Parenting and Cultures of Risk: a Comparative Analysis of Infidelity

ROBERT J. QUINLAN MARSHA B. QUINLAN

Parenting and Cultures of Risk: A Comparative Analysis of Inﬁdelity, Aggression, and Witchcraft

ABSTRACT Parenting behavior may respond ﬂexibly to environmental risk to help prepare children for the environment they can expect to face as adults. In hazardous environments where child outcomes are unpredictable, unresponsive parenting could be adaptive. Child development associated with parenting practices, in turn, may inﬂuence cultural patterns related to insecurity and aggression (which we call the “risk-response model”). We test these propositions in a cross-cultural analysis. The Standard Cross-Cultural Sample (SCCS) includes indicators of parental responsiveness: father–infant sleeping proximity, father involvement, parental response to infant crying, and breastfeeding duration (age at weaning). Unresponsive parenting was associated with cultural models including greater acceptance of extramarital sex, aggression, theft, and witchcraft. Socialization practices in later childhood were not better predictors of the outcomes than was earlier parenting. We conclude that some cultural adaptations appear rooted in parenting practices that affect child development. [Keywords: life history, behavioral ecology, personality development, cross-cultural research, cultural transmission]

HILDHOOD EXPERIENCE SHAPES ADULT BEHAV- effort to achieve an adaptive balance with their environ- CIOR, but pathways to adulthood are poorly un- ment (Roff 2002). When should we stop growing and start derstood. Parenting, child development, and cultural reproducing to enhance fitness—genetic representation in models can be integrated in life history theory. In this per- future generations (Walker et al. 2006)? When should we spective, parental care prepares children for an expected stop reproducing (Hawkes et al. 1998)? And how should adult environment. Development targets divergent strate- we best care for our offspring to improve their chance gies that are adaptive in different circumstances (Belsky of success, given local conditions (Clutton-Brock 1991)? et al. 1991; Chisholm 1999b; Draper and Harpending 1982). Anthropologists and psychologists have recently applied These strategies lie along a continuum from mating to par- life history theory to answer these questions for humans enting effort. Mating effort emphasizes early reproduction, (Belsky et al. 1991; Borgerhoff Mulder 1992; Chisholm fluid pair-bonds, low-intensity parental care, and generally 1999b; Cronk 2000; Ellis et al. 2003; Geary 2002; Hawkes risk-prone behavior. These traits may be adaptive in envi- et al. 1998; Hewlett et al. 2000b; Hill and Hurtado 1996; ronments in which relationships are unstable and risky so- Kaplan and Lancaster 2003; Kaplan et al. 2000; Mace 2000b; cial behavior is rewarded with cultural success.1 Conversely, Pennington and Harpending 1993; Quinlan 2003; Walker parenting effort emphasizes delayed reproduction, stable et al. 2006; Worthman 1999). These applications are be- pair-bonds, intensive parenting, and generally risk-averse ginning to have an impact on medicine and public health behavior: a suite of traits that may be adaptive in environ- (e.g., Campbell in press; Chisholm and Burbank 2001; Gant ments with stable relationships and little reward for risky et al. in press; Hurtado et al. 2006; McDade 2003; Sugiyama behavior. We suggest that parenting affects child develop- 2004; Weinstein and Ciszek 2002; Wilson and Daly 1997). ment, which, in turn, shapes adaptive cultural models for Here we use a life history approach to explore cross-cultural relationships and risk taking. variation related to environmental risk. Life history thinking is a massively influential develop- Humans show flexible responses to local conditions. ment in evolutionary biology that can potentially integrate Theory suggests that humans respond to environmental large and seemingly disparate bodies of knowledge—an ul- risk by adjusting their parenting behaviors (Quinlan in timate goal of science. The premise is that all organisms press). Basic parental care should usually benefit children, face a series of trade-offs in the allocation of energy and but there are diminishing returns to parental effort. For

