I regret having only one upvote to give to these wonderful gems you keep digging up gwern. When ems of you are available on the market I'll buy at least two.
I think it's time for a meta-post in which gwern discusses summarizing articles and gives advice.
eminent scientists tend to be
Base rate?
I think it's time for a meta-post in which gwern discusses summarizing articles and gives advice.
http://lesswrong.com/lw/5me/scholarship_how_to_do_it_efficiently/ is pretty good, IMO.
That's advice for the skimming/reading/intensive study of 1,000 papers to get their knowledge, balancing completeness, depth, breadth, and the like.
I want advice on summarizing 100 individual articles, each one fairly completely read, so that many other people can do that and share the results with each other. The thing you do best, rather than the thing lukeprog does best.
Hm. I dunno then. I read, excerpt the interesting bits, and paste them into a file. Is there anything else to explain?
"Scientists are curious and passionate and ready to argue":
This psychological assessment owes nothing to surveys or personality testing; it pays no heed to the zodiac. Instead, researchers took the linguistic data from 200 tweets each of nearly 130,000 Twitter users across more than 3,500 occupations to assess their “personality digital fingerprints”.
They used machine learning to identify the traits and values that distinguish professions from each other.
This “21st century approach for matching one’s personality with congruent occupations,” dubbed the robot career adviser, is more reliable than existing career guidance methods based on self-reports through questionnaires, the researchers argue in a January paper in Proceedings of the National Academy of Sciences.
Personality characteristics of Twitter users were inferred using IBM Watson's Personality Insights tool. The study focused on five specific traits (extraversion, agreeableness, conscientiousness, emotional stability, and openness) and five values (helping others, tradition, taking pleasure in life, achieving success, and excitement).
The analysis revealed that scientists combine low agreeableness and low conscientiousness with high openness.
“The combination is characteristic of people who tend to be unconventional and quirky, consistent with the image of scientists as curious and even sometimes eccentric boffins,” says one of the authors, Paul McCarthy+, an adjunct professor at the University of New South Wales (UNSW Sydney) in Australia. “In some ways it does confirm stereotypes.”
Within the sciences, a spectrum of personality traits emerged. Those dealing with more abstract or inanimate things (mathematicians, geologists) were more open than those in the life sciences (bio-statisticians, horticulturalists), who “tended to be more extroverted and agreeable”, says McCarthy.
Scientists and software programmers, whose personality characteristics aligned closely, were generally more open to experiencing a variety of new activities, tended to think in symbols and abstractions, and found repetition boring, the researchers found.
On the spectrum of occupations, scientists are especially different from professional tennis players. “Tennis professionals are a lot more agreeable and conscientious than all others in the study — especially scientists,” says McCarthy.
“It makes sense, because to be a tennis player, you have to be highly conscientious and be willing to take direction, whereas scientists are almost the complete opposite. They don’t take direction, their openness to experiences is very high, but so is their openness to being disagreeable.”
He noted that the occupation of research director had the highest median openness scores of any of the 3,513 occupations in the study.
"Now for the matter of drive. You observe that most great scientists have tremendous drive. I worked for ten years with John Tukey at Bell Labs. He had tremendous drive. One day about three or four years after I joined, I discovered that John Tukey was slightly younger than I was. John was a genius and I clearly was not. Well I went storming into Bode's office and said, "How can anybody my age know as much as John Tukey does?" He leaned back in his chair, put his hands behind his head, grinned slightly, and said, "You would be surprised Hamming, how much you would know if you worked as hard as he did that many years." I simply slunk out of the office!
What Bode was saying was this: "Knowledge and productivity are like compound interest." Given two people of approximately the same ability and one person who works ten percent more than the other, the latter will more than twice outproduce the former. The more you know, the more you learn; the more you learn, the more you can do; the more you can do, the more the opportunity - it is very much like compound interest. I don't want to give you a rate, but it is a very high rate. Given two people with exactly the same ability, the one person who manages day in and day out to get in one more hour of thinking will be tremendously more productive over a lifetime. I took Bode's remark to heart; I spent a good deal more of my time for some years trying to work a bit harder and I found, in fact, I could get more work done. I don't like to say it in front of my wife, but I did sort of neglect her sometimes; I needed to study. You have to neglect things if you intend to get what you want done. There's no question about this."
Richard Hamming, "You and Your Research". (Note that Hamming is, of course, talking about top scientists, not people in general - so he's assuming that the basic thresholds for IQ and Openness are already met, even though I doubt he ever conceptualized that explicitly.)
