PsychOS
The Harvard Study of Adult Development: How 85 Years of Tracking the Same People Revealed What Actually Predicts a Good Life
In 1938, researchers began tracking 724 teenage boys through every decade of their adult lives. Most of the men have since died. The study hasn't. It's now the longest-running study of human life e...
Explore PsychOS
PsychOS combines psychology education with structured self-reflection tools. Explore assessments, articles, experiments, and related resources while treating results as educational signals rather than diagnosis.
In 1938, a group of Harvard researchers picked 268 sophomores, mostly from privileged backgrounds, and started asking them questions: about their health, their relationships, their ambitions, their fears. Around the same time, a separate team led by Sheldon and Eleanor Glueck began tracking 456 boys from disadvantaged Boston neighborhoods, many from families struggling with poverty. Neither team could have known they were starting something that would still be running eight decades later, long after every original researcher, and most of the original participants, had died.
The two studies eventually merged under Harvard's continuing direction, and the tracking never stopped. Researchers followed these men through World War II, through careers and divorces and retirements, through the slow accumulation of an entire life, and eventually began tracking their children too. What emerged wasn't a single dramatic result. It was something arguably more useful: a genuine, decades-deep answer to a question most research is too short-term to actually address — what predicts whether a person ends up thriving by the time they're old.
THE HYPOTHESIS
Unlike a controlled experiment testing a specific manipulated variable, the Harvard Study of Adult Development, formally the Grant and Glueck Study, was built around a broader, more open-ended research question: what factors, measured across an entire adult lifespan, actually predict physical health, psychological well-being, and life satisfaction in old age. Founding director Arlie Bock's original hypothesis reflected the medical thinking of his era, focused heavily on physical constitution, habits, and personality traits measured in early adulthood as likely predictors of later outcomes.
This is a genuinely different kind of research design than a classic experiment, and it's worth being direct about that distinction. There's no manipulated variable and no control group in the traditional sense — this is a naturalistic, longitudinal observational study, which means its findings describe strong, well-documented associations rather than experimentally proven causation. What the design does offer, uniquely, is decades of real data on the same individuals, rather than a single snapshot or a short-term follow-up, which is precisely what allows it to speak to genuinely long-term outcomes in a way almost no other study in psychology can.
THE METHODOLOGY
Participants across both original cohorts were interviewed and given physical exams at regular intervals, initially every two years, gathering data on physical health, career trajectory, marriage and family life, mental health, social relationships, and general life satisfaction. Over time, the data collection expanded considerably: brain scans in participants' later decades, detailed interviews with spouses and adult children, and, as George Vaillant took over direction of the study in the 1960s and led it for several decades, an increasingly sophisticated focus on psychological and relational variables rather than purely physical ones.
Under Vaillant's direction, and later under Robert Waldinger's, researchers began systematically coding decades of accumulated data for specific variables: quality and stability of close relationships, patterns of emotional coping, career satisfaction, and physical health markers tracked longitudinally rather than at a single point in time. This allowed researchers to statistically examine which factors, measured in someone's 30s and 40s, actually predicted their health and well-being in their 70s and 80s, decades later, with the benefit of knowing exactly how each life actually played out rather than relying on prediction alone.
STEP-BY-STEP EXECUTION
Findings accumulated across decades rather than emerging from a single analysis, but one pattern stood out with unusual consistency across both the Harvard cohort and the Glueck cohort, despite their very different starting socioeconomic circumstances. The strength and quality of a person's close relationships in midlife, not wealth, career achievement, IQ, or social class, was the strongest predictor of physical health and reported happiness measured decades later, in old age.
Vaillant's analysis, synthesized in his 2012 book Triumphs of Experience , found that men who had warm, stable relationships in their 40s and 50s were significantly more likely to be physically healthy and psychologically well in their 80s, while men who reported chronic relational conflict or isolation showed measurably worse outcomes on both fronts, independent of their income or occupational success. Waldinger's later analysis, widely popularized through his 2015 TED talk, reinforced this: participants who felt genuinely satisfied in their relationships at age 50 were physically healthier at 80 than participants with lower cholesterol readings at 50, a comparison Waldinger used specifically to illustrate how much more predictive relational quality was than a standard medical risk factor.
Both cohorts also revealed that the specific number of relationships mattered far less than their quality — participants with a smaller number of genuinely trusted, emotionally close relationships fared consistently better than participants with a wide social circle lacking real depth.
PSYCHOLOGICAL REVELATION
This research demonstrated something that ran directly against the assumptions many of the original researchers started with in 1938: that a good life could be substantially predicted by physical constitution, achievement, or intelligence. Decades of data instead pointed toward something far less measurable at the time anyone was young enough to still change course, the depth and reliability of close relationships, as the single strongest predictor available.
This connects directly to the research explored in The Gottman Love Lab , which identified specific relational patterns that predict whether a couple's relationship actually lasts. Where Gottman's research zooms into the mechanics of a single relationship, the Harvard study zooms out across an entire lifetime, and both point toward the same underlying conclusion: relational quality isn't a soft, secondary factor in a well-lived life. It's one of the most consistently measurable ones available.
It also offers a longitudinal counterpoint to the pattern explored in Why Adult Friendships Are So Hard : the difficulty of maintaining close relationships past early adulthood isn't a minor inconvenience of modern life. Based on this research, it may be one of the more consequential challenges a person navigates across their entire lifespan, whether or not it's treated with that level of seriousness at the time.
See Your Own Relational Patterns More Clearly
Reading about eight decades of data is different from understanding what your own relational baseline actually looks like right now, and how it might be shaping your long-term well-being.
→ Take the Relational Baseline Assessment
Free. No account required.
Curious how your closest friendships and social bonds specifically compare?
→ Explore the Friendship & Social Bonds Assessment in the PsychOS Assessment Lab
Frequently Asked Questions
What is the Harvard Study of Adult Development?
It's the longest-running longitudinal study of human life, beginning in 1938 with two combined cohorts, one of Harvard sophomores and one of disadvantaged Boston youths, tracking participants and eventually their children across more than 80 years to identify what actually predicts long-term health and well-being.
What was the study's most significant finding?
Across both original cohorts, the quality and stability of close relationships in midlife was the strongest predictor of physical health and reported happiness decades later, outperforming wealth, career success, IQ, and even some standard medical risk factors.
Is this a controlled experiment?
No. It's a naturalistic, longitudinal observational study rather than a controlled experiment with a manipulated variable. This means its findings describe strong, well-documented associations across decades of real data rather than experimentally proven causation.
Does the study track anyone alive today?
Yes. While most of the original participants have died, the study now includes their children, sometimes called the "second generation," continuing the same long-term tracking approach into a new cohort.
Why does this study matter more than shorter studies on happiness?
Most psychological research captures a single snapshot or a short follow-up period. This study's unique value comes from tracking the same individuals across their entire adult lives, allowing researchers to see what actually played out decades later rather than relying on prediction alone.