AMERICAN ANTHROPOLOGIST, Vol. 109, Issue 1, pp. 164–179, ISSN 0002-7294 online ISSN 1548-1433. C 2007 by the American Anthropological Association. All rights reserved. Please direct all requests for permission to photocopy or reproduce article content through the University of California Press’s Rights and Permissions website, http://www.ucpressjournals.com/reprintInfo.asp. DOI: 10.1525/AA.2007.109.1.164. Quinlan and Quinlan • Parenting and Cultures of Risk 165 example, there is point at which more food does not im- predicted to affect behavioral “strategies” later in life. They prove children’s well-being, and once parents reach that argued that pair-bond stability and father involvement are “saturation point” they should turn their attention to other reliable indicators of local resource stability and the long- resources (see Harpending et al. 1990). Certain kinds of risk term benefits of parental care. Their model is based in can also create substantial diminishing returns to parental parental investment theory (Clutton-Brock 1991; Trivers effort. If there are hazards in the environment that par- 1972). Parental investment is defined as expenditures (cash, ents cannot reduce or avoid by increasing the amount or time, attention, etc.) benefiting one offspring at a cost to intensity of childcare, then parents might enhance their parents’ ability to invest in other offspring or reproductive fitness by focusing “effort” on mating rather than parent- opportunities (Clutton-Brock 1991:9). The theory predicts ing: Parental effort may be wasted in the face of unpre- father abandonment in environments in which paternal dictable hazards. By shifting from parenting to mating effort care does not benefit men’s fitness (Trivers 1985). For exam- an individual can produce more offspring, increasing the ple, where avuncular relationships are important for chil- chance that at least some will succeed despite the hazards dren, investment from fathers may be less crucial; hence, (Quinlan in press). This low-intensity childcare may gener- conjugal relations may be unstable and relatively unimpor- ate a developmental “trajectory” in children that gears them tant (e.g., Goody 1969:39–90; Gough 1959) as men pursue toward life in a risky and uncertain environment. The de- other avenues to enhance fitness. Draper and Harpending velopmental trajectory could lead to adults with psycholog- also draw inspiration from the Whitings’ classic study of ical dispositions geared toward a high-risk strategy includ- aloofness and intimacy of husbands and wives (Whiting ing multiple sex partners, aggressive behavior, hostility, and and Whiting 1975). “Aloofness” between women and their so forth. Although each individual has some unique com- mates may influence socialization practices that shape adult bination of experiences, members of a community share behavior. Hence, an aloof relationship between parents acts many aspects of the environment that may shape general as a cue for a child to develop a mating effort strategy. features of their psyche. We suggest that shared psycholog- Conversely, an intimate relationship between parents cues ical tendencies can support cultural models that are geared development of a parental effort strategy. If childhood en- toward risk taking including acceptance of extramarital sex, vironment reliably predicts the demands of the adult en- aggression, and hostile “social cognition” such as beliefs in vironment, then developmental “canalization” is adaptive. witchcraft. At the individual level, risk-taking psychologi- Although Draper and Harpending’s basic hypothesis was cal tendencies are associated with “mating effort strategies”; innovative and integrative, both the theory and potential at the population level, such tendencies may be associated mechanisms needed further specification. Subsequent ad- with “cultures of risk.” vances in life history theory and developmental research Are parenting practices associated with cultural mod- have filled in gaps. els related to risk taking and sexual behavior? This ques- Jay Belsky and colleagues (1991) and James Chisholm tion is reminiscent of John Whiting and Irvin Child (1953) (1999b) refined the Draper-Harpending model and hu- and other classic works (Rohner 1975; Whiting and Whit- man life history theory by suggesting that parent–child ing 1975), and we owe much inspiration to those ground- attachment (Ainsworth et al. 1978; Bowlby 1969) medi- breaking efforts. Those studies dealt with descriptive the- ates between early father involvement and adult behavior. ory using simple analyses. In particular, predictor variables Generally, attachment style refers to relatively stable “inter- such as parental “warmth” (Rohner 1975:194–196) or “sat- nal working models of social relationships” (Hewlett et al. isfaction potential” (Whiting and Child 1953:50–52) in ear- 2000a; Pietromonaco and Barrett 2000) or “mental repre- lier studies make it difficult to suggest pathways by which sentations” of social relationships (Hughes et al. 2005). An parenting affects development. Recent theoretical and an- insecure internal model views social relationships as unsta- alytical advances, however, allow for better specified and ble, exploitative, and based in mistrust; psychologists re- more complex models. We offer a socioecological approach fer to such mental representations as “hostile social cog- in which parents’ specific responses to environmental risk nition” (e.g., Guyll and Madon 2003). Conversely, secure in early childhood encodes adaptive behavioral strategies attachment includes stable relationships that are support- that, in turn, generate cultural models. We call this the “risk ive and based in trust. Elements of attachment theory may response” hypothesis. Here we outline the historical devel- reflect Western culture models (Rothbaum et al. 2000); how- opment of these ideas, present a set of predictions, and ever, the notion that aspects of parent–child interaction test those predictions with data from the Standard Cross- might promote development of hostile social cognition is Cultural Sample (SCCS) (Murdock and White 1969). not controversial (Rothbaum and Morelli 2005:108–113). Cross-cultural variation in parental availability and responsiveness to children’s needs is a promising avenue for inves- LIFE HISTORY, PARENTING, AND CHILD DEVELOPMENT tigation. In an early influential work, Patricia Draper and Henry Father absence may create a family environment in Harpending (1982) proposed that humans have psycho- which mothers are more taxed and, hence, less respon- logical mechanisms that use cues in the childhood fam- sive to their offspring (Belsky et al. 1991; Chisholm ily environment to shape development of adult reproduc- 1999b). Maternal unresponsiveness during a sensitive pe- tive behavior. Father involvement during development was riod (birth to seven years of age) results in insecure internal 166 American Anthropologist • Vol. 109, No. 1 • March 2007

