyone to a warm marriage or a lonely one. That leaves the direction of the arrow genuinely tangled. Warm relationships may protect health. It is also true that people who are healthier, steadier, and less burdened tend to find and keep good relationships more easily, which would produce the same correlation without the relationships doing the causal work. Waldinger himself has called the mechanism behind the effect the open question of the whole project.

The second misreading is the money one. “Close relationships, more than money or fame” is a fair description of what differentiated outcomes inside this study, which was not designed to test the health effects of poverty, housing insecurity, or material deprivation directly. It is not a finding that material security is irrelevant to health. For people in real poverty, money buys medical care, housing, and food, and the absence of it harms the body directly. The study’s point is better read as: once the basics are reasonably met, connection is what separates the people who flourish from the people who do not. That is a narrower and more honest claim than “money doesn’t matter.”

The third is the self-help turn, in which a population pattern becomes a personal instruction. The study describes what tends to be true across many lives over many years. It cannot tell any individual reader what is wrong, or hand them a five-step plan. The most the data supports is that relationships are worth treating as something you tend to rather than assume, which Waldinger has framed as a kind of social fitness. That is a useful reframe. It is not a prescription.

Who the original men were

The sample is the limitation that gets mentioned least and matters most. The original 724 were all men, and the first generation was white. The Harvard group were privileged, selected at one university in the late 1930s. The Boston group were disadvantaged, but still male, still from one city, still shaped by one narrow slice of the twentieth century. Women entered the study only later, through the wives of the original men and, more recently, through more than a thousand descendants now being followed as a second generation.

This does not void the finding. The relationship pattern showed up in two very different groups within the study, and it lines up with other research drawn from far broader populations. But it does mean that any sentence beginning “the study proves that people…” is reaching past what an all-male, two-cohort, single-era design can carry on its own. The honest version keeps the result and keeps the caveat in the same breath.

What is worth keeping

Strip away the overstatement and a quieter claim survives, and it is the more useful one. Across this study and a wider literature, the quality of a person’s close relationships is one of the few things that both tracks long-term wellbeing and sits partly within reach. You cannot easily change your genes, your childhood, or your luck. You can, with effort, pay more attention to the people already in your life. That is probably why the finding has travelled so far. It offers something to do.

It is worth holding that lightly. A long study can name a pattern that holds across hundreds of lives. It cannot deliver a verdict on any single one, and a hard stretch of loneliness is a circumstance, not a sentence. People rebuild their connections at every age the study has ever measured.

The Harvard project is now in its ninth decade and still running, with the children of the original participants supplying the next long arc of data. The interesting question for the years ahead is not whether relationships matter, which the study has illuminated as well as a long observational study can. It is whether the pattern holds as steadily for the women and the more varied lives now entering the data as it did for a few hundred men who started filling in questionnaires before the Second World War.</p>