Do non-scientists
really live longer?
SIR-In 1954 the influential Stanford psychologist Lewis Terman asked: "Are scientists different?" He was wondering how to recruit more scholars to be scientists and how to smooth scientists’ relations with politicians and lawyers.! For an answer, Terman turned to his large-scale longitudinal study of gifted children, which began in 1922. Terman searched for the psychosocial qualities that distinguished the scientists from the businessmen and lawyers in his sample of 788 bright men followed up since childhood. He concluded that "the physical scientists and engineers are at the opposite pole from the businessmen and lawyers in abilities, occupational interests and in social behavior".1 He also found that scientists were unsociable: they were more shy and less interested in social relations at school and in young adulthood, and Terman speculated that these differences accounted for friction between scientists and government officials and politicians. Differences in sociability have become a major focus of attention in health psychology research, since there is good evidence that social ties are related to good health.2 The Terman sample is an excellent one in which to examine sociability and health issues because the sample is highly intelligent, educated, and ethnically homogeneous (99% white), with access to health knowledge, nutrition, and medical care. Any differences between scientists and non-scientists in health outcomes are likely to be largely attributable to differences in health-relevant behaviours or in stress-relevant factors. To address these issues, we gathered death certificates for the Terman sample up to 1991. The Terman Life-Cycle Study is the longest prospective study of a single cohort ever conducted. Terman’s aim was to secure a reasonably random sample of bright California children, and most public sector schools in San Francisco and Los Angeles in the 1920s were searched for bright children. The average birth date was 1910. In 1951, Terman focused on 284 scientists and 327 non-scientists. From Terman’s archives at Stanford University we recreated his groups, except for a slight discrepancy in the classification of 4 men. Our study began with 288 scientists and 326 non-scientists; elimination of those who had left the study by 1950 and those who were not of school age in 1922 left 240 scientists and 269 non-scientists of average age 40-2 and 39-7 years, respectively, in 1950. Using death certificate information from state bureaus, we computed death rates from 1950 to 1991 by survival analysis. Of the 509 men, 238 (47%) are recorded as having died. Previous research on the Terman sample33 indicated childhood conscientiousness/social-responsibility as a predictor of longevity, and so this and a measure of sociability in childhood33 were also examined. Mortality was predicted by Cox’s proportional hazards regression. The non-scientists have a marginally greater mortality risk than scientists: in any given year, non-scientists were 26% more likely to die than scientists (relative hazard 1 ’26, p < 009). 72% of scientists but only 67% of non-scientists reached 70. If we assume that those not known to be dead are alive the figures become 80% and 74%. Might the scientists have been protected from specific cause(s) of death? Computation of death rates by cause followed by Gompertz modelling (details not shown) revealed that any increase in risk associated with being a non-scientist operates equally across the causes of death. A comparison of the two groups on our childhood sociability in line measure revealed a significant difference (p<0’001) with Terman’s claims about the greater sociability of nonscientists. Contrary to expectation, scientists were slightly (but non-significantly) less conscientious than non-scientists in childhood so conscientiousness cannot account for scientists’
296
longevity. Scientists reported drinking less than non-scientists so alcohol was entered into a Cox’s model. However, controlling for alcohol hardly affected the relative hazard associated with being a non-scientist. In a much smaller subsample we found that scientists and non-scientists differed little in their smoking habits by pack-years or by "ever smoking" (64% of scientists, 69% of non-scientists). Obesity did not predict longevity and did not differ between scientists and non-scientists. The scientist/non-scientist distinction might seem crude for it lumps together in one category those oriented towards humanities, non-research social sciences, business, and law, and collapses those in engineering, medicine, and the natural sciences. Yet when Terman compared the two on a whole host of measures in 1951 (he used the newly developed IBM electrical computers) he found the two groups were either alike or very diverse on most variables. For example, some scientists (doctors) and some non-scientists (lawyers) earned a very good income. The groups consistently differed in two ways: in their interest in science and in their social relations. Terman’s analysis left "no doubt that non-scientists tend to score higher than scientists in social relations" across a wide range of attributes, indices, and times.! For example, non-scientists saw themselves (and were seen by others) as more sociable, and they were more likely to engage in public service and community affairs. The effects of social support on all-cause mortality risk2,4 are comparable with those of better-known risk factors such as hypertension and possibly even smoking. It is sometimes suspected that the better health of those high in social ties is due to some advantage of personality. Our study argues against that interpretation. It is not true that intelligent men who are highly sociable in their youth will outlive their unsociable, scienceoriented peers. As a footnote, we would point out that, fortunately (in light of Terman’s original aims), the findings do bode well for attracting the brightest scholars into science.
Supported by research grant from National Institute on Aging AG08825. Part of the data were made available from the Terman Life-Cycle Study, begun by Lewis Terman, and followed up by Robert Sears and others; further assistance provided by Eleanor Walker. Howard S Friedman, Joan S Tucker, Leslie R Martin, Carol Tomlinson-Keasey, Jospeh E Schwartz, Deborah L Michael H Criqui
Wingard,
Psychology Department, University of California, Riverside, CA 92521, USA; Office of Provost, University of California, Davis; Department of Psychiatry and Behavioral Sciences, SUNY Stony Brook, New York; and Department of Family and Preventive Medicine, University of California, San Diego 1 2
Terman LM. Are scientists different? Sci Am 1955 (Jan): 7-11. House JS, Landis KR, Umberson D. Social relationships and health. Science 1988; 241: 540-45.
3
Friedman HS, Tucker JS, Tomlinson-Keasey C, Schwartz JE, Wingard DL, Criqui MH. Does childhood personality predict longevity? J Pers Social Psychol 1993; 65: 176-85. Orth-Gomer K, Rosengren A, Wilhelmsen L. Lack of social support and incidence of coronary heart disease in middle-aged Swedish men. Psychosom Med 1993; 55: 37-43.
4
Drug legalisation SIR-I was the honoured guest of Mayor Schmoke of Baltimore in November, in appreciation of the harm-reduction policies with respect to drug addiction, initiated by Allan Parry and myself in the Mersey region in the 1980s. Early in December, 1993, the Swiss authorities began a heroinprescribing experiment, stimulated by a lecture tour Dr Russell Newcombe and myself undertook to Switzerland in 1990, extolling the effects of controlled availability of drugs in Merseyside. In the UK, however, this practice is often condemned. But the real message from Baltimore came from Senator Gomez HurtadoUS ambassador to France and High