FIGURE 1. Significant cross-cultural associations between extrinsic risk and parenting in the SCCS (adapted from Quinlan in press). models, a mating effort strategy, and unstable conjugal (and more on stochastic factors than on individual competence, other) relationships (Belsky et al. 1991). Conversely, secure then people are predicted to reproduce early and invest rel- attachment leads to secure internal representations, likely atively little in many offspring in hopes of beating the odds. stable conjugal unions, and a parental effort strategy (Belsky The nature of risk in an environment should influ- et al. 1991). ence an individual’s strategic choices. “Risk” can be de- J. Belsky (1998; see also Belsky et al. 1991) further sug- fined as “unpredictable variability in the outcome of an gests that parental unresponsiveness promotes the devel- adaptively significant behavior” (Winterhalder and Leslie opment of “externalizing behaviors,” which often involve 2002:61). We call this “extrinsic” risk comparable to “ex- delinquency, violence, and crime, risky behaviors geared trinsic mortality” in life history theory. Extrinsic mortality is toward exploiting social and material resources in unsta- independent of parental care, which means that changes in ble and unpredictable environments (see also Chisholm parental behavior do not reduce extrinsic mortality (Stearns 1999b). A longitudinal study, for example, demonstrates 1992:182), creating a special set of pressures (Borgerhoff- that negative and unresponsive maternal care is associ- Mulder 1992; Chisholm 1999b; Harpending et al. 1990; ated with more externalizing behaviors among toddlers, Quinlan 2006, in press). Local extrinsic risk may influence and positive and responsive paternal care is associated with parenting behaviors.2 If extrinsic risk is high, then parental inhibition (Belsky 1998). Similarly, a study of Anglo and effort might be wasted, because child outcomes are de- East Indian British families found that parental responsive- termined by chance rather than parenting (e.g., Scheper- ness was inversely associated with externalizing behaviors Hughes 1992). Hence, high fertility and lower levels of in both ethnic groups (Deater-Deckard et al. 2004). Exter- parental investment per child are predicted to enhance par- nalizing and lack of inhibition in early childhood may es- ents’ fitness. But if extrinsic risk is low, then more chil- tablish a developmental trajectory toward risk-taking later dren survive to reproduce, and responsive parenting can in life. We expect such differences in development to shape promote children’s ultimate success as a member of their cross-cultural trends. society. If knowledge, skill, and social competence lead to Finally, because developing systems are especially sensi- success (Flinn 2004; Kaplan et al. 2000), then low fertility tive to early influences (West-Eberhard 2003:89–139), there and higher levels of parental investment are predicted to en- may be a sensitive period in infancy and early child- hance parents’ long-term fitness (Mace 2000a). Consistent hood (birth to seven years) for these effects. In fact, re- with these predictions, a recent cross-cultural study showed cent research shows that early childhood environment had reduced maternal care, paternal involvement, and breast- stronger effects on subsequent conjugal stability and re- feeding duration in societies with high extrinsic (unavoid- productive behavior than did later environment (Quinlan able) risk including pathogen stress, famine, and warfare 2003). (Quinlan in press; see Figure 1). Cross-cultural associations between extrinsic risk and parenting were not confounded by social complexity, economy, subsistence strategy, popu- CROSS-CULTURAL APPROACH TO RISK RESPONSE lation size, division of labor by sex, male and female con- Life history reflects two basic “decisions” (Roff 2002): tribution to subsistence, community structure, parenting whether to reproduce now or later, and the amount of care style, weather patterns, latitude, world region, accultura- to invest in each offspring. These two decisions may be as- tion, modernization or contact with “Western-industrial” sociated with a broad suite of behaviors involving conju- populations (Quinlan in press; see Figure 1). gal unions, interpersonal relations, and risk-taking. “Choos- Other research shows some of the predicted rela- ing” a strategy (whether consciously or not) depends on the tions between parenting and life history development costs and benefits of alternatives. When local success de- (see Table 1). Stressful family environments and low lev- pends on high levels of skill and social competence, then els of parental care are associated with development of an people are predicted to delay reproduction and invest a lot orientation toward mating effort behaviors; stable families in a few offspring. When local success depends as much or and higher levels of parental care are associated with Quinlan and Quinlan • Parenting and Cultures of Risk 167