We’ve discussed the Big Five in the past, such as the relationship of Openness to parasites & signaling or whether hallucinogens increase Openness and parasites decrease it, along with my little notes on the value of Conscientiousness. This is another entry in the topic of ‘what is Big Five good for’.
I researched the topic of how and whether Conscientiousness and Openness correlate with scientific achievement for Luke for the Intelligence Explosion paper; here is some of what I found:
“Creativity, Intelligence, and Personality”, 1981 review:
“Studies of creative adult artists, scientists, mathematicians, and writers find them scoring very high on tests of general intelligence (e.g. Barron 1969; Bachtold & Werner 1970; Helson & Crutchfield 1970b; Cattell 1971; Helson 1971; Bachtold & Werner 1973; Gough 1976a), though rs between tested intelligence and creative achievement in these samples range from insignificantly negative (r = -.05, Gough 1976a) to mildly and significantly positive (r = +.31, Helson 1971).”
I found this one amusing:
“It should be noted that creative people are often perceived and rated as more intelligent than less creative people even in samples where no corresponding correlations between tested intelligence and creativity obtain. Despite an r of -.08 between Terman’s Concept Mastery Test and professionally rated creativity among the top 40 IPAR architects (MacKinnon 1962a), e.g., staff ratings of the single adjective “intelligent” correlated +.39 with the index of creativity (MacKinnon 1966). While such an r may reflect some spurious halo effects, it may also tell us something about the true overlap in meaning of these terms in the natural language."
An embarrassment of riches; no summary, but a starting point if one needs more:
“SCIENCE AND TECHNOLOGY: Personality correlates of scientific achievement and creativity were studied in elementary school children (Milgram et al 1977); high school students (Schaefer & Anastasi 1968, Parloff et al 1968, Anastasi & Schaefer 1969, Schaefer 1969a, b, Walberg 1969a); undergraduates, young adults, and graduate students (Rossman & Horn 1972, Schaefer 1973, Gough 1979, Korb & Frankiewicz 1979); psychologists (Chambers 1964, Wispe 1965, Bachtold & Werner 1970); inventors (Bergum 1975, Albaum 1976, Albaum & Baker 1977); mathematicians (Helson 1967b, 1968a, Parloff et al 1968; Helson & Crutchfield 1970a, b; Helson 1971; Gough 1979); chemists (Chambers 1964); and assorted engineers and research scientists (McDermid 1965, Owens 1969, Bachtold & Werner 1972, Bergum 1973, Eiduson 1974, Gough 1979).”
“How development and personality influence scientific thought, interest, and achievement”, GJ Feist, Review of General Psychology, 2006; important bits start on pg 9:
"In 1998, I published a quantitative review of the literature on personality and scientific interest and creativity (Feist, 1998. In this meta-analytic review of which personality traits make interest and creativity in science more likely, I found every published (and some unpublished studies) that examined the role in personality in scientific interest or scientific creativity from 1950 to 1998. There were 26 studies that reported quantitative effects of personality in scientists compared to non-scientists.
…The two strongest effect sizes (medium in magnitude) were for the positive and negative poles of conscientiousness (C; see Table 1). Being high in conscientiousness (Cϩ) consists of scales and items such as careful, cautious, conscientious, fastidious, and self-controlled, whereas being low in conscientiousness (CϪ) consists only of two scales/items, namely, direct expression of needs and psychopathic deviance. Although the CϪ dimension comprised only five comparisons, it is clear that relative to non-scientists, scientists are roughly a half a standard deviation higher on conscientiousness and controlling of impulses. In addition, low openness to experience had a median d of .30, whereas introversion had a median effect size of .26.
A consistent finding in the personality and creativity in science literature has been that creative and eminent scientists tend to be more open to experience and more flexible in thought than less creative and eminent scientists (see Table 2). Many of these findings stem from data on the flexibility (Fe) and tolerance (To) scales of the California Psychological Inventory (Feist & Barron, 2003; Garwood, 1964; Gough, 1961; Helson, 1971; Helson & Crutchfield, 1970; Parloff & Datta, 1965). The Fe scale, for instance, taps into flexibility and adaptability of thought and behavior as well as the preference for change and novelty (Gough, 1987). The few studies that have reported either no effect or a negative effect of flexibility in scientific creativity have been with student samples (Davids, 1968; Smithers & Batcock, 1970).
For instance, Feist and Barron (2003) examined personality, intellect, potential, and creative achievement in a 44-year longitudinal study. More specifically, they predicted that personality would explain unique variance in creativity over and above that already explained by intellect and potential. Results showed that observer-rated Potential and Intellect at age 27 predicted Lifetime Creativity at age 72, and yet personality variables (such as Tolerance and Psychological Mindedness) explained up to 20% of the variance over and above Potential and Intellect. Specifically, two measures of personality—California Psychological Inventory scales of Tolerance (To) and Psychological Mindedness (Py)–resulted in the 20% increase in variance explained (20%) over and above potential and intellect. The more tolerant and psychologically minded the student was, the more likely he was to make creative achievements over his lifetime. Together, the four predictors (Potential, Intellect, Tolerance, and Psychological Mindedness) explained a little more than a third of the variance in lifetime creative achievement. I should point out that these findings on To and Py mirror very closely those reported by Helson and Pals (2000) in a longitudinal study of women from age 21 to 52.
…Busse and Mansfield (1984), for example, studied the personality characteristics of 196 biologists, 201 chemists, and 171 physicists, and commitment to work (i.e., “need to concentrate intensively over long periods of time on one’s work”) was the strongest predictor of productivity (i.e., publication quantity) even when holding age and professional age constant. Helmreich, Spence, Beane, Lucker, and Matthews (1980) studied a group of 196 academic psychologists and found that different components of achievement and drive had different relationships with objective measures of attainment (i.e., publications and citations). With a self-report measure, they assessed three different aspects of achievement: “mastery” preferring challenging and difficult tasks; “work” enjoying working hard; and “competitiveness” liking interpersonal competition and bettering others….Helmreich and his colleagues found that mastery and work were positively related to both publication and citation totals, whereas competitiveness was positively related to publications but negatively related to citations
…Helson (1971) compared creative female mathematicians with less creative female mathematicians, matched on IQ. Observers blindly rated the former as having more “unconventional thought processes,” as being more “rebellious and non-conforming,” and as being less likely to judge “self and others in conventional terms.” More recently, Rushton, Murray, and Paunonen (1987) conducted factor analyses of the personality traits most strongly loading on the “research” factor (in contrast to a “teaching” factor) in two separate samples of academic psychologists. Among other results, they found that “independence” tended to load on the research factor, whereas “extraversion” tended to load on the teaching factor."
Pretty substantial. The 26 study meta-analysis was Feist, G. J. (1998). “A meta-analysis of the impact of personality on scientific and artistic creativity”. Personality and Social Psychological Review, 2, 290–309
Feist 1998 was also cited in a chapter of The Cambridge Handbook of Creativity, “the relationship between creativity and intelligence”, but I couldn’t get the book; further reading, if anyone wants some.
One useful bit from Feist; I also ran into a lot of CP Benbow-keyworded studies of the gifted SMPY cohorts to the effect that the kids’ early interest in science predicts later careers in science, which would tie in nicely to this:
“The empirical consensus is that early levels of high productivity do regularly predict continued levels of high productivity across one’s lifetime (Cole, 1979; Dennis, 1966; Helson & Crutchfield, 1970; Horner et al., 1986; Lehman, 1953; Over, 1982; Reskin, 1977; Roe, 1965; Simonton, 1988a, 1991)….In other words, the younger NAS members were when they and others recognized their scientific talent, when they wanted to be a scientist, and when they first conducted scientific research, the younger they were when they published their first paper. Age of first publication in turn predicted total publication rate over the lifetime, meaning that the earlier one publishes, the more productive one will be. This pattern of relationships—from precocity to age of first publication to lifetime productivity—implies an indirect connection between precocity and publication rate. The only precocity variable that reached the .05 level of significance with lifetime productivity was age that one first conducted formal research.”
“Scientific talent, training, and performance: Intellect, personality, and genetic endowment”, Simonton 2008; the abstract caught my eye:
"After specifying the ideal data requirements for the application of the three estimators, the procedures were applied to previously published results. Personality traits were illustrated with the use of the California Psychological Inventory and the Eysenck Personality Questionnaire with respect to two criteria (scientists versus non-scientists and creative scientists versus less creative scientists) and intellectual traits with the use of the Miller Analogies Test with respect to seven criteria (graduate grade-point average, faculty ratings, comprehensive examination scores, degree attainment, research productivity, etc.). The outcome provides approximate, lower-bound estimates of the genetic contribution to scientific training and performance.
…Sawyer reviewed three investigations that allegedly disconfirm the role of genetic endowment in any form of creative achievement (viz., Barron, 1972; Reznikoff, Domino, Bridges, & Honeyman, 1973; Vandenberg, Stafford, & Brown, 1968).2 Likewise, Ericsson, Roring, and Nandagopal (2007) cited two behavior genetic investigations in drawing the same conclusion about talent in general (viz., Bouchard & Lykken, 1999; Klissouras et al., 2001).
…This argument certainly applies to scientific talent. For instance, a comprehensive longitudinal study of the mathematically precocious (Lubinski, Webb, Morelock, & Benbow, 2001) has extremely few twins in the sample (D. Lubinski, personal communication, March 15, 2007). Simonton’s (1991a) study of 2,026 eminent scientists contained only one twin (Auguste Picard). And, needless to say, there are no twins, monozygotic or dizygotic, among Nobel laureates in the sciences. Thus, not only may we lack direct evidence for scientific talent, but also it may never be possible to establish such substantiation with the use of standard behavior genetic methods.
…Bouchard and Lykken (1999) demonstrated that the personality characteristics associated with scientific productivity display heritabilities ranging between .32 and .57, meaning that between 32% and 57% of the variance in those traits can be attributed to genetic endowment. Similarly, the Creativity Personality Scale (CPS) of the Adjective Check List (ACL) not only predicts scientific creativity (Gough, 1979) but also has a heritability of .54 (Bouchard & Lykken, 1999; Waller, Bouchard, Lykken, Tellegen, & Blacker, 1993). Hence, 54% of the variance in the CPS has a genetic contribution. Moreover, because predictive validities are known for this measure, we can draw a more powerful inference. For instance, CPS scores correlated .31 with the creativity ratings that expert judges assigned 57 mathematicians (Gough, 1979). This signifies that almost 10% of the variance in that criterion can be attributed to CPS (i.e., .312 ϭ .096). Multiplying the squared criterion–trait correlation by the heritability coefficient then yields .052, which implies that over 5% of the variance in the rated creativity of these mathematicians might be ascribed to the genetic part of the CPS scores.
“Bouchard & Lykken 1999” = Bouchard, T. J., Jr., & Lykken, D. T. (1999). “Genetic and environmental influence on correlates of creativity”. In N. Colangelo & S. G. Assouline (Eds.), Talent development III: Proceedings from the 1995 Henry B. & Jocelyn Wallace National Symposium on Talent Development (pp. 81–97). I couldn’t find this, but I did find http://cogprints.org/611/1/genius.html which describes it a little more (C-f ‘in press’).
Moving onwards, pg 9–12 discuss Feist 1998’s meta-analysis:
…Even worse, “science” was obliged to include the physical, biological, and social sciences as well as mathematics, engineering, and invention. To the extent that the personality profiles are closely tailored to domain-specific training or performance criteria, this definitional inclusiveness implies that the hc2 estimates are too low. This criticism is not intended to fault Feist’s (1998) meta-analytic review. To obtain sound effect size estimates he had no other option but to collate many diverse findings. Nevertheless, in this analysis of intellectual traits it is feasible to substitute somewhat more specific criteria for these more global contrasts.
For example, spatial ability has been identified as a crucial component of math–science talent that exhibits predictive utility beyond that provided by both mathematical and verbal ability (Webb, Lubinski, & Benbow, 2007). Yet measures of spatial intelligence display heritabilities almost as high as general intelligence (Bratko, 1996; McClearn, Johansson, Berg, & Pedersen, 1997). Moreover, these more specialized intellectual abilities may be especially useful in differentiating distinct types of scientific talents. For instance, in Roe’s (1953) classic study of 64 eminent scientists it was found that theoretical physicists, experimental physicists, biologists, psychologists, and anthropologists display distinctive profiles with respect to verbal, mathematical, and spatial intelligence.
Concluding:
…It is not possible to say exactly when the final sum would be, but a conservative guess might be that between 10% and 20% of the variance in these criteria could be potentially attributed to genetic effects (cf. Simonton, 2007).
For the sake of this discussion, then, suppose that .10 Յ hc2 Յ .20 holds for the training and performance criteria examined here. Does this outcome imply that scientific talent is an important substantive phenomenon? To answer this question requires that we obtain some kind of baseline for comparison….[the range of estimates] can be qualitatively expressed as medium to large (Cohen, 1988). This range is about as good as can be expected of most effects in the behavioral sciences (Meyer et al., 2001; Rosenthal, 1990). To offer specific comparisons, the lower-end estimate is about the same magnitude as the relation between psychotherapy and subsequent well-being, whereas the upper-end estimate is about the same size as the correlation between height and weight among U.S. adults (Meyer et al., 2001).