The longest-running study of human happiness ever conducted has been tracking the same group of Harvard students since 1938, when they were teenagers — and after nearly nine decades of interviews and medical records, the single factor most closely tied to a good life is one almost nobody would guess

The 724 teenage boys who agreed to be studied for the rest of their lives in 1938 did not know that the single most reliable predictor of their eventual happiness would turn out to be something none of them had heard mentioned in the initial recruitment interviews. The Harvard sophomores had been selected on the assumption that intellectual ability, family background, and physical constitution would be the substantive determinants of successful adult outcomes — the underlying research hypothesis of the study’s first director, Clark Heath, reflected the biological-deterministic view of human development that dominated late-1930s American psychology. The Boston teenagers, recruited in parallel by the criminologists Sheldon and Eleanor Glueck, had been selected as a matched control group to a study of juvenile delinquency, on the assumption that early social disadvantage would produce specific adult outcomes that a longitudinal study could measure. Neither cohort was told that they were being enrolled in what would eventually become an 88-year investigation of the underlying structure of human wellbeing. Neither cohort’s initial recruitment materials mentioned the possibility that the study’s core finding, when it was finally synthesised across the accumulated eight decades of interviews, medical records, brain scans, marriage certificates, career histories, autopsies, and eventually the interviews of their children and grandchildren, would be that the quality of a person’s close relationships — specifically the depth and warmth of their emotional connections to spouses, family members, friends, and broader community — would matter substantially more, for essentially every measurable outcome the study tracked, than intelligence, family wealth, physical constitution, career achievement, or any other variable the original 1938 investigators had considered.

According to the Harvard Gazette’s 2023 summary of the study’s cumulative findings and the current director Robert Waldinger’s synthesis of them in his co-authored book The Good Life, the core substantive finding of the study — refined across 88 years of longitudinal data and now supported by essentially every measurement the researchers have collected — is that people who have warm, close relationships in their 50s are, at the age of 80, substantially healthier, cognitively sharper, and more emotionally stable than people who reported difficult or absent relationships at the same earlier ages. The relationship-quality finding predicts subsequent physical health better than cholesterol levels do. It predicts subsequent cognitive decline better than IQ does. It predicts subsequent life satisfaction better than income does. It predicts mortality — controlling for the standard medical risk factors — better than smoking status does. And, most substantively, it holds across both original cohorts of the study: the Harvard sophomores from wealthy backgrounds and the Boston teenagers from the poorest neighbourhoods of the city produced statistically indistinguishable relationship-vs-outcome curves, meaning that the finding was not an artefact of privilege or of educational background but a substantive feature of how human wellbeing appears to work.

What the researchers thought they were measuring

The gap between what the study set out to measure in 1938 and what it eventually discovered was, in essential respects, the accidental structure of the study’s own methodology. As detailed in the John Templeton Foundation’s 2023 interview with study director Robert Waldinger on the Harvard Study of Adult Development’s evolution across its 85-year history, the original research team collected data on essentially every dimension of participant life that seemed at the time to be relevant to adult flourishing — physical measurements, cognitive testing, personality assessment, family relationships, career progression, income, marital satisfaction, alcohol consumption, exercise patterns, medical history — without knowing which of these variables would eventually turn out to matter and which would not. The study’s fourth director, Robert Waldinger, took over in 2005 and has spent the subsequent two decades working with his colleague Marc Schulz to synthesise the accumulated data into the findings that the study’s Grant/Glueck predecessors had been assembling for six decades without the analytical framework to fully interpret. The 2015 TED Talk in which Waldinger presented the core finding — that “good relationships keep us happier and healthier” — has been viewed more than 44 million times, making it one of the ten most-watched TED talks ever recorded.

The specific statistical relationships that support the finding are, by any reasonable measure of longitudinal research methodology, substantial. Per the Harvard T.H. Chan School of Public Health’s summary of Waldinger’s 2023 discussion of the study’s findings, participants’ satisfaction with their relationships at age 50 predicted their physical health at age 80 more accurately than their cholesterol levels at age 50 did. Participants who reported warm connections in their 50s exhibited slower cognitive decline in their 70s and 80s. Participants who reported loneliness or social isolation in mid-life experienced earlier onset of chronic disease and shorter overall lifespans than the socially connected members of both cohorts. The finding did not require the relationships to be uniformly pleasant — the study documented substantial numbers of long-married couples who bickered essentially continuously across their entire lives together and who nonetheless benefitted from the underlying emotional connection between them. The distinction that mattered was between substantive emotional attachment and its absence, not between harmonious relationships and difficult ones.

What the finding actually says

The substantive implication of the 88-year Harvard result, set against the broader landscape of contemporary Western culture, is one that Waldinger and Schulz have described as “quietly countercultural.” As reported in the Harvard Gazette’s 2017 profile of the study on its 80th anniversary, the modern American cultural emphasis on individual achievement, career advancement, financial success, physical fitness, and personal optimisation as the substantive determinants of a good life is, on the accumulated evidence of the Harvard Study, substantially misaligned with what the data actually shows. The 724 men whose lives were tracked from 1938 onward pursued essentially every combination of these goals across the subsequent decades. Some became wealthy. Some became famous — John F. Kennedy was among the original Harvard sophomore cohort, though his individual data has been kept sealed for privacy reasons. Some became distinguished academics, physicians, lawyers, business executives. Some remained in the working-class neighborhoods where they had grown up. The consistent finding across all of these trajectories was that the participants who reported strong, warm relationships in mid-life — regardless of their income level, educational attainment, or social status — were, at the end of their lives, both happier and healthier than the participants who had achieved essentially any other measurable form of external success without also maintaining substantive emotional connections to other human beings. The 88-year finding is, in essential respects, that the specific thing modern American culture spends the most collective effort optimising for is not the specific thing that, on the available longitudinal evidence, actually matters most.