TABLE 1. Studies Showing Predicted Associations between Parenting Variables and Child Outcomes in Various Populations

Parenting variable Child outcome Population Reference

Father absence unstable conjugal relationships U.S.; Hungary; Dominica, Amato 1999, 1996; Bereczkei and W.I. Csanaky 1996; Quinlan 2003; Quinlan and Flinn 2003 multiple sex partners U.S. Barber 1998; Quinlan 2003 accepting attitude to casual sex U.S. Barber 1998 insecure attachment U.S. Barber 1998 delinquency and behavior Hungary; U.S. Bereczkei and Csanaky 1996; problems (African-American and Matsueda and Heimer 1987 White) “deviance” U.S. Dornbusch et al. 1985 aggression and competition U.S. Thornton and Camburn 1987 among boys unstable relationships for boys U.S. Beaty 1995 less willing to maintain conjugal Belize (Maya) Waynforth et al. 1998 relationship teen pregnancy U.S. Quinlan 2003 early age at ﬁrst sex U.S. Quinlan 2003 aggression Cross-cultural analysis Ember and Ember 2002 Parental relationship strained insecure attachment U.S. Chisholm 1999a multiple sex partners U.S. Chisholm 1999a Family stress insecure attachment Australia Chisholm et al. 2005 unstable conjugal relationships Antigua, W.I. Handwercker 2003 multiple sex partners Australia; Antigua, W.I. Chisholm et al. 2005 Insecure parent–child hostile social cognition U.S. Simons et al. 2001 attachment aggressive behavior U.S. Simons et al. 2001 Unresponsive parental care “externalizing” behavior U.S.; U.K. (Anglo and East Belsky 1998; Deater-Deckard et al. Indian) 2004 Lack of parental warmth “externalizing” behavior Cross-cultural review Rohner et al. 2005 teen pregnancy New Zealand; U.S. Ellis et al. 2003 early age at ﬁrst sex New Zealand; U.S. Ellis et al. 2003 insecure attachment Cross-cultural review Rohner et al. 2005

development of an orientation toward parental effort be- Second, we suggest that ethnographic data for attitudes haviors (Chisholm et al. 2005; Ellis et al. 1999; Quinlan about extramarital sex are likely to be much more reliable 2003). The nature of parental care during childhood may in- than qualitative estimates of the frequency of extramarital dicate appropriate adult strategies and affect developmental sex (available in the SCCS), which are likely to suffer many pathways tuned to socioenvironmental risk. In sum, Table 1 methodological vagaries. presents associations from many studies that are consistent We operationalize insecurity and hostile social cogni- with the risk-response model. The table does not present tion as “importance of witchcraft.” Witchcraft gauges at- all of the relevant research, but it is suggestive of important tachment security because it taps into similar, analytically effects that warrant further cross-cultural analysis. useful, psychological processes. Here we are concerned with We expect associations between early childhood envi- the attribution of hostile intent to others as one element of ronment and subsequent development in individual level so called “attachment style” related to parenting. We follow studies to show broad cross-cultural trends. Extrinsic risk G. Murdock and colleagues who deﬁne witchcraft as “the affects community-level parenting tendencies (Quinlan in ascription of the impairment of health to the suspected vol- press) that are predicted to shape shared ideas concern- untary or involuntary aggressive action of a member of a ing social relations, including conjugal bonds, and more special class of human beings believed to be endowed with general risk-taking behavior (see also Hewlett et al. 2000a). a special power and propensity for evil” (1978:456). Accu- Data from the Standard Cross-Cultural Sample (Murdock sations and hints of witchcraft and related “fear of others” and White 1969) can be used to test this hypothesis. appear to have a “paranoid” quality (Whiting and Child We operationalize conjugal stability as “acceptance of 1953:269–270), and such attribution of hostile intentions extramarital sex,” which simultaneously captures attitudes has been associated with insecure parent–child relations about sex with multiple partners and the strength of (Simons et al. 2001). Classic studies suggest that witchcraft spousal attachments. In fact, extramarital affairs are the notions require uncertainty and mistrust of others (Nash most frequent cause of divorce cross-culturally (Betzig 1967:128–129): “You cannot easily tell a witch and never 1989). Attitudes about extramarital sex offer a nearly ideal know who may be one” (Middleton 1967:60). And after measure for our analysis. First, it allows us test whether a witch commits his evil act, “later he returns to console risk-response predictions are consistent with cross-cultural with kind words, but he is gleeful in his heart” (Middleton data concerning models of acceptable sexual behavior. 1967:60). Who may be a witch in a particular society is 168 American Anthropologist • Vol. 109, No. 1 • March 2007

TABLE 2. Multiple Regression Analysis Showing Effects of Parenting on Acceptance of Extramarital Sex

Unstandardized Log Log odds Model Stats Model Variables Beta 1-tail p Odds 1-tail p R2 Np

1 Intercept 6.94 0.000 Father’s sleeping proximity to infant −0.45 0.004 −1.08 0.001 Father involvement in early childhood −1.80 0.005 −4.67 0.001 0.32 42 .000 Father involvement in early childhood squared 0.25 0.010 0.65 0.004 Parental responsiveness to crying −0.30 0.051 −1.62 0.053 Age at weaning 0.08 0.141 0.92 0.180 2∗∗ Intercept 7.46 0.000 Father’s sleeping proximity to infant −0.45 0.001 −1.19 0.000 Father involvement in early childhood −1.59 0.024 −4.41 0.002 0.34 45 .001 Father involvement in early childhood squared 0.22 0.034 0.62 0.005 Parental responsiveness to crying −0.39 0.006 −0.95 0.002 Inculcation of sexual restraint in late childhood −0.01 0.371 −0.02 0.387

Note. Beta represents the effect of a change of one level in the predictor variable on the outcome variable after controlling for other variables in the model. R2 is the proportion of variance in the outcome variable in the linear model accounted for by all the predictor variables combined. 1-tail p is based on robust estimates of the standard errors clustered by region (SCCS variable 1858). Log odds estimates refer to the probability of being in the next higher level of the outcome variable from an ordinal logit model used because the outcome variable has only four levels (see Appendix). ∗∗Age at weaning was removed because n/k < 10 (see Appendix). not relevant. The importance of witchcraft here is as an good indicator of direct parental investment (see Quinlan explanation for misfortune that requires hostile social cog- et al. 2003, 2005). nition. Operationizing insecure internal representations and hostile social cognition as importance of witchcraft allows us to tap into normative reasoning about suspicious METHODS actions and intent of others in the community. It also al- Data are from George Murdock and Douglas White’s (1969) lows us to extend the risk-response model to understand Standard Cross Cultural Sample (SCCS), containing 186 the construction of larger features of ideational culture. mostly “nonindustrial” or “traditional” cultures. The anal- Finally, we operationalize risk-taking and “externaliz- ysis uses information for 37 to 55 societies (depending on ing” behaviors as prevalence of aggression including assault, the analyses) from the SCCS. We used previously published homicide, and theft. These are behaviors that form an im- codes (Divale 2004), which were recoded or combined in portant part of the “externalizing” construct used in psy- some cases. Details for all variables are presented in the chological studies in Western populations. Furthermore, vi- appendix. olence and theft are usually very high-risk behaviors, which We used multiple linear regression (and ordinal logit) invite retribution or otherwise severe consequences. models to explore relations between parenting behavior and Parental “responsiveness” in early childhood, we sug- cross-cultural models. The strategy is to simultaneously ex- gest, is the behavioral mechanism underlying these devel- amine associations between four parenting variables (father opmental effects (see also Belsky 1997; Chisholm 1996; Ellis sleeping location, father involvement in early childhood, et al. 2003; Quinlan 2003). Fathers have multiple roles: One parents’ response to infant crying, and age at weaning) and is to provide direct care (e.g., Hewlett 1991), and another is cultural models of extramarital sex, aggression, theft, and to support responsive maternal care by freeing the mother witchcraft. The risk-response model predicts that higher lev- from tasks that conflict with childcare (Quinlan and Quin- els of parental care will be associated with less importance lan in press; Quinlan et al. 2003). We have chosen measures of each of the four outcomes. For example, in cultures in of normative behavior rather than more abstract constructs which parents are more responsive to infant crying and fa- based on multiple attributes (e.g., Rohner 1975). Our intent thers are relatively close to children, we expect extramarital is to capture aspects of parenting behavior more than “par- sex, aggression, and so forth to be relatively unimportant. enting style.” We use fairly direct ethnographic measures Statistically, these relationships are expressed as “Unstan- for father involvement (proximity maintenance) in early dardized Beta” in Tables 2 through 6, which is the effect childhood and parental response to infant crying. As with that one unit or level of increase in the predictor variable most cross-cultural research, we are forced to deal with some has on the outcome variable. An unstandardized Beta of imperfect proxy measures (Ember and Ember 2001). Father .70, for example, indicates that an increase in the average sleeping proximity to infants is a proxy of father availabil- value of a predictor variable (in this case parenting) is asso- ity, and Carol Ember and Melvin Ember (2002) suggest that ciated with an increase of .7 units in the average value of is the best available cross-cultural measure of father involve- the outcome variable (e.g., acceptance of extramarital sex). ment. Breastfeeding duration (age at weaning) is a proxy for Conversely a Beta of −1.30 indicates that an increase in the maternal willingness and ability to provide care, and it is a predictor variable (e.g., age at weaning) is associated with Quinlan and Quinlan • Parenting and Cultures of Risk 169 a decrease of 1.3 units of the outcome variable (e.g., aggression), which is predicted here. We consider Betas with a one-tail p of less than .05 to be worth considering as a real association. The “Model R2” tells how much variation in the outcome variable is accounted for by all of the predictors considered simultaneously. If, for example, Model R2 = .40, then all of the predictors account for 40 percent of the variance in the outcome variable, which means that 60 percent is left “unexplained.” Generally, a predictive social model that accounts for about 20 percent of the variance in an outcome is very promising, 40 percent is an accurate model, and greater than 60 percent is cause for celebration (see Jaeger 1990:339). Model R2 is also sensitive to reliability and precision of the data. One important feature of multiple regression is that FIGURE 2. Association between father involvement in early child- each predictor variable is “controlled” for all the other hood and acceptance of extramarital sex (solid line from Table 2, predictor variables in the analysis. When we include fa- model 2) and prevalence of witchcraft (dashed line from Table 3, ther’s sleeping proximity, age at weaning, and so forth into model 3). one analysis, then any relationship between those predictor variables is “partialled out” so that the Beta represents the independent effect of father sleeping proximity, and the (Leakey 1930:202). Maasai men are equally free (Leakey independent effect of age at weaning and so forth on the 1930:201). Double standards for extramarital sex also vary outcome variable. cross-culