Social Psychology Quarterly Vol. 73, No. 1, 33–57 Ó American Sociological Association 2010 DOI: 10.1177/0190272510361602 http://spq.sagepub.com

Why Liberals and Atheists Are More Intelligent

SATOSHI KANAZAWA London School of Economics and Political Science

The origin of values and preferences is an unresolved theoretical question in behavioral and social sciences. The Savanna-IQ Interaction Hypothesis, derived from the Savanna Principle and a theory of the evolution of general intelligence, suggests that more intelligent individ- uals may be more likely to acquire and espouse evolutionarily novel values and preferences (such as liberalism and atheism and, for men, sexual exclusivity) than less intelligent indi- viduals, but that general intelligence may have no effect on the acquisition and espousal of evolutionarily familiar values (for children, marriage, family, and friends). The analyses of the National Longitudinal Study of Adolescent Health (Study 1) and the General Social Sur- veys (Study 2) show that adolescent and adult intelligence significantly increases adult lib- eralism, atheism, and men’s (but not women’s) value on sexual exclusivity.

Keywords: origin of values and preferences, evolutionary psychology, the Savanna Princi- ple, the Savanna-IQ Interaction Hypothesis, general intelligence

here do individual values and pref- and preferences, although Becker (1996) has erences come from? Why do people since attempted to explain them. A theory of Wwant what they want? Some social revealed preferences, which is often used in scientists and biologists have explored the microeconomics, only measures individuals’ origin of values (Emerson 1987; Hechter, preferences empirically but does not explain Nadel, and Michod 1993) while economists where they come from or why actors have have remained mute on the issue. Their tradi- them. Despite many attempts and some prom- tional answer to the question of individual ising starts (Hechter et al. 1999; Schwartz values and preferences is: De gustibus non 1992; Wildavsky 1987), there currently is no est disputandum (Stigler and Becker 1977). satisfactory general theory of values. There is no accounting for tastes, and one can- Some argue that evolutionary psychology not explain individuals’ idiosyncratic values can provide such a general theory of values (Ben-Ner and Putterman 2000; Horne 2004; Kanazawa 2001). Evolutionary psychology is This research uses data from Add Health, a program the study of universal human nature, or sex- project designed by J. Richard Udry, Peter S. Bearman, specific male human nature and female human and Kathleen Mullan Harris, and funded by a grant P01- HD31921 from the National Institute of Child Health nature, and their interaction with the environ- and Human Development, with cooperative funding ment. It can therefore in principle explain both from 17 other agencies. Special acknowledgment is due universal preferences (as a function of the uni- Ronald R. Rindfuss and Barbara Entwisle for assistance versal human nature) and individual preferen- in the original design. Persons interested in obtaining ces (as a function of the interaction between data files from Add Health should contact Add Health, Carolina Population Center, 123 West FranklinStreet, the universal human nature and individual cir- Chapel Hill, NC 27516-2524; [email protected]. I cumstances and experiences). thank David de Meza, Norman P. Li, Andrew J. In this paper I discuss recent theoretical Oswald, Diane J. Reyniers, Pierre L. van den Berghe, developments in evolutionary psychology and especially Jeremy Freese for their comments on and offer one possible explanation of earlier drafts. Direct all correspondence to Department of Management, London School of Economics and individual values and preferences that I Political Science, Houghton Street, London WC2A 2AE, call the Savanna-IQ Interaction Hypothesis. United Kingdom; [email protected]. It explains how the level of general intelligence 34 SOCIAL PSYCHOLOGY QUARTERLY affects the acquisition of certain evolution- social exchange and absolutely no possibility arily novel values and preferences. I then of knowing future interactions (which makes test the Savanna-IQ Interaction Hypothesis the game truly one-shot). Neither of these sit- with respect to three evolutionarily novel val- uations existed in the ancestral environment, ues (liberalism, atheism, and monogamy) and but they are crucial for the game-theoretical show that, consistent with the Hypothesis, prediction of universal defection. more intelligent individuals are more likely As another illustration of the Savanna Prin- to espouse liberal political ideology and to ciple, individuals who watch certain types of be atheists, and more intelligent men (but TV shows are more satisfied with their friend- not women) are more likely to value sexual ships, just as they are if they had more friends exclusivity. In contrast, and consistent with or socialized with them more frequently the prediction, general intelligence does not (Kanazawa 2002). This may be because realis- affect the espousal of evolutionarily familiar tic images of other humans, such as television, values for children, marriage, family, and movies, videos, and photographs, did not exist friends. in the ancestral environment, where all realis- tic images of other humans were other humans. THE SAVANNA PRINCIPLE As a result, the human brain may have implicit difficulty distinguishing their ‘‘TV friends’’ Adaptations, physical or psychological, (the characters they repeatedly see on TV are designed for and adapted to the conditions shows) and their real friends. Caughey of the environment of evolutionary adapted- (1984:31-76) details the ‘‘imaginary social ness, not necessarily to the current environ- relations’’ that many Americans have with ment (Tooby and Cosmides 1989). This media figures, including TV characters. fundamental principle of evolution holds equally for psychological adaptations as it EVOLUTION OF GENERAL INTELLIGENCE does for physical adaptations. Pioneers of evolutionary psychology (Crawford 1993; General intelligence refers to the ability to Symons 1990; Tooby and Cosmides 1990) reason deductively or inductively, think all recognized this. Kanazawa (2004a) sys- abstractly, use analogies, synthesize informa- tematizes these observations into what he tion, and apply it to new domains (Gottfredson calls the Savanna Principle: The human brain 1997; Neisser et al. 1996). The concept of has difficulty comprehending and dealing general intelligence poses a problem for evo- with entities and situations that did not exist lutionary psychology. Evolutionary psycholo- in the ancestral environment. Hagen and gists contend that the human brain consists of Hammerstein (2005:341-3) refer to the same domain-specific evolved psychological mech- observation as the ‘‘mismatch hypothesis,’’ anisms, which evolved to solve specific adap- while Burnham and Johnson (2006:130-31) tive problems (problems of survival and call it the ‘‘evolutionary legacy hypothesis.’’ reproduction) in specific domains. If the con- The Savanna Principle can potentially tents of the human brain are domain-specific, explain why some otherwise elegant scientific how can evolutionary psychology explain theories of human behavior, such as the sub- general intelligence? jective expected utility maximization theory In contrast to views expressed by Miller or game theory, often fail empirically, because (2000), Cosmides and Tooby (2002), and they posit entities and situations that did not Chiappe and MacDonald (2005), Kanazawa exist in the ancestral environment. For exam- (2004b) proposes that what is now known as ple, many players of one-shot Prisoner’s general intelligence may have originally Dilemma games may make the theoretically evolved as a domain-specific adaptation to irrational choice to cooperate with their part- deal with evolutionarily novel, nonrecurrent ner, possibly because the human brain has dif- problems. Kanazawa’s (2004b) theory of the ficulty comprehending completely anonymous evolution of general intelligence represents WHY LIBERALS AND ATHEISTS ARE MORE INTELLIGENT 35 some departure from the orthodox in intelli- of food because none of our normal gence research. It provides a theoretical defi- sources of food are working. What nition of general intelligence as well as an else can we eat? What else is safe explanation for its evolution, whereas the to eat? How else can we procure orthodox intelligence research defines it food? only empirically as the highest-order latent 3. A flash flood has caused the river to factor—the g factor—that emerges from swell to several times its normal a factor analysis of a large number of cogni- width, and I am trapped on one side tive (‘‘IQ’’) tests. For the orthodox, ‘‘general of it while my entire band is on the intelligence’’ and ‘‘g’’ are completely synony- other side. It is imperative that I mous, whereas for Kanazawa (2004b) g is an rejoin them soon. How could I cross indicator or measure of general intelligence, the rapid river? Should I walk across not general intelligence itself (Kanazawa it? Or should I construct some sort of 2007:284n). buoyant vehicle to use to get across The human brain consists of a large num- it? If so, what kind of material should ber of domain-specific evolved psychological I use? Wood? Stones? mechanisms to solve recurrent adaptive prob- lems. In this sense, our ancestors did not To the extent that these evolutionarily really have to think in order to solve such novel, nonrecurrent problems happened fre- recurrent problems. Evolution has already quently enough in the ancestral environment done all the thinking, so to speak, and equip- (different problem each time) and had serious ped the human brain with the appropriate enough consequences for survival and repro- psychological mechanisms, which engender duction, any genetic mutation that allowed preferences, desires, cognitions, and emo- its carriers to think and reason would have tions, and motivate adaptive behavior in the been selected for, and what we now call context of the ancestral environment. ‘‘general intelligence’’ could have evolved as Even in the extreme continuity and con- a domain-specific adaptation for the domain stancy of the ancestral environment, however, of evolutionarily novel, nonrecurrent prob- there were likely occasional problems that lems. General intelligence may have become were evolutionarily novel and nonrecurrent, universally important in modern life which required our ancestors to think and rea- (Gottfredson 1997; Herrnstein and Murray son in order to solve. Such problems may 1994; Jensen 1998) only because our current have included, for example: environment is almost entirely evolutionarily novel. The new theory suggests, and empirical 1. Lightning has struck a tree near the data confirm, that more intelligent individuals camp and set it on fire. The fire is are better than less intelligent individuals at now spreading to the dry underbrush. solving problems only if they are evolution- What should I do? How could I stop arily novel, but that more intelligent individu- the spread of the fire? How could I als are not better than less intelligent and my family escape it? (Since individuals at solving evolutionarily familiar lightning never strikes the same problems, such as those in the domains of place twice, this is guaranteed to be mating, parenting, interpersonal relationships, a nonrecurrent problem.) and wayfinding (Kanazawa 2004b, 2007). 2. We are in the middle of the severest Three recent studies, employing widely varied drought in a hundred years. Nuts and methods, have shown that the average intelli- berries at our normal places of gath- gence of a population appears to be a strong ering, which are usually plentiful, are function of the evolutionary novelty of its not growing at all, and animals are environment (Ash and Gallup 2007; Bailey scarce as well. We are running out and Geary 2009; Kanazawa 2008). 36 SOCIAL PSYCHOLOGY QUARTERLY

