Triumphs of Experience: The Men of the Harvard Grant Study (11 page)

1938–1945

    
8–10 psychiatric interviews

    
Complete physical exam by Dr. Heath

    
Interviews by Dr. Heath and Miss Gregory; she also made home visits

    
Anthropological and physiological testing

    
EEG, Rorschach, and handwriting analysis (on many of the men)

    
Complete psychometric testing and some projective testing by Dr. Wells

    
The 26 personality traits assigned by Dr. Woods (
Appendix E
)

    
Brief childhood assessment (1–3) by Miss Gregory

    
At age 21 the ABC College adjustment (soundness) ratings assigned

1946–1950

    
Debriefing of combat experience by Dr. Monks

    
Annual questionnaires begun

    
Interviews with wives by social anthropologist Dr. Lantis

    
Thematic Apperception Test by Dr. McArthur

    
Full staff conferences on each man at age 29; ABCDE ratings of personality soundness for future adjustment made

1950–1967

    
Mostly biennial questionnaires; little other contact

1967–1985

    
All men re-interviewed, mostly by Eva Milofsky or Dr. Vaillant (
Appendix A
) Beginning age 45, complete physical exams every five years until the present

    
Objective health scored 1–5 (Well, minor illness, chronic illness, disabling illness, dead)

    
Childhood environment assessed, blind to events after age 19

    
Wives twice sent questionnaires

    
Adjustment at work, love, and play assessed from age 30–47 on all men (
Appendix D
)

    
Adaptive coping style (narcissistic, neurotic, empathic) assessed

    
NEO (Costa and McCrae) administered by mail

    
Lazare Personality Inventory administered by mail

1985–2002

    
All men re-interviewed (
Appendix A
)

    
Adjustment at work, love, and play assessed from age 49–65 and from age 65–80 (
Appendix D
)

    
Wives and children sent questionnaires

    
Physical exams every five years and biennial questionnaires continued

    
Gallup
Organization’s Wellsprings of a Positive Life administered by mail

    
Aging at 80 computed (subjective and objective mental and physical health)

2002–2010

    
Physical exams every five years and biennial questionnaires continued

    
Joint marital interviews and “daily diaries”

    
Telephone Interview for Cognitive Status (TICS) at ages 80, 85, and 90

    
Re-interviewed at retirement, 1985–2005

2010

    
Decathlon scores compiled

Case in point: the scientific pendulum has swung once more, and in the early twenty-first century genetic research once more dominates studies of environment. Instead of handwriting samples, the Study is collecting DNA. We don’t yet know how it will be used. What will
that
look like to readers seventy-five years from now?

EARLY DATA ANALYSIS

By 1941, the Study had examined 211 sophomores. A debate arose as to whether this was enough. Frederic Wells, perhaps the most established researcher employed by the Study, sent a letter to Clark Heath imploring that he stop accepting new students so that the staff could begin evaluating the huge amount of data that they already accumulated. But others worried that 211 cases was only a small number, after all, when the goal was to find conclusive correlations between
the
psychological and anthropological. Eventually a compromise was reached, and the researchers agreed to take on one last group of students from the class of 1944. They were studied as sophomores in 1942 and completed the total Study cohort of 268. Between 1938 and 1943 the Grant Foundation had contributed $450,000 ($7 million in 2009 dollars) to the research.

But now the Study had to decide how to harvest this vast planting, and that was not proving easy. As I’ve said, a longitudinal study has to stay a bit ahead of itself, both in its data collecting and in its theoretical hunches. After all, it can’t know in advance what will prove relevant (if it did, there would be no need for the study), and has to rely on best guesses. This can be an advantage as well as a disadvantage. But the Grant Study was a very early longitudinal study, so it didn’t have the benefit of others’ experience, and it began in something of a theoretical vacuum, when very little was known about adult growth and development. It had been very ambitious in collecting data, but not always very thoughtful about what it would do with the results. Many of the investigators were feeling overwhelmed, and by 1944 William Grant was beginning to express serious doubts about the administration of the Study and about whether he should continue to fund it. The trustees of the Grant Foundation put increasing pressure on the researchers to produce a summary of the Study’s early results. For the next two years, the Study archives are filled with desperate efforts to discover publishable material.

Only three papers appeared before 1945 (see
Appendix F
), and in 1945, two monographs were hastily published to meet Grant’s demand for results and a popular book that would dramatize the Study’s findings.

Earnest Hooton was a Harvard professor, a brilliant physical anthropologist, and a fluent writer. He had been Study anthropologist Carl Seltzer’s mentor (also my archaeologist father’s). And he was
firmly
committed to the world of constitutional medicine. “If we wish to study the whole man, we must begin with his physique,” he maintained, and he imagined that research out of the Grant Study might one day lead to “effective control of individual quality through genetics, or breeding.”
16
It was he who had encouraged early researchers to expect that “[O]n the whole . . . ‘normality’ goes closely with a ‘strong masculine component.’
”
17
Hooton’s summary of the Grant Study was published in 1945 as
Young Man, You Are Normal.
18
That was a far cry from Grant’s own proposed title,
The Grant Study of Social Adjustment,
which did not help to bridge the growing rift between the Study administration and its funder. Still, for the next thirty years, Hooton’s was the leading book on the study.