SAVANNA-IQ INTERACTION HYPOTHESIS individuals possibly because they are better able to recognize and deal with evolutionarily The logical conjunction of the Savanna novel threats and dangers to health in modern Principle and the theory of the evolution of society (Deary et al. 2004; Gottfredson and general intelligence suggests a qualification of Deary 2004; Kanazawa 2006b). Consistent the Savanna Principle. If general intelligence with the Hypothesis, however, general intelli- evolved to deal with evolutionarily novel gence does not appear to affect health and lon- problems, then the human brain’s difficulty in gevity in sub-Saharan Africa, where many of comprehending and dealing with entities and the health threats and dangers are more evolu- situations that did not exist in the ancestral tionarily familiar than elsewhere in the world. environment (proposed in the Savanna Princi- Finally, criminologists have long known ple) should interact with general intelligence, that criminals on average have lower intelli- such that the Savanna Principle holds stronger gence than the general population (Wilson among less intelligent individuals than among and Herrnstein 1985; Herrnstein and Murray more intelligent individuals. More intelligent 1994). From the perspective of the Hypothesis, individuals should be better able to comprehend there are two important points to note (Kana- and deal with evolutionarily novel (but not zawa 2009). Much of what we call interper- evolutionarily familiar) entities and situations sonal crime today, such as murder, assault, than less intelligent individuals. robbery, and theft, were probably routine There has been accumulating evidence for means of intrasexual male competition in the this Savanna-IQ Interaction Hypothesis. First, ancestral environment. This is how men likely individuals’ tendency to respond to TV char- competed for resources and mating opportuni- acters as if they were real friends, first discov- ties for much of human evolutionary history; ered by Kanazawa (2002), appears to be they beat up and killed each other, and they limited to those with below-median intelli- stole from each other if they could get away gence (Kanazawa 2006a); individuals with with it. We may infer this from the fact that above-median intelligence do not become behavior that would be classified as criminal more satisfied with their friendships by watch- if engaged in by humans, like murder, rape, ing more television. assault, and theft, is quite common among Second, net of age, race, sex, education, other species (Ellis 1998), including other pri- marital history, and religion, less intelligent mates such as chimpanzees (de Waal 1998), individuals have more children than more bonobos (de Waal 1992), and capuchin mon- intelligent individuals, even though they do keys (de Waal, Luttrell, and Canfield 1993). not want to do so. This may possibly be At the same time, the institutions that con- because they have greater difficulty effec- trol, detect, and punish criminal behavior in tively employing evolutionarily novel means society today—the police, the courts, and of modern contraception (Kanazawa 2005). the prisons—are all evolutionarily novel; Another indication that less intelligent indi- there was very little formal third-party viduals may have greater difficulty employ- enforcement of norms in the ancestral envi- ing modern contraception effectively is the ronment, only second-party enforcement (vic- fact that the correlation between the lifetime tims and their kin and allies) or informal number of sex partners and the number of third-party enforcement (ostracism). Thus it children is positive among the less intelligent makes sense from the perspective of the but negative among the more intelligent. The Savanna-IQ Interaction Hypothesis that men more sex partners less intelligent individuals with low intelligence may be more likely to have, the more children they have; the more resort to evolutionarily familiar means of sex partners more intelligent individuals competition for resources (theft rather than have, the fewer children they have. full-time employment) and mating opportuni- Third, more intelligent individuals stay ties (rape rather than computer dating) and healthier and live longer than less intelligent not to comprehend fully the consequences WHY LIBERALS AND ATHEISTS ARE MORE INTELLIGENT 37 of criminal behavior imposed by evolution- experiential novelty (entities and situations arily novel entities of law enforcement. that individuals have not personally experi- There thus appears to be some evidence enced in their own lifetime). While the Five- for the Savanna-IQ Interaction Hypothesis. Factor Model does not specify the type of nov- Applied to the origin of preferences and val- elty that open individuals are more likely to ues, the Hypothesis suggests that more intelli- seek, the Savanna-IQ Interaction Hypothesis gent individuals may be more likely to acquire suggests that more intelligent individuals are and espouse evolutionarily novel preferences more likely to seek only evolutionary novelty, and values than less intelligent individuals, not necessarily experiential novelty. while general intelligence may make no differ- For example, everybody who is alive in ence for the acquisition and espousal of evolu- the United States today has lived their entire tionarily familiar values. In particular, the lives in a strictly monogamous society, and, Hypothesis leads to predictions about three despite recent news events, very few contem- evolutionarily novel values of liberalism, porary Americans have any personal experi- atheism, and, for men, sexual exclusivity, ences with polygyny.1 Therefore monogamy and how general intelligence may affect their is experientially familiar for most Americans acquisition and espousal. whereas polygyny is experientially novel. The Five-Factor Model may therefore predict that more intelligent individuals are more GENERAL INTELLIGENCE AND likely to be open to polygyny as an experien- OPENNESS TO EXPERIENCE tially novel idea or action. In contrast, humans Research in personality psychology has have been mildly polygynous throughout their shown that one of the Five-Factor Model per- evolutionary history (Alexander et al. 1977; sonality factors — openness to experience — Leutenegger and Kelly 1977), and socially is significantly positively correlated with intel- imposed monogamy is a relatively recent his- ligence (Ackerman and Heggestad 1997). The torical phenomenon (Kanazawa and Still similarity and overlap between intelligence 1999). In other words, polygyny is evolution- and openness are apparent from the fact that arily familiar, whereas monogamy is evolu- some researchers call this personality factor tionarily novel. The Savanna-IQ Interaction ‘‘intellect’’ rather than ‘‘openness’’ (Goldberg Hypothesis would therefore predict that more 1992; McRae 1994). While it is widely intelligent individuals are more likely to be accepted by personality psychologists that open to monogamy and less likely to be open intelligence and openness covary across indi- to polygyny. In fact, as the empirical analysis viduals, it is not known why (Chamorro- below shows, more intelligent men are more Premuzic and Furnham 2006). The Savanna- likely to value monogamy and sexual exclu- IQ Interaction Hypothesis can potentially sivity than less intelligent men. explain why more intelligent individuals are As another example, for most contempo- more open to new experiences and are there- rary Americans, traditional names derived fore more prone to seek novelty. It is instruc- from the Bible, such as John and Mary, are tive to note from this perspective that only the actions, ideas, and values facets of openness to 1 experience are significantly correlated with There is much confusion about terminology for dif- ferent institutions of marriage, even among social scien- general intelligence, not the fantasy, esthetics, tists. Monogamy is the marriage of one man to one and feelings facets (Gilles, Stough, and woman. Polygyny is the marriage of one man to more than Loukomitis 2004; Holland et al. 1995). one woman, while polyandry is the marriage of one At the same time, the Hypothesis suggests woman to more than one man. Polygamy (although it is a possible need to refine the concept of nov- often used synonymously with polygyny in casual conver- sations) refers to both polygyny and polyandry. Because elty and to distinguish between evolutionary of its ambiguity, the word polygamy should not be used novelty (entities and situations that did not unless it specifically and simultaneously refers to both exist in the ancestral environment) and polygyny and polyandry. 38 SOCIAL PSYCHOLOGY QUARTERLY experientially more familiar than untradi- welfare of genetically unrelated others and tional names like Winner and Loser (Levitt the willingness to contribute larger propor- and Dubner 2005). So the Five-Factor Model tions of private resources for the welfare of may predict that more intelligent individuals such others. In the modern political and are more likely to name their children untra- economic context, this willingness usually ditional names like Winner and Loser than translates into paying higher proportions of less intelligent individuals. From the perspec- individual incomes in taxes toward the gov- tive of the Hypothesis, however, both John ernment and its social welfare programs. and Winner are equally evolutionarily novel Defined as such, liberalism is evolutionarily (because the Bible itself and all the traditional novel. Humans (like other species) are designed names derived from it are evolutionarily by evolution to be altruistic toward their genetic novel), so it would not predict that more intel- kin (Hamilton 1964), their repeated exchange ligent individuals are more likely to name partners (Trivers 1971), and members of their their children untraditional names. In fact, deme (a group of intermarrying individuals) there is no evidence at all that more intelli- or ethnic group (Whitmeyer 1997). They are gent individuals are more likely to prefer not designed to be altruistic toward an indefi- untraditional names for their children (Fryer nite number of complete strangers whom they and Levitt 2004; Lieberson and Bell 1992). are not likely ever to meet or exchange with. The Savanna-IQ Interaction Hypothesis This is largely because our ancestors lived in underscores the need to distinguish between a small band of 50 to 150 genetically related in- evolutionary novelty and experiential novelty. dividuals all their lives, and large cities and na- It can potentially explain why more intelligent tions with thousands and millions of people are individuals are more likely to seek evolution- themselves evolutionarily novel. ary novelty, but not necessarily experiential In order to make reasonable inferences novelty. It further suggests that the established about what values our ancestors might have correlation between openness and intelligence held during the course of human evolution, may be limited to the domain of evolutionary I have relied on two sources. First, I have novelty, not necessarily experiential novelty. consulted the ten-volume compendium The Encyclopedia of World Cultures (Levinson EVOLUTIONARILY NOVEL AND FAMILIAR 1991-1995), which extensively describes all PREFERENCES AND VALUES human cultures known to anthropology (more than 1,500) in great detail. Second, I have con- Liberalism sulted the following extensive (monograph- It is difficult to provide a precise definition length) ethnographies of traditional (hunter- of a whole school of political ideology like gatherer, pastoral, and horticultural) societies liberalism. Further, what passes as liberalism around the world: Yanomamo¨ (Chagnon varies by place and time. The Liberal Demo- 1992); From Mukogodo to Maasai: Ethnicity cratic Party in the United Kingdom is middle- and Cultural Change in Kenya (Cronk of-the-road, while the Liberal Democratic 2004); Ache Life History: The Ecology and Party in Japan is conservative. The political Demography of a Foraging People (Hill and philosophy which originally emerged as ‘‘lib- Hurtado 1996); The !Kung San: Men, Women, eralism’’ during the Enlightenment is now and Work in a Foraging Society (Lee 1979); called ‘‘classical liberalism’’ or ‘‘libertarian- and Sacha Runa: Ethnicity and Adaptation ism,’’ and represents the polar opposite of of Ecuadorian Jungle Quichua (Whitten what is now called ‘‘liberalism’’ in the United 1976). While contemporary hunter-gatherers States (Murray 1998). are not exactly the same as our ancestors dur- In this paper I will adopt the contemporary ing the Pleistocene, they are the best analog American definition of liberalism. I provi- that we have available for close examination sionally define liberalism (as opposed to con- and are thus often used for the purpose of servatism) as the genuine concern for the making inferences about our ancestral life. WHY LIBERALS AND ATHEISTS ARE MORE INTELLIGENT 39

These ethnographic sources make it clear Atheism that, while sharing of resources, especially food, is quite common and often normatively While religion is a cultural universal prescribed among hunter-gatherer tribes, and (Brown 1991), recent evolutionary psycho- while trade with neighboring tribes may logical theories (Atran 2002; Boyer 2001; have taken place (Ridley 1996), there is no Guthrie 1993; Haselton and Nettle 2006; evidence that people in contemporary Kirkpatrick 2005) suggest that religiosity hunter-gatherer bands freely share resources (belief in higher powers) may not be an adap- with members of other tribes. Because all tation in itself. It may instead be a byproduct members of a hunter-gatherer tribe are genetic of other evolved psychological mechanisms, kin or at the very least repeated exchange part- variously known as ‘‘animistic bias’’ (Guthrie ners (friends and allies for life), sharing re- 1993) or ‘‘the agency-detector mechanisms’’ sources among them does not qualify as an (Atran 2002). expression of liberalism as defined above. It When our ancestors faced some ambigu- may therefore be reasonable to infer that, ous situation, such as rustling noises nearby while sharing of food and other resources with at night or a large fruit falling from a tree genetic kin may be part of universal human branch and hitting them on the head, they nature, sharing of the same resources with could attribute it either to impersonal, inani- total strangers that one has never met or is mate, unintentional forces (wind blowing not likely ever to meet is not part of evolved gently to make the rustling noises among human nature. The Savanna-IQ Interaction the bushes and leaves, a mature fruit falling Hypothesis would therefore predict that more by its own weight from the branch by the intelligent individuals are more likely to force of gravity and hitting them on the espouse liberal political ideology than less head purely by accident) or to personal, ani- intelligent individuals. mate, intentional forces (a predator sneaking In an earlier study, Eaves and Eysenck up on them to attack, an enemy hiding in (1974) discover that political attitude (on the tree branches and throwing fruits at their the ‘‘radical-conservative’’ scale) has the her- head). itability of .65. More recently, Alford, Funk, Given that the situation is inherently and Hibbing (2005) show that roughly 43 per- ambiguous, our ancestors could have made cent of the variance in political attitudes on one of two errors of inference. They could the conservative-liberal dimension is deter- have attributed the events to intentional mined by genes, and parental socialization forces when they are in fact caused by unin- has a relatively minor role, accounting for tentional forces (false-positive or Type I only 22 percent of the total variance. In error) or they could have attributed them to a comprehensive meta-analysis, Jost et al. unintentional forces when they were in fact (2003) uncover a large number of personality caused by intentional forces (false-negative correlates with conservatism such as death or Type II error). The consequences of anxiety and intolerance of ambiguity. Their Type I errors were that our ancestors became study, however, does not include general unnecessarily paranoid and looked for pred- intelligence as a correlate of political attitude, ators and enemies where there were none. except that they show that openness to expe- The consequences of Type II errors were rience is negatively correlated with conserva- that our ancestors were attacked and killed tism, and we know from studies cited above by predators or enemies when they least sus- that openness correlates positively with intel- pected an attack. The consequences of com- ligence. Consistent with the prediction mitting Type II errors are far more derived from the Hypothesis, Deary, Batty, detrimental to survival and reproduction and Gale’s (2008a, b) recent studies show than the consequences of committing Type that more intelligent British children are I errors. Evolution should therefore favor more likely to become liberal adults. psychological mechanisms which predispose 40 SOCIAL PSYCHOLOGY QUARTERLY their carriers to commit Type I errors but conclude that atheism may not be part of avoid Type II errors, and thus overinfer the universal human nature, and widespread (rather than underinfer) intentions and practice of atheism may have been a recent agency behind potentially harmless phenom- product of Communism in the twentieth cen- ena caused by inanimate objects. Evolution- tury. The Hypothesis would therefore suggest arily speaking, it is good to be paranoid, that more intelligent individuals are more because it might save your life (Haselton likely to be atheist than less intelligent and Nettle 2006). individuals. Recent evolutionary psychological theo- ries therefore suggest that evolutionary origin Monogamy of religious beliefs in supernatural forces may stem from such an innate bias to commit Throughout human evolutionary history, Type I errors rather than Type II errors. The humans were mildly polygynous. A species- human brain may be biased to perceive inten- typical degree of polygyny correlates with tional forces (the hands of God at work) the extent of sexual dimorphism in size; the behind a wide range of natural physical phe- more sexually dimorphic the species (where nomena whose exact causes are unknown. If males are bigger than females), the more these theories are correct, then it means that polygynous the species (Alexander et al. religion and religiosity have an evolutionary 1979; Leutenegger and Kelly 1977). This is origin. It is evolutionarily familiar and natural either because males of polygynous species to believe in God, and evolutionarily novel become larger in order to compete with other not to be religious. males and monopolize females (Alexander Once again, in order to make reasonable et al. 1979; Leutenegger and Kelly 1977) or inferences about the religious beliefs of our because females of polygynous species ancestors during the course of human evolu- become smaller in order to mature early and tion, I have consulted the same primary eth- start mating (Harvey and Bennett 1985; nographic sources on which I relied to make Kanazawa and Novak 2005; Pickford 1986). inferences about their liberalism. Out of Thus strictly monogamous gibbons are sexu- more than 1,500 distinct cultures throughout ally monomorphic (males and females are the world described in The Encyclopedia of about the same size), whereas highly polygy- World Cultures,only19containanyreferen- nous gorillas are equally highly sexually ces to atheism. Not only do all these 19 cul- dimorphic in size. On this scale, humans are tures exist far outside of our ancestral home mildly polygynous, not as polygynous as in sub-Saharan Africa, but all 19 without an gorillas, but not strictly monogamous like exception are former Communist societies gibbons. (Abkhazians in Georgia, Ajarians in Georgia, Consistent with this comparative evi- Albanians, Bulgarians, Chuvash in Russia, dence, an analysis of the Standard Cross- Czechs, Germans in Russia [but not in Ger- Cultural Sample (Murdock 1967) shows many],GypsiesinRussia,ItelmeninRussia, that an overwhelming majority of traditional Kalmyks in Russia, Karakalpaks in Russia, culturesintheworld(83.39percent)practice Koreans in Russia (but not in Korea), Lat- polygyny, with only 16.14 percent practicing vians, Nganasan in Russia, Nivkh in Russia, monogamy and 0.47 percent practicing poly- Poles, Turkmens, Ukrainian peasants). There andry. Once again, while present-day hunter- are no non-former-Communist cultures gatherer societies are not the same as our described in The Encyclopedia as containing ancestors in every detail, they are our best any significant segment of atheists. Nor is analogs available to us to study. The fact there any reference to any individuals who that polygyny is widespread in such socie- do not subscribe to the local religion in any ties, combined with the comparative data of the monograph-length ethnographies cited discussed above, strongly suggests that our above. It may therefore be reasonable to ancestors might have practiced polygyny WHY LIBERALS AND ATHEISTS ARE MORE INTELLIGENT 41 throughout most of human evolutionary I will test these predictions derived from history. the Savanna-IQ Interaction Hypothesis with Under polygyny, one man is married to regard to the origin of preferences and values several women, so a woman in a polygynous using two separate representative data sets marriage still (legitimately) mates only with from the United States (Add Health and one man as a woman in a monogamous mar- GSS). riage does. In contrast, a man in a polygynous marriage concurrently mates with several women quite unlike a man in a monogamous STUDY 1 marriage who mates with only one woman. Method So throughout human evolutionary history, men have mated with several women while Data. In Study 1, I use the National Longitu- women have mated with only one man.2 dinal Study of Adolescent Health (Add Sexual exclusivity prescribed under socially Health). A sample of 80 high schools and 52 imposed monogamy today is therefore evolu- middle schools from the United States was tionarily novel for men, but not for women. selected with an unequal probability of selec- The Hypothesis would therefore suggest that tion. Incorporating systematic sampling meth- more intelligent men may value sexual exclu- ods and implicit stratification into the Add sivity more than less intelligent men, but Health study design ensures this sample is rep- intelligence may not affect women’s likeli- resentative of U.S. schools with respect to hood of espousing the value of sexual region of country, urbanicity, school size, exclusivity. school type, and ethnicity. A sample of 20,745 adolescents were personally inter- Evolutionarily Familiar Values viewed in their homes in 1994 through 1995 (Wave I) and again in 1996 (Wave II; n 5 Unlike liberalism, atheism, and monog- 14,738). In 2001 through 2002, 15,197 of the amy, values in such evolutionarily familiar original Wave I respondents, now aged 18 to entities as children, marriage (pair-bonding), 28, were interviewed in their homes. My sam- family, and friends should themselves be evo- ple consists of Wave III respondents in their lutionarily familiar. It has always been impor- early adulthood. For further details on the tant to value these entities throughout human design features of Add Health, see http:// evolutionary history; our ancestors who did www.asanet.org/journals/spq/health.cfm. not value these entities are not likely to Dependent variables. For liberal political have left many descendents. The Hypothesis ideology, I use the respondents’ response to would therefore predict that general intelli- the following question: ‘‘In terms of politics, gence may make no difference for the acqui- do you consider yourself conservative, lib- sition and espousal of these evolutionarily eral, or middle-of-the-road?’’ Their re- familiar values. sponses are coded as follows: 1 5 very conservative,25 conservative,35 middle 2 Of course, polygynous marriage in any society is of the road,45 liberal,55 very liberal. mathematically limited to a minority of men; most men For religiosity, I use the respondents’ in polygynous societies either have only one wife or no response to the following question: ‘‘To wife at all. However, at least some men throughout evolu- tionary history were polygynous, and we are dispropor- what extent are you a religious person?’’ tionately descended from polygynous men with a large Their responses are coded as follows: 1 5 number of wives (because they had more children than not religious at all,25 slightly religious, monogamous or wifeless men). Nor does the human evo- 3 5 moderately religious,45 very lutionary history of mild polygyny mean that women religious. Because both of these dependent var- always remained faithful to their legitimate husband. There is anatomical evidence to suggest that women have iables are measured on an ordinal scale, I use the always been mildly promiscuous (Baker and Bellis 1995; ordinal regression (McCullagh 1980) to esti- Gallup et al. 2003). mate these models. For sexual exclusivity, I 42 SOCIAL PSYCHOLOGY QUARTERLY use the respondent’s response to the following academic achievement. Stanovich, Cunning- question: ‘‘Using a scale from 1 to 10, where 1 ham, and Feeman’s (1984) study of first, third, means not important at all and 10 means and fifth graders shows that the correlation extremely important, how important do you between the PPVT and Raven’s Progressive think each of the following elements is for Matrices (which is widely regarded as the a successful marriage or serious committed best measure of general intelligence) is .22 relationship? Being faithful—that is, not (ns, n 5 56) among the first graders, .52 (p \ cheating on your partner by seeing other peo- .05, n 5 18) among the third graders, and .52 ple.’’ I use OLS regression to analyze this (p \ .05, n 5 20) among the fifth graders. It response. appears that the PPVT becomes a better mea- Independent variable. Add Health meas- sure of general intelligence as children get ures respondents’ intelligence with the Pea- older. body Picture Vocabulary Test (PPVT). The In order to establish the direction of cau- raw scores (0–87) are age-standardized and sality more clearly, I will use the measure converted to the IQ metric, with a mean of of intelligence taken in Wave I (in 1994– 100 and a standard deviation of 15. The 1995 when the respondents were in junior PPVT is properly a measure of verbal intelli- high and high school) to predict their adult gence, not general intelligence. However, values in Wave III (in 2001–2002 when the verbal intelligence is known to be highly cor- respondents are in their early adulthood). related with (and thus heavily load on) gen- Despite the fact that correlation between eral intelligence. Miner’s (1957) extensive measures of intelligence at Waves I and III review of 36 studies shows that the median (taken seven years apart) is not extremely correlation between vocabulary and general high (r 5 .5844, p \ .00001, n 5 13,943), intelligence is .83. Wolfle (1980) reports all of my substantive conclusions (and even that the correlation between a full-scale IQ the size of the coefficients) remain the same test (Army General Classification Test) and if I use Wave III’s measure of intelligence. the GSS synonyms measure (which I use later Control variables. In addition to adolescent in Study 2) is .71. As a result, the GSS syno- intelligence, I control for the following var- nyms measure has been used widely by intel- iables: age (even though there is very little ligence researchers to assess trends in general variance in it given that these are cohort intelligence (Huang and Hauser 1998), as data); sex (1 if male); race (with three dum- well as in research by sociologists who do mies for Asian, black, and Native American, not primarily study intelligence (Alwin and with white as the reference category, even McCammon 1999; Glenn 1999; Wilson and though Add Health respondents can choose Gove 1999). more than one racial category and a small With respect specifically to PPVT, Zagar proportion [4.15%] of them do); education and Mead’s (1983) hierarchical cluster analysis (years of formal schooling); earnings (in dol- of the Wechsler Intelligence Scale for Children- lars); and religion (with four dummies for Revised (WISC-R), the Peabody Individual Catholic, Jewish, Protestant, and other, Achievement Test (PIAT), the Beery Develop- with none as the reference category). For mental Test of Visual-Motor Integration (VMI) models predicting the value on sexual exclu- and the PPVT shows that the PPVT and the sivity, I also control for the number of times VMI, along with some components of the that the respondent has been married; 81.1 WISC-R, load on a first-order factor which percent of the Wave III respondents have they term ‘‘perceptual motor ability,’’ which never been married. in turn loads on a second-order factor which they term ‘‘general intelligence.’’ As Results a result, their conclusion is that the WISC-R, VMI, and PPVT are all good tests of general Figure 1a shows a clear monotonic bivariate intelligence, whereas the PIAT is a test of relationship between adolescent intelligence WHY LIBERALS AND ATHEISTS ARE MORE INTELLIGENT 43