In it he wrote, “When physique, studied from different standpoints, turns out to be so intimately related to various personality traits, it is clear that body build must also furnish clues to the social capacities of the individual.”
19
Instead of adducing any experimental evidence in support of these claims, however, he simply dismissed his opposition as “crass environmentalists.”
20
This is a striking term. It conveys not just simple cluelessness about the environmental considerations that are so crucial now, but real antagonism to an entire way of thinking. Yet Arlie Bock had already been wondering about the nature/nurture issue in his first visions of the Study seven years before; it’s possible to see in the difference between his attitude and Hooton’s a scientific paradigm nearing its tipping point.

The second monograph was
What People Are,
by Study director Clark Heath. Heath’s book relied heavily on anthropometric measurements too, and also on an untested personality profiling scheme devised by William Woods, a staff psychiatrist without research training.
21

Woods’s schema (see
Appendix E
) scored Study members on twenty-six personality traits, many of them dichotomies such as
Vital
(warm,
expressive)
Affect vs. Bland
(colorless)
Affect, Well-Integrated Personality vs. Unintegrated Personality, Verbal vs. Inarticulate, Sociable vs. Asocial
etc. An attempt was then made to correlate the resulting profiles with body build but the results were unconvincing.

Neither of these two hastily published books attracted much attention, and certainly they did nothing to halt the shift toward environmentalism in the social sciences. In the years since, somatotype categories have not proven particularly useful when matched against other independent ratings, and neither have (most of) Woods’s personality traits. Even at the time, what evidence there was, was not persuasive. Worse, the early investigators were not blind to each other’s ratings, so there’s no way to know how much apparent early correlations between body build and personality were a function of halo effect or observer bias. In 1970 I tried to replicate the correlations reported in some of the early papers and could not.

Before we contemptuously (and prematurely) dismiss the early investigators’ work as a wash, however, we have to consider the realities of statistical methodology in those days. There were no computers to absorb infinitudes of entries and magically align them along dozens of axes at the click of a key. Study members had to be listed down the left-hand side of huge ledger sheets. Scores and measurements were plotted along the top, individually, in exquisite handwriting. And that was only the beginning. To be useful, the data had to be pulled out of the ledgers, again by hand, and subjected to individual calculations, some of them quite complex. The tests that the original investigators used are out of fashion today, but analysis was laborious in a way that is unimaginable now. The Study in the 1940s did not even own a Monroe calculator (as the early electric adding machines were called). Ironically, in 1944, one of the world’s first reliable computers, the Mark I, was being installed only 300 yards away. But the techniques and technologies of complex data analysis that could (and in time did) reveal
unexpected correlations in the early data were not yet available to the Grant Study researchers. When the rich crop they had sown finally did come to fruition after the explosion of computer technology in the late twentieth century, it was the early investigators who had done much of the heavy lifting.

EARLY CONCLUSIONS AND SOME CORRECTIONS

The specter of World War II engaged early the Study’s interest in possible military applications for its research, particularly the identification of potential officers and appropriate placements for them. John Monks was a patrician internist who joined the staff in 1946 to study the men’s response to the war, and produced a well-researched if little-appreciated monograph,
College Men at War.
22
Monks treated Woods’s personality traits as if their scientific validity had been proven, with the suggestion that they be used to identify desirable traits in officer candidates. His monograph was fascinating in the stories and backgrounds it gave of the College men. It was disappointing, however, because Monks, like the original investigators, had not yet grasped the power of longitudinal study to actually test a hypothesis.

And the fresh empirical follow-up of 2010 (which came out of the questions I was trying to answer for this book as well as the Decathlon challenge) found that such promising variables as
Masculine Body Build, Well Integrated, Vital Affect,
or such foreboding ones as
Lack of Purpose and Values
and
Shy,
bore no relation at all to attained military rank. The only college trait that did predict high rank was
Political,
while low rank was predicted only by . . .
Cultural
and
Creative and Intuitive!
The scriptwriters of
M*A*S*H
might have anticipated that result, but Bock’s original investigators certainly did not.

Woods’s predictive schema has had more general problems than its failure to identify good officer material. Most of his traits failed to
correlate
significantly with any of the events in the 2010 Decathlon of Flourishing, and only one correlated significantly with more than three. That one, however, was a notable exception.
Well Integrated
(defined as “steady, dependable, thorough, sincere, and trustworthy”) correlated with eight Decathlon events, and for the last twenty-five years has been a staple variable in our data analyses. It signals a bundle of traits that enable a young man “to surmount common problems which confront him such as career choice, competitive environment, and moral and religious attitudes.”
23
It was assigned to 60 percent of the men, while 15 percent were called
Incompletely Integrated.
These latter men were deemed to lack perseverance, and were seen as “erratic, unreliable, sporadic, undependable, ill directed and little organized.” (The remaining 25 percent of the men were unclassified for this dichotomized variable.) Half a century later, proportionately four times as many of the
Well Integrated
enjoyed good marriages than the
Incompletely Integrated.
Another finding of great interest was that as of 2012, the
Well Integrated
have lived on average seven years longer. In contrast, the two variables that the original investigators thought would be most predictive of positive outcome,
Vital Affect
and
Sociable,
were unimportant after the first ten years.