Figure 1. Mean Adolescent Intelligence by Political Ideology and Religiosity Add Health Data, Wave III (2001–2002). Error bars indicate standard error of the mean. and adult political ideology. The higher the Figure 1b shows a similarly clear mono- intelligence of Add Health respondents in tonic bivariate relationship between adoles- junior high and high school, the more liberal cent intelligence and adult religiosity. The they grow up to be in their early adulthood. higher the intelligence of Add Health The mean adolescent intelligence of young respondents in junior high and high school, adults who identify themselves as ‘‘very lib- the less religious they grow up to be in their eral’’ is 106.42, while that of those who iden- early adulthood. The absolute difference in tify themselves as ‘‘very conservative’’ is mean adolescent intelligence between the 94.82. The differences in mean adolescent extreme categories of religiosity is not as intelligence by adult political ideology is great as that between the extreme categories highly statistically significant (F(4, 13053) 5 of political ideology. The mean adolescent 83.6327, p \ .00001). intelligence of young adults who identify 44 SOCIAL PSYCHOLOGY QUARTERLY themselves as ‘‘not at all religious’’ is 103.09, Table 1. The Effect of Adolescent Intelligence on while that of those who identify themselves Evolutionarily Novel Values (Liberalism and Atheism) as ‘‘very religious’’ is 97.14. The difference Add Health Data, Wave III (2001–2002) is still statistically very significant (F(3, Liberal Political 14273) 5 78.0381, p \ .00001). Ideology Religiosity Correlation between adolescent intelli- (1) (2) gence and the value on sexual exclusivity is r 5 .0572 (n 5 7,657) among women, and Adolescent .0158**** –.0121**** intelligence (.0015) (.0014) r 5 .0849 (n 5 6,756) among men. The corre- .2380 –.1821 lation is statistically significantly larger Age –.0131 –.0021 among men than among women (t 5 2.9774, (.0119) (.0112) n 5 14,413, p \ .01). –.0232 –.0037 Table 1 presents the results from the two Sex –.2524**** –.1822**** multiple ordinal regression models predicting (.0396) (.0372) adult political ideology and religiosity from –.1260 –.0910 Race adolescent intelligence and a set of control Asian –.0184 –.1173** variables. Column 1 shows that, net of age, (.0463) (.0444) sex, race, education, earnings, and religion, –.0096 –.0612 Add Health respondents who have higher ado- Black .1931**** .3463**** lescent intelligence are significantly more lib- (.0437) (.0411) eral in early adulthood than those who have .1115 .2000 lower adolescent intelligence (b 5 .0158, Native American –.1174* –.1292* (.0556) (.0529) p \ .0001, standardized coefficient 5 .2380). –.0551 –.0606 A comparison of standardized coefficients Education .0290** .0542**** (an unstandardized coefficient multiplied by (.0110) (.0104) the standard deviation of the independent .0750 .1404 variable to capture the effect of one standard Earnings –.0000** –.0000** deviation increase in the independent vari- (.0000) (.0000) able on the dependent variable) shows that –.0640 –.0470 Religion adolescent intelligence has a larger effect Catholic –.6143**** 2.3695**** on adult political ideology than any other (.0595) (.0618) variable in the model except for religion –.2655 1.0243 (Catholicism, Protestantism, and other reli- Jewish 1.1356**** 1.7496**** gion). Despite the fact that past studies show (.2236) (.2150) that women are more liberal than men (Lake .0976 .1503 and Breglio 1992; Shapiro and Mahajan Protestant –1.0957**** 2.9244**** (.0688) (.0709) 1986; Wirls 1986) and blacks are more –.3847 1.0266 liberal than whites (Kluegel and Smith Other –.8246**** 2.8744**** 1989; Sundquist 1983), the results here show (.0556) (.0598) that adolescent intelligence is twice as strong –.4036 1.4070 an influence on adult political ideology as Threshold sex (.2380 vs. –.1260) or being black (.2380 Y51 –2.7625 –.2491 vs. .1115). (.3085) (.2845) Y52 –.3552 1.8936 Table 1, Column 2 shows that, net of the (.3036) (.2857) same control variables, adolescent intelli- Y53 2.3153 4.0332 gence has an equally significant effect on (.3017) (.2867) adult religiosity (b 5 –.0212, p \ .0001, stan- Y54 4.8826 — dardized coefficient 5 –.1821). Add Health (.3119) — respondents who have higher adolescent (continued) WHY LIBERALS AND ATHEISTS ARE MORE INTELLIGENT 45

Table 1. continued Table 2. The Effect of Adolescent Intelligence on Evolutionarily Novel Values (Sexual Exclusivity for Liberal Political Men) Add Health Data, Wave III (2001–2002) Ideology Religiosity (1) (2) Sexual Exclusivity Men Women x2 goodness of fit 39500.1374 32834.1840**** Cox and Snell .0579 .2714 Adolescent .0038** .0006 2 pseudo R intelligence (.0012) (.0008) Number of cases 9,952 10,673 .0465 .0110 Age .0051 –.0074 Note: Main entries are unstandardized regression (.0100) (.0068) coefficients. Numbers in parentheses are standard errors. .0076 –.0163 Numbers in italics are standardized coefficients (b*sx). *p \ .05; **p \ .01; ***p \ .001; ****p \ .0001. Race Asian .0492 –.0067 (.0380) (.0247) intelligence are significantly less religious in .0227 –.0042 early adulthood than those who have lower Black –.1753**** –.0610** adolescent intelligence. Once again, despite (.0355) (.0236) the fact that past studies show that women –.0862 –.0409 are much more religious than men (Miller Native American .1174** .0140 (.0410) (.0340) and Hoffmann 1995; Miller and Stark 2002), .0519 .0067 adolescent intelligence is twice as strong an Education .0006 –.0031 influence on adult religiosity as sex (–.1821 (.0092) (.0061) vs. –.0910). It is remarkable that adolescent .0009 –.0078 intelligence is a significant and strong deter- Earnings –.0000 .0000 minant of adult religiosity when religion itself (.0000) (.0000) is controlled for (with no religion as the refer- –.0061 –.0044 Religion ence category). Catholic .1551*** .1468**** Table 2 presents the results for the value on (.0468) (.0332) sexual exclusivity, broken down by sex. The left .0582 .0822 column shows that, net of the same control var- Jewish –.1163 .0324 iables as before and the number of marriages, (.2003) (.1195) adolescent intelligence significantly increases –.0081 .0037 men’s value on sexual exclusivity; the more Protestant .2940**** .1337**** (.0559) (.0377) intelligent male Add Health respondents are in .0860 .0610 junior high and high school, the more value Other .2142**** .1367**** they grow to place on sexual exclusivity. In (.0436) (.0308) sharp contrast, the right column shows that ado- .0885 .0861 lescent intelligence has no significant effect on Number of marriages .1710*** .0245 women’s value on sexual exclusivity; more (.0444) (.0262) intelligent female Add Health respondents are .0557 .0134 Constant 8.9640 9.8878 no more likely to grow up to espouse a value (.2539) (.1636) on sexual exclusivity than their less intelligent R2 .0176 .0066 counterparts. The comparison of standardized Number of cases 5,263 5,480 coefficients shows that adolescent intelligence has more than four times as strong an effect Note: Main entries are unstandardized regression coefficients. Numbers in parentheses are standard errors. on the value for sexual exclusivity for men as Numbers in italics are standardized coefficients (betas). it does for women (.0465 vs. .0110). This is *p \ .05; **p \ .01; ***p \ .001; ****p \ .0001. 46 SOCIAL PSYCHOLOGY QUARTERLY consistent with the prediction from the Hypoth- Surveys (GSS) to examine the effect of intel- esis. However, the effect of adolescent intelli- ligence on the espousal of both evolutionarily gence on the value on sexual exclusivity is not novel values (liberalism and atheism) and as strong as its effects on liberalism or atheism.3 evolutionarily familiar values (on children, Because it is impossible to have a one stan- marriage, family, and friends). dard deviation increase in truly dichotomous The National Opinion Research Center at variables like sex or race (as a dummy), one the University of Chicago has administered may suggest that the proper comparison is the GSS either annually or biennially since between an unstandardized coefficient for the 1972. Personal interviews are conducted dichotomous variables (measuring the differ- with a nationally representative sample of ence between men and women and that non-institutionalized adults in the United between blacks and whites) and an unstandard- States. The sample size is about 1,500 for ized coefficient for adolescent intelligence each annual survey, and about 3,000 for multiplied by its standard deviation (to mea- each biennial one. The exact questions asked sure the change in the dependent variable asso- in the survey vary by the year. Some of the ciated with one standard deviation increase in following analyses includes samples from adolescent intelligence). These comparisons multiple years (when the relevant questions show that the effect of sex on liberal political are asked multiple times) or a sample from ideology is comparable to that of adolescent one year (when the relevant questions are intelligence (–.2524 vs. .2457), the effect of asked only once). being black on liberal political ideology is Dependent variables: Evolutionarily novel much smaller than that of adolescent intelli- values. For liberal political ideology, I use gence (.1931 vs. .2457), and the effect of sex the GSS respondents’ response to the follow- on religiosity is comparable to the effect of ing question: ‘‘We hear a lot of talk these adolescent intelligence (–.1822 vs. –.1882). days about liberals and conservatives. I’m going to show you a seven-point scale on which the political views that people might STUDY 2 hold are arranged from extremely liberal— Method point 1 to extremely conservative—point 7. Where would you place yourself on this Data. Because Add Health is cohort data scale?’’ Their response are reverse-coded as and includes only respondents in the same follows: 1 5 extremely conservative,25 generation and similar age group (in their ear- conservative,35 slightly conservative,45 lier twenties), the findings from them may or moderate,55 slightly liberal,65 liberal, may not generalize to all Americans across 7 5 extremely liberal. The GSS asks this ques- generations. In order to ascertain whether the effect of intelligence can be generalized tion in every survey year since 1974. For religiosity, I use two different ques- to all contemporary Americans in the last 30 tions. The first question asks: ‘‘Please look years, I now use data from the General Social at this card and tell me which statement comes closest to expressing what you believe about God: 1. I don’t believe in God; 2. I 3 While both height (Case and Paxson 2008; Jensen don’t know whether there is a God and I and Sinha 1993) and physical attractiveness (Kanazawa and Kovar 2004) are correlated with intelligence, entering don’t believe there is any way to find out; them as additional controls does not alter the substantive 3. I don’t believe in a personal God, but I findings on the effect of adolescent intelligence. Physical do believe in a Higher Power of some kind; attractiveness and height have no effect on liberalism; 4. I find myself believing in God some of physical attractiveness significantly (p \ .0001) in- the time, but not at others; 5. While I have creases, and height significantly (p \ .001) decreases, religiosity; physical attractiveness significantly (p \ doubts, I feel that I do believe in God; 6. I .05) increases men’s (but not women’s) value on sexual know God really exists and I have no doubts exclusivity while height has no effect for either sex. about it.’’ The GSS asks this question in WHY LIBERALS AND ATHEISTS ARE MORE INTELLIGENT 47 1988, 1991, 1993, 1994, 1998, and 2000. The score as a measure of verbal intelligence, second question asks: ‘‘Would you call your- which is known to be highly correlated with self a strong [respondent’s religion] or a not general intelligence (Huang and Hauser very strong [respondent’s religion]?’’ Their 1998; Miner 1957; Wolfle 1980). responses are coded as follows: 1 5 no reli- Control variables. In addition to intelli- gion,25 somewhat strong,35 not very gence, I control for the following variables: strong,45 strong. The GSS asks this ques- age (in years); sex (1 5 male); race (1 5 tion in every survey year since 1974. black); education (years of formal schooling); Dependent variables: Evolutionarily familiar earnings (measured in 12 to 23 equidistant values. For evolutionarily familiar values on ordinal categories, here treated as continu- children and marriage, I use the following ous); religion (with four dummies for Catho- question: ‘‘I’m going to read you a list of lic, Jewish, Protestant and other, with none as some things that different people value. the reference category); and survey year (only Some people say these things are very impor- for questions asked in multiple years). tant to them. Other people say they are not so In addition, for models predicting evolu- important. Please tell me how important each tionarily familiar values on marriage, children, thing is to you personally, using the response family, and friends, I control for whether on this card. How about having children? or not the respondent is currently married How about being married?’’ Their responses (1 5 yes), whether or not the respondent has are coded as follows: 1 5 not at all important, ever been married (1 5 yes), and the total 2 5 not too important,35 somewhat impor- number of children. tant,45 very important,55 one of the most important. The GSS asks these questions only Results in 1993. Table 3 presents the analysis of the GSS For evolutionarily familiar values on family data with respect to the evolutionarily novel and friends, I use the following question: ‘‘On values of liberalism and atheism. Column 1 these cards are various aspects of life. We shows that, net of age, sex, race, education, would like to know how important each of earnings, religion, and survey year, GSS these aspects of life is for you. On each of respondents who are more intelligent are these cards you see on the right hand side significantly more liberal than those who a scale with seven points. The lowest point are less intelligent (b 5 .0335, p \ .001, with number 1 indicates that this aspect of standardized coefficient 5 .0726). Columns life is unimportant to you. Point 7 at the top in- 2 and 3 show that, controlling for the same dicates that the particular aspect of life is very set of variables, more intelligent individuals importanttoyou.Thenumbersinbetween have a significantly weaker belief in God indicate varying degrees of importance. Please (b 5 –.1048, p \ .0001, standardized coeffi- take a look at all the cards first. Then, tell me cient 5 –.2271) and significantly less intense for each card its letter and the number you’ve religiosity (b 5 –.0283, p\.01, standardized decided on. One’s own family and children. 4 coefficient 5 –.0613). These results are Friends and acquaintances.’’ The GSS asks these questions only in 1982. Because all of the dependent variables are measured on ordi- 4 In the model predicting religious intensity (Column nal scales, I once again use ordinal regression. 3), inclusion of all four religion dummies simultaneously Independent variable. The GSS measures results in complete separation of data, and the maximum the verbal intelligence of its respondents by likelihood estimates thus do not exist. This is because the asking them to select a synonym for a word reference category then is ‘‘no religion,’’ while the lowest out of five candidates. Half of the respondents category of the dependent variable is also ‘‘no religion.’’ In order to avoid complete separation, I only enter three in each GSS sample answer 10 of these ques- religion dummies (Catholic, Jewish, Protestant), thus tions, and their total score (the number of cor- leaving the reference category ‘‘no religion and other rect responses) varies from 0 to 10. I use this religion.’’ 48 SOCIAL PSYCHOLOGY QUARTERLY

Table 3. The Effect of Intelligence on Evolutionarily Novel Values: General Social Survey Data, 1972–2004

Liberal Political Ideology Belief in God Religious Intensity (1) (2) (3)

Intelligence .0335*** –.1048**** –.0283*** (.0091) (.0215) (.0096) .0726 –.2271 –.0613 Age –.0098**** .0072* .0157**** (.0013) (.0030) (.0013) –.1714 .1259 .2745 Sex (15Male) –.1994**** –.5084**** –.2644**** (.0336) (.0774) (.0352) –.0989 –.2522 –.1311 Race (15Black) .5420**** .6619**** .3195**** (.0480) (.1360) (.0498) .1867 .2280 .1101 Education .0142* –.0249 .0563**** (.0070) (.0159) (.0073) .0450 –.0789 .1783 Earnings –.0148**** .0020 –.0090* (.0035) (.0073) (.0036) –.0866 .0117 –.0526 Religion Catholic –.7686**** 2.0418**** 4.5118**** (.0589) (.1247) (.0750) –.3312 .8798 1.9442 Jewish –.0137 .7157** 4.5643**** (.1224) (.2474) (.1340) –.0020 .1026 .6545 Protestant –1.0498**** 2.2861**** 4.6800**** (.0548) (.1154) (.0718) –.5123 1.1156 2.2838 Other –.4659**** 1.4427**** — (.1056) (.2085) — –.0807 .2500 — Year –.0095**** .0187* .0028 (.0023) (.0090) (.0022) –.0923 .1817 .0272 Threshold Y51 –23.7067 34.2737 7.5566 (4.5650) (17.8854) (4.4433) Y52 –21.6938 35.4099 9.2032 (4.5645) (17.8857) (4.4435) Y53 –20.7137 36.4118 11.6322 (4.5642) (17.8862) (4.4439) Y54 –19.0873 36.7681 — (4.5636) (17.8865) — Y55 –18.2036 37.9383 — (4.5634) (17.8872) — Y56 –16.2599 — — (4.5634) — — x2 goodness of fit 77390.5215 16874.8644**** 63737.5261**** Cox and Snell pseudo R2 .0571 .1807 .3494 Number of cases 13,034 3,200 13,904

Note: Main entries are unstandardized regression coefficients. Numbers in parentheses are standard errors. Numbers in italics are standardized coefficients (b*sx). *p \ .05; **p \ .01; ***p \ .001; ****p \ .0001. WHY LIBERALS AND ATHEISTS ARE MORE INTELLIGENT 49 identical to the results from the Add Health government to reduce the differences in data in Study 1 reported above. income between people with high incomes Table 4 presents the analysis of the GSS and those with low incomes’’ (r 5 .208, p \ data with respect to the evolutionarily famil- .00001, n 5 9,306), and with the statement iar values of children, marriage, family and ‘‘The government in Washington ought to friends. Columns 1 through 4 show that intel- reduce the income differences between the ligence does not have a significant effect on rich and the poor, perhaps by raising the taxes any of these evolutionarily familiar values of wealthy families or by giving income assis- (children: b 5 –.0486, ns, standardized coeffi- tance to the poor’’ and less likely to agree with cient 5 –.1053; marriage: b 5 –.0229, ns, the statement ‘‘The government should not standardized coefficient 5 –.0469; family: concern itself with reducing this income dif- b 5 .0611, ns, standardized coefficient 5 ference between the rich and the poor’’ (r 5 .1324; friends: b 5 –.0335, ns, standardized .217, p \ .00001, n 5 12,122). coefficient 5 –.0726). Now because all four However, even though more intelligent questions were asked only in a single year, GSS respondents are more likely to identify the sample size for all models in Table 4 are themselves as liberals (Table 3, Column 1), smaller than that for models in Table 3. The they are actually less likely to agree with the smaller sample size is not likely to be the main statement ‘‘It is the responsibility of the gov- reason for the lack of significant effects, how- ernment.’’ (r 5 –.236, p \ .00001, n 5 ever, because other variables included in these 5,849) or the statement ‘‘The government in models continue to have significant effects. Washington ought to reduce.’’ (r 5 –.167, For example, despite the small sample size, p \ .00001, n 5 5,814). Net of the same number of children has a highly significant demographic controls as in Table 3 (age, (p \ .0001) and very large (standardized sex, race, education, earnings, religion, and coefficient 5 .5829), if very predictable, survey year), intelligence is significantly neg- positive effect on the value of children. Less atively associated with agreement with the predictably, more educated individuals place first statement (b 5 –.147, p \ .00001) or asignificantly(p \ .05) and moderately the second statement (b 5 –.067, p \ (standardized coefficient 5 .2423) greater .00001) in multiple ordinal regression value on children, and older individuals place equations. asignificantly(p \ .01) and moderately In this paper, however, I provisionally (standardized coefficient 5 –.3532) smaller define liberalism as the genuine concern for value on them. the welfare of genetically unrelated others and the willingness to contribute larger pro- portions of private resources for the welfare WHAT DOES IT MEAN TO BE A LIBERAL IN THE of such others. There is some evidence in CONTEMPORARY UNITED STATES? the GSS data that self-identified liberals The results presented in Tables 1 and 3 indeed espouse these values. Reflecting their indicate that more intelligent individuals are genuine concern for the welfare of genetically more likely to identify themselves as liberal unrelated others, nonblack liberal GSS as opposed to conservative. But what does respondents are significantly more likely to such self-identification mean? Is it consistent agree with the statement ‘‘Blacks have been with the operational definition of liberalism discriminated against for so long that the gov- used in this paper? How are self-identified ernment has a special obligation to help liberals different from self-identified conser- improve their living standards’’ and less vatives? The GSS data can shed some light likely to agree with the statement ‘‘The gov- on these questions. ernment should not be giving special treat- Self-identified liberals in the GSS samples ment to blacks’’ (r 5 .209, p \ .00001, n 5 are significantly more likely to agree with the 19,290). Reflecting their willingness to con- statements ‘‘It is the responsibility of the tribute larger proportions of private resources 50 SOCIAL PSYCHOLOGY QUARTERLY

Table 4. The Effect of Intelligence on Evolutionarily Familiar Values: General Social Survey Data, 1972–2004

Children Marriage Family Friends (1) (2) (3) (4)

Intelligence –.0486 –.0229 .0611 –.0335 (.0419) (.0417) (.0563) (.0308) –.1053 –.0469 .1324 –.0726 Age –.0202** .0041 –.0175 .0187**** (.0066) (.0066) (.0090) (.0049) –.3532 .0717 –.3060 .3270 Sex (15Male) –.4279** .2724 –.5934** –.0222 (.1491) (.1481) (.2259) (.1203) –.2122 .1351 –.2943 –.0110 Race (15Black) –.4836* .0696 .2256 –.4491**** (.2440) (.2426) (.2492) (.1328) –.1666 .0240 .0777 –.1547 Education .0765* .0228 –.0307 –.0468 (.0316) (.0313) (.0471) (.0248) .2423 .0722 –.0972 –.1482 Earnings .0026 –.0162 .0288 –.0244 (.0142) (.0141) (.0245) (.0134) .0152 –.0947 .1684 –.1427 Religion Catholic .7397** .4360 1.1443** .3560 (.2583) (.2578) (.3488) (.2201) .3187 .1879 .4931 .1534 Jewish 1.1874* .5651 1.4268 .8562 (.5280) (.5197) (1.0875) (.4731) .1703 .0810 .2046 .1228 Protestant .7384** .7155** .7174* .4433* (.2341) (.2349) (.2928) (.2036) .3603 .3492 .3501 .2163 Other .3176 .2922 .8132 .5311 (.4836) (.4853) (.8806) (.5232) .0550 .0506 .1409 .0920 Currently married .5324** 2.1984**** .1620 –.3187* (.1901) (.2037) (.3117) (.1422) .2646 1.0924 .0805 –.1584 Ever married .6497** –.0574 1.3718**** –.2517 (.2517) (.2490) (.3435) (.1868) .2565 –.0227 .5416 –.0994 Number of children .3215**** –.0550 .1302 –.0881* (.0591) (.0570) (.0844) (.0365) .5829 –.0997 .2361 –.1597 Threshold Y51 –.8876 –.9053 –3.2897 –6.2999 (.5058) (.5066) (.7093) (.6301) Y52 .0126 .6433 –3.0873 –4.6761 (.5012) (.5007) (.7004) (.4444) Y53 1.2256 1.9560 –2.6374 –3.4213 (.5037) (.5064) (.6861) (.4027) Y54 3.1364 3.8336 –2.3133 –2.1552 (.5143) (.5203) (.6791) (.3893) (continued) WHY LIBERALS AND ATHEISTS ARE MORE INTELLIGENT 51

Table 4. continued

Children Marriage Family Friends (1) (2) (3) (4)

Y55 — — –1.6922 –1.0349 — — (.6712) (.3849) Y56 — — –.9300 .0546 — — (.6674) (.3836) x2 goodness of fit 2733.6806 2712.0050 6586.0304 6911.0090 Cox and Snell pseudo R2 .1835 .2719 .0749 .0545 Number of cases 683 683 1,125 1,126

Note: Main entries are unstandardized regression coefficients. Numbers in parentheses are standard errors. Numbers in italics are standardized coefficients (b*sx). *p \ .05; **p \ .01; ***p \ .001; ****p \ .0001. for the welfare of such others, liberals of all note that two recent studies, conducted in races are significantly more likely to state that another nation (the United Kingdom), using the amount of federal income tax that they pay valid measures of general intelligence, and is too low, rather than too high or about right entirely different indicators of liberalness (r 5 .048, p \ .00001, n 5 25,174). (nontraditional social attitudes and voting Consistent with the prediction derived for the Green Party or the Liberal Democratic from the Savanna-IQ Interaction Hypothesis, Party), reach the same conclusion that child- more intelligent nonblack GSS respondents hood intelligence increases adult liberal atti- are more likely to agree that the government tudes (Deary et al. 2008a, b). has a special obligation to help blacks (r 5 .060, p \ .00001, n 5 8,610) and more intel- CONCLUSION ligent GSS respondents of all races are more likely to state that the amount of their federal The Savanna-IQ Interaction Hypothesis, income tax is too low (r 5 .039, p \ .0001, derived from the logical conjunction of the n 5 12,463). Net of the same control varia- Savanna Principle and a theory of the evolu- bles, intelligence among nonblack respond- tion of general intelligence, suggests that ents is significantly positively associated more intelligent individuals may be more with agreement with the statement that the likely to acquire and espouse evolutionarily government has a special obligation to help novel values, such as liberalism, atheism, blacks (b 5 .054, p \ .001), and intelligence and, for men, sexual exclusivity, than less is significantly positively associated with the intelligent individuals, while general intelli- likelihood that the GSS respondents of all gence may have no effect on the acquisition races state that their federal income tax is and espousal of evolutionarily familiar values. too low (b 5 .052, p \ .001). Data from two large representative samples, It therefore appears that, while they are the National Longitudinal Study of Adolescent opposed to government-imposed income Health and the General Social Surveys, sup- transfers from the rich to the poor, more intel- port the predictions (although some of the ligent individuals are more liberal in the sense standardized coefficients, while highly statisti- that they exhibit genuine concern for the wel- cally significant, are relatively small in size). fare of genetically unrelated others (non- Both adolescent and adult intelligence predict blacks’ concern for the welfare of blacks), adult espousal of liberalism, atheism, and sex- and that they are willing to contribute larger ual exclusivity for men (but not for women), proportions of private resources for the wel- while intelligence is not associated with the fare of such others (in the forms of higher adult espousal of evolutionarily familiar values federal income tax). It is also instructive to on children, marriage, family, and friends. 52 SOCIAL PSYCHOLOGY QUARTERLY Given the high heritability of intelligence Savanna-IQ Interaction Hypothesis because (Jensen 1998:169-202), and the moderate her- many of the current beliefs in our society, itability of political attitudes (Alford et al. derived as they are from evolved human 2005; Eaves and Eysenck 1974) and religios- nature, are evolutionarily familiar. However, ity (Bouchard et al. 1999; Koenig et al. 2005), the alternative hypothesis should lead to the one alternative explanation for the effect of prediction that, because contemporary Amer- adolescent intelligence on adult political ican society is predominantly monogamous, ideology and religiosity is the genetic trans- more intelligent individuals should prefer mission of all three traits. Intelligent parents polygyny. As the results presented above beget intelligent children; liberal parents show, however, this is not the case; more beget liberal children; religious parents beget intelligent men are more likely to prefer religious children. monogamy and sexual exclusivity. Further, Such behavior genetic explanations, while both in the relatively more conservative cap- undoubtedly true, cannot explain the origin of italist United States and in the relatively more covariance between general intelligence and liberal socialist United Kingdom, more intel- certain values. Why do intelligent parents ligent individuals are more likely to be liberal tend simultaneously to be liberal and atheist, than less intelligent individuals (Deary et al. to pass on their genetic tendencies toward lib- 2008a, b). So it appears that it is not the pre- eralism and atheism to their intelligent chil- dominant values of society that affect the val- dren? Why are there not an equal (or ues of more or less intelligent individuals, but greater) number of intelligent parents who their evolutionary novelty. are conservative and/or religious, to pass on While it may be reasonable to posit that their conservative and religious tendencies liberalism, atheism, and sexual exclusivity to their intelligent children? Why are there for men may be evolutionarily novel values, not many less intelligent parents who are lib- they are far from the only ones. For example, eral and atheist? Further, behavior genetics while the contemporary American definition cannot explain why the value on sexual of left-wing liberalism may be the polar exclusivity (if heritable) is transmitted only opposite of genetic self-interestedness, nepo- to sons but not to daughters. The Savanna- tism, reciprocal altruism, and ethnocentrism IQ Interaction Hypothesis can offer one pos- of our ancestors, there are other political val- sible explanation for the coexistence of gen- ues that deviate from them, such as commu- eral intelligence and certain values. nism, monarchism, and libertarianism, to Another alternative hypothesis is that more name just a few. Future research would intelligent individuals are more likely not have to examine whether more intelligent in- necessarily to acquire and espouse evolution- dividuals are also more likely to adopt these arily novel values, as the Hypothesis predicts, evolutionarily novel political ideologies. but not to conform to others in the society What other values are evolutionarily (Millet and Dewitte 2007). It may be adaptive novel? Another such value is vegetarianism. for the survival of humans to adopt the rule Humans are naturally omnivorous, and ‘‘The less intelligent you are relative to those anyone who eschewed animal protein and around you, the more you should mimic their ate only vegetables in the ancestral environ- beliefs and actions.’’ Because the majority, ment, in the face of food scarcity and precar- under most reasonable conditions, is always iousness of its supply, was not likely to have mathematically more likely to be correct survived long and stayed healthy enough to than the average individual (as the Condorcet have become our ancestors. Vegetarianism Jury Theorem holds), such a decision rule would therefore be an evolutionarily novel may save the lives of individuals of less value, and the Hypothesis would predict than average intelligence. more intelligent individuals are more likely It is a bit difficult to separate this alterna- to choose to be a vegetarian than less intelli- tive hypothesis from the prediction of the gent individuals. WHY LIBERALS AND ATHEISTS ARE MORE INTELLIGENT 53 Consistent with this prediction, Gale et al. list of such evolutionarily novel values is (2007) find in their analysis of the 1970 Brit- potentially very long, the Hypothesis prom- ish Cohort Study that, net of sex, social class, ises to provide a theoretical explanation for and education, childhood intelligence at age why intelligent individuals acquire preferen- 10 significantly increases the probability ces and values in many different domains of that individuals become vegetarian as adults life. at age 30. In the Add Health Data, adolescent intelligence (in junior high and high school) REFERENCES significantly (p \ .05) increases the likeli- hood of vegetarianism in early adulthood, Ackerman, Phillip L. and Eric D. Heggestad. 1997. net of age, sex, race, education, and earnings. ‘‘Intelligence, Personality, and Interests: Evidence for Overlapping Traits.’’ Psychological Bulletin However, once I control for religion, the 121:219–45. effect of adolescent intelligence becomes only Alexander, Richard D., John L. Hoogland, Richard D. marginally significant (p \ .07). (Results Howard, Katharine M. Noonan, and Paul W. Sher- available upon request.) man. 1979. ‘‘Sexual Dimorphisms and Breeding The current work must be extended into Systems in Pinnipeds, Ungulates, Primates and Humans.’’ Pp. 402–35 in Evolutionary Biology several directions in the future. First, the and Human Social Behavior: An Anthropological Hypothesis must be tested against other com- Perspective, edited by Napoleon A. Chagnon and peting theories of the origin of values and William Irons. North Scituate, RI: Duxbury Press. preferences. Second, future empirical work Alford, John R., Carolyn L. Funk, and John R. Hib- must consider other evolutionarily novel and bing. 2005. ‘‘Are Political Orientation Genetically Transmitted?’’ American Political Science Review familiar values besides the ones considered 99:153–67. and tested in this paper. For example, in addi- Alwin, Duane F. and Ryan J. McCammon. 1999. tion to vegetarianism referred to above, the ‘‘Aging Versus Cohort Interpretations of Interco- Hypothesis would predict that more intelli- hort Differences in GSS Vocabulary Scores.’’ gent individuals are more likely to espouse American Sociological Review 64:272–86. Ash, Jessica and Gordon G. Gallup, Jr. 2007. ‘‘Pale- such other evolutionarily novel values as pac- oclimatic Variation and Brain Expansion During ifism (for men), feminism, or environmental- Human Evolution.’’ Human Nature 18:109–24. ism. Third, future empirical work must be Atran, Scott. 2002. In Gods We Trust: The Evolution- carried out with samples from nations other ary Landscape of Religion. Oxford, UK: Oxford than the United States, such as Gale et al.’s University Press. Bailey, Drew H. and David C. Geary. 2009. ‘‘Hominid (2007) paper discussed above, which uses Brain Evolution: Testing Climatic, Ecological, and a large British sample to test the effect of Social Competition Models.’’ Human Nature childhood intelligence on the acquisition of 20:67–79. vegetarianism. Baker, R. Robin and Mark A. Bellis. 1995. Human The origin of values and preferences Sperm Competition: Copulation, Masturbation and Infidelity. London, UK: Chapman and Hall. remains a very important theoretical puzzle Becker, Gary S. 1996. Accounting for Tastes. Cam- for social and behavioral sciences, and the bridge, MA: Harvard University Press. Savanna-IQ Interaction Hypothesis, at the Ben-Ner, Avner and Louis Putterman. 2000. ‘‘On intersection of evolutionary psychology and Some Implications of Evolutionary Psychology intelligence research, provides one deductive for the Study of Preferences and Institutions.’’ Journal of Economic Behavior and Organization explanation from theoretical first principles 43:91–99. for why individuals have certain values and Bouchard, Jr., Thomas J., Matt McGue, David preferences. By explaining how general Lykken, and Auke Tellegen. 1999. ‘‘Intrinsic and intelligence may interact with the evolution- Extrinsic Religiousness: Genetic and Environmen- ary constraints of the human brain, the tal Influences and Personality Correlates.’’ Twin Research 2:88–98. Hypothesis can account for the effect of Boyer, Pascal. 2001. Religion Explained: The Evolu- intelligence on the acquisition and espousal tionary Origins of Religious Thought. New York: of evolutionarily novel values. Because the Basic. 54 SOCIAL PSYCHOLOGY QUARTERLY

Brown, Donald E. 1991. Human Universals. New Emerson, Richard M.. 1987. ‘‘Toward a Theory of York: McGraw-Hill. Value in Social Exchange.’’ Pp. 11–46 in Social Burnham, Terence C. and Dominic D. P. Johnson. Exchange Theory, edited by Karen S. Cook. New- 2005. ‘‘The Biological and Evolutionary Logic bury Park, CA: Sage. of Human Cooperation.’’ Analyse & Kritik Fryer, Jr., Roland G. and Steven D. Levitt. 2004. ‘‘The 27:113–35. Causes and Consequences of Distinctly Black Case, Anne and Christina Paxson. 2008. ‘‘Stature and Names.’’ Quarterly Journal of Economics Status: Height, Ability, and Labor Market Out- 119:767–805. comes.’’ Journal of Political Economy 116:499– Gale, Catharine R., Ian J. Deary, Ingrid Schoon, and 532. G. David Batty. 2007. ‘‘IQ in Childhood and Veg- Caughey, John L. 1984. Imaginary Social Worlds: A etarianism in Adulthood: 1970 British Cohort Cultural Approach. Lincoln: University of Ne- Study.’’ British Medical Journal 334:245–8. braska Press. Gallup, Gordon G., Rebecca L. Burch, Mary L. Zap- Chagnon, Napoleon. 1992. Yanomamo¨, 4th ed. Fort pieri, Rizwan A. Parvez, Malinda L. Stockwell, Worth, TX: Harcourt Brace Jovanovich. and Jennifer A. Davis. 2003. ‘‘The Human Penis Chamorro-Premuzic, Tomas and Adrian Furnham. as a Semen Displacement Device.’’ Evolution 2006. ‘‘Intellectual Competence and the Intelli- and Human Behavior 24:277–89. gent Personality: A Third Way in Differential Gilles, Gignac E., Con Stough, and S. Sue Loukomiti, Psychology.’’ Review of General Psychology S. 2004. ‘‘Openness, Intelligence, and Self-Report 10:251–67. Intelligence.’’ Intelligence 32:133–43. Chiappe, Dan and Kevin MacDonald. 2005. ‘‘The Glenn, Norval D. 1999. ‘‘Further Discussion of the Evi- Evolution of Domain-General Mechanisms in dence for an Intercohort Decline in Education- Intelligence and Learning.’’ Journal of General Adjusted Vocabulary.’’ American Sociological Psychology 132:5–40. Review 64:267–71. Cosmides, Leda and John Tooby. 2002. ‘‘Unraveling Goldberg, Lewis R. 1992. ‘‘The Development of the Enigma of Human Intelligence: Evolutionary Markers for the Big-Five Factor Structure.’’ Psy- Psychology and the Multimodular Mind.’’ Pp. chological Assessment 4:26–42. 145–98 in The Evolution of intelligence, edited Gottfredson, Linda S. 1997. ‘‘Why g Matters: The by Robert J. Sternberg and James C. Kaufman. Complexity of Everyday Life.’’ Intelligence Mahwah, NJ: Lawrence Erlbaum. 24:79–132. Crawford, Charles B. 1993. ‘‘The Future of Sociobiol- Gottfredson, Linda S. and Ian J. Deary. 2004. ‘‘Intelli- ogy: Counting Babies or Proximate Mechanisms?’’ gence Predicts Health and Longevity, But Why?’’ Trends in Ecology and Evolution 8:183–6. Current Directions in Psychological Science Cronk, Lee. 2004. From Mukogodo to Maasai: Eth- 13:1–4. nicity and Cultural Change in Kenya. Boulder, Guthrie, Stewart Elliott. 1993. Faces in the Clouds: A CO: Westview. New Theory of Religion. New York: Oxford Uni- Deary, Ian J., Martha C. Whiteman, John M. Starr, versity Press. Lawrence J. Whalley, and Helen C. Fox. 2004. Hamilton, William D. 1964. ‘‘Genetical Evolution of ‘‘The Impact of Childhood Intelligence on Later Social Behavior.’’ Journal of Theoretical Biology Life: Following Up the Scottish Mental Surveys 7:1–52. of 1932 and 1947.’’ Journal of Personality and Hagen, Edward H. and Peter Hammerstein. 2006. Social Psychology 86:130–47. ‘‘Game Theory and Human Evolution: A Critique Deary, Ian J., G. David Batty, and Catharine R. Gale. of Some Recent Interpretations of Experimental 2008a. ‘‘Bright Children Become Enlightened Games.’’ Theoretical Population Biology Adults.’’ Psychological Science 19:1–6. 69:339–48. ———. 2008b. ‘‘Childhood Intelligence Predicts Harvey, P. H. and P. M. Bennett. 1985. ‘‘Sexual Voter Turnout, Voting Preferences, and Political Dimorphism and Reproductive Strategies.’’ Pp. Involvement in Adulthood: The 1970 British 43–59 in Human Sexual Dimorphism, edited by Cohort Study.’’ Intelligence 36:548–55. J. Ghesquiere, R. D. Martin and F. Newcombe. Eaves, L. J. and H. J. Eysenck. 1974. ‘‘Genetics and London, UK: Taylor and Francis. the Development of Social Attitudes.’’ Nature Haselton, Martie G. and Daniel Nettle. 2006. ‘‘The 249:288–9. Paranoid Optimist: An Integrative Evolutionary Ellis, Lee. 1998. ‘‘Neodarwinian Theories of Violent Model of Cognitive Biases.’’ Personality and Criminality and Antisocial Behavior: Photo- Social Psychology Review 10:47–66. graphic Evidence from Nonhuman Animals and Hechter, Michael, Lynn Nadel, and Richard E. a Review of the Literature.’’ Aggression and Vio- Michod, eds. 1993. The Origin of Values. New lent Behavior 3:61–110. York: Aldine de Gruyter. WHY LIBERALS AND ATHEISTS ARE MORE INTELLIGENT 55

Hechter, Michael, James Ranger-Moore, Guillermina ———. 2007. ‘‘Mating Intelligence and General Jasso, and Christine Horne. 1999. ‘‘Do Values Intelligence as Independent Constructs.’’ Pp. Matter? An Analysis of Advanced Directives for 283–309 in Mating Intelligence: Sex, Relation- Medical Treatment.’’ European Sociological ships, and the Mind’s Reproductive System, edited Review 15:405–30. by Geher Glenn and Geoffrey Miller. Mahwah, NJ: Herrnstein, Richard J. and Charles Murray. 1994. The Lawrence Erlbaum. Bell Curve: Intelligence and Class Structure in ———. 2008. ‘‘Temperature and Evolutionary Nov- American Life. New York: Free Press. elty as Forces behind the Evolution of General Hill, Kim and A. Magdalena Hurtado. 1996. Ache Life Intelligence.’’ Intelligence 36:99–108. History: The Ecology and Demography of a Forag- ———. 2009. ‘‘Evolutionary Psychology Crime.’’ ing People. New York: Aldine. Pp. 90–110 in Biosocial Criminology: New Direc- Holland, Daniel C., Stephen J. Dollinger, Cornelius J. tions in Theory and Research, edited by Anthony Holland, and Douglas A. MacDonald. 1995. ‘‘The Walsh and Kevin M. Beaver. New York: Relationship between Psychometric Intelligence Routledge. and the Five-Factor Model of Personality in Kanazawa, Satoshi and Jody L. Kovar. 2004. ‘‘Why a Rehabilitation Sample.’’ Journal of Clinical Beautiful People Are More Intelligent.’’ Intelli- Psychology 51:79–88. gence 32:227–43. Horne, Christine. 2004. ‘‘Values and Evolutionary Kanazawa, Satoshi and Deanna L. Novak. 2005. Psychology.’’ Sociological Theory 22:477–503. ‘‘Human Sexual Dimorphism in Size May Be Trig- Huang, Min-Hsiung and Robert M. Hauser. 1998. gered by Environmental Cues.’’ Journal of Bioso- ‘‘Trends in Black-White Test-Score Differentials: cial Science 37:657–65. II. The WORDSUM Vocabulary Test.’’ Pp. 303– Kanazawa, Satoshi and Mary C. Still. 1999. ‘‘Why 32 in The Rising Curve: Long-Term Gains in IQ Monogamy?’’ Social Forces 78:25–50. and Related Measure, edited by Ulric Neisser. Kirkpatrick, Lee A. 2005. Attachment, Evolution, and Washington, DC: American Psychological the Psychology of Religion. New York: Guilford. Association. Kluegel, James R. and Eliot R. Smith. 1986. Beliefs Jensen, Arthur R. 1998. The g Factor: The Science of About Inequality: Americans’ View of What Is Mental Ability. Westport, CT: Praeger. and What Ought to Be. New York: Aldine. Jensen, Arthur R. and S. N. Sinha. 1993. ‘‘Physical Koenig, Laura B., Matt McGue, Robert F. Krueger, Correlates of Human Intelligence.’’ Pp. 139–242 and Thomas J. Bouchard, Jr. 2005. ‘‘Genetic and in Biological Approaches to the Study of Human Environmental Influences on Religiousness: Find- Intelligence, edited by Philip A. Vernon. Nor- ings for Retrospective and Current Religiousness wood, CA: Ablex. Ratings.’’ Journal of Personality 73:471–88. Jost, John T., Jack Glaser, Arie W. Kruglanski and Lake, Celinda C. and Vincent J. Breglio. 1992. ‘‘Dif- Frank J. Sulloway. 2003. ‘‘Political Conservatism ferent Voices, Different Views: The Politics of as Motivated Social Cognition.’’ Psychological Gender.’’ Pp. 178–201 in The American Woman, Bulletin 129:339–75. 1992-93: A Status Report, edited by Paula Ries Kanazawa, Satoshi. 2001. ‘‘De Gustibus Est Disputan- and Anne J. Stone. New York: Norton. dum.’’ Social Forces 79:1131–63. Lee, Richard Borshay. 1979. The !Kung San: Men, ———. 2002. ‘‘Bowling with Our Imaginary Women, and Work in a Foraging Society. Cam- Friends.’’ Evolution and Human Behavior bridge, UK: Cambridge University Press. 23:167–71. Leutenegger, Walter and James T. Kelly. 1977. ‘‘Rela- ———. 2004a. ‘‘The Savanna Principle.’’ Manage- tionship of Sexual Dimorphism in Canine Size and rial and Decision Economics 25:41–54. Body Size to Social, Behavioral, and Ecological ———. 2004b. ‘‘General Intelligence as a Domain- Correlates in Anthropoid Primates.’’ Primates Specific Adaptation.’’ Psychological Review 18:117–36. 111:512–23. Levinson, David, ed. 1991-1995. Encyclopedia of ———. 2005. ‘‘An Empirical Test of a Possible Solu- World Cultures. Ten Volumes. Boston, MA: tion to ‘‘the Central Theoretical Problem of G.K. Hall. Human Sociobiology’’’’ Journal of Cultural and Levitt, Steven D. and Stephen J. Dubner. 2005. Frea- Evolutionary Psychology 3:249–60. konomics: A Rogue Economist Explores the Hid- ———. 2006a. ‘‘Why the Less Intelligent May Enjoy den Side of Everything. London, UK: Penguin. Television More than the More Intelligent.’’ Journal Lieberson, Stanley and Eleanor O. Bell. 1992. ‘‘Child- of Cultural and Evolutionary Psychology 4:27–36. ren’s First Names: An Empirical Study of Social ———. 2006b. ‘‘Mind the Gap.in Intelligence: Re- Taste.’’ American Journal of Sociology 98:511–54. examining the Relationship between Inequality McCullagh, Peter. 1980. ‘‘Regression Models for and Health.’’ British Journal of Health Psychology Ordinal Data.’’ Journal of the Royal Statistical 11:623–42. Society, Series B 42:109–42. 56 SOCIAL PSYCHOLOGY QUARTERLY

McRae, Robert R. 1994. ‘‘Openness to Experience: Stigler, George J. and Gary S. Becker. 1977. ‘‘De Gus- Expanding the Boundaries of Factor V.’’ European tibus Non Est Disputandum.’’ American Economic Journal of Personality 8:251–72. Review 67:76–90. Miller, Alan S. and John P. Hoffmann. 1995. ‘‘Risk Sundquist, James L. 1983. Dynamics of the Party Sys- and Religion: An Explanation of Gender Differen- tem, Revised Edition. Washington DC: Brookings ces in Religiosity.’’ Journal for the Scientific Study Institution. of Religion 34:63–75. Symons, Donald. 1990. ‘‘Adaptiveness and Adapta- Miller, Alan S. and Rodney Stark. 2002. ‘‘Gender and tion.’’ Ethology and Sociobiology 11:427–44. Religiousness: Can Socialization Explanations Tooby, John and Leda Cosmides. 1989. ‘‘Evolutionary be Saved?’’ American Journal of Sociology Psychology and the Generation of Culture, Part I: 107:1399–1423. Theoretical Considerations.’’ Ethology and Socio- Miller, Geoffrey F. 2000. ‘‘Sexual Selection for Indi- biology 10:29–49. cators of Intelligence.’’ Pp. 260–75 in The Nature ———. 1990. ‘‘The Past Explains the Present: Emo- of Intelligence, edited by Gregory R. Bock, Jamie tional Adaptations and the Structure of Ancestral A. Goode, and Kate Webb. New York: John Wiley. Environments.’’ Ethology and Sociobiology Millet, Kobe and Siegfried Dewitte. 2007. ‘‘Non- 11:375–424. conformity May Be Hidden Driver Behind Rela- Trivers, Robert L. 1971. ‘‘The Evolution of Reciprocal tion.’’ British Medical Journal 334:327–8. Altruism.’’ Quarterly Review of Biology 46: Miner, John B. 1957. Intelligence in the United States: 35–57. A Survey — with Conclusions for Manpower Utili- de Waal, Frans B. M. 1989. ‘‘Food Sharing and Recip- zation in Education and Employment. New York: rocal Obligations among Chimpanzees.’’ Journal Springer. of Human Evolution 18:433–59. Murdock, George P. 1967. Culture and Society. Pitts- ———. 1992. ‘‘Appeasement, Celebration, and Food burgh, PA: University of Pittsburgh Press. Sharing in the Two Pan Species.’’ Pp. 37–50 in Murray, Charles. 1998. What It Means to Be a Liber- Topics in Primatology: Human Origins,editedby tarian: A Personal Interpretation. New York: Nishida, Toshisada, William C. McGrew, and Peter Broadway. Marler. Tokyo, Japan: University of Tokyo Press. Neisser, Ulric, Gwyneth Boodoo, Thomas J. Bouchard, de Waal, Frans B. M., Lesleigh M. Luttrell, and M. Jr., A. Wade Boykin, Nathan Brody, Stephen J. Eloise Canfield. 1993. ‘‘Preliminary Data on Vol- Ceci, Diane F. Halpern, John C. Loehlin, Robert untary Food Sharing in Brown Capuchin Mon- Perloff, Robert J. Sternberg, and Susana Urbina. keys.’’ American Journal of Primatology 29:73–8. 1996. ‘‘Intelligence: Known and Unknowns.’’ Whitmeyer, Joseph M. 1997. ‘‘Endogamy as a Basis American Psychologist 51:77–101. for Ethnic Behavior.’’ Sociological Theory Pickford, Martin. 1986. ‘‘On the Origins of Body Size 15:162–78. Dimorphism in Primates.’’ Pp. 77–91 in Sexual Whitten, Norman E., 1976. Sacha Runa: Ethnicity and Dimorphism in Living and Fossil Primates, edited Adaptation of Ecuadorian Jungle Quichua. Ur- by Martin Pickford and Brunetto Chiarelli. Flor- bana: University of Illinois Press. ence, Italy: Il Sedicesimo. Wildavsky, Aaron. 1987. ‘‘Choosing Preferences by Ridley, Matt. 1996. The Origins of Virtue: Human In- Constructing Institutions: A Cultural Theory of stincts and the Evolution of Cooperation. New Preferences Formation.’’ American Political Sci- York: Viking. ence Review 81:3–21. Schwartz, Shalom H. 1992. ‘‘Universals in the Content Wirls, Daniel. 1986. ‘‘Reinterpreting the Gender and Structure of Values: Theoretical Advances and Gap.’’ Public Opinion Quarterly 50:316–30. Empirical Tests in 20 Countries.’’ Pp. 1–65 in Wilson, James A. and Walter R. Gove. 1999. ‘‘The Advances in Experimental Social Psychology, Intercohort Decline in Verbal Ability: Does It edited by Mark Zanna. New York: Academic Exist?’’ American Sociological Review 64:253–66. Press. Wilson, James Q. and Richard J. Herrnstein. 1985. Shapiro, Robert Y. and Harpreet Mahajan. 1986. Crime and Human Nature: The Definitive Study ‘‘Gender Differences in Policy Preferences: A of the Causes of Crime. New York: Touchstone. Summary of Trends from the 1960s to the Wolfle, Lee M. 1980. ‘‘The Enduring Effects of Edu- 1980s.’’ Public Opinion Quarterly 50:42–61. cation on Verbal Skills.’’ Sociology of Education Stanovich, Keith E., Anne E. Cunningham, and Doro- 53:104–14. thy J. Feeman. 1984. ‘‘Intelligence, Cognitive Zagar, Robert and John D. Mead. 1983. ‘‘Analysis of Skills, and Early Reading Progress.’’ Reading Short Test Battery for Children.’’ Journal of Clin- Research Quarterly 19:278–303. ical Psychology 39:590–7. WHY LIBERALS AND ATHEISTS ARE MORE INTELLIGENT 57 Satoshi Kanazawa is Reader in Management at the London School of Economics and Political Sci- ence, and Honorary Research Fellow in the Department of Psychology at University College London and in the Department of Psychology at Birkbeck College University of London. He has written over 80 articles and chapters in psychology, sociology, political science, economics, anthropology, and biology. He shares his evolutionary psychological observations in his popular blog, The Scientific Fundamentalist, at Psychology Today (http://www.psychologytoday.com/blog/the-scientific- fundamentalist). He is the author of Why Beautiful People Have More Daughters (Penguin 2007) and Escaping Biology: Why Intelligent People Are the Ultimate Losers in Life (tentative title) (Wiley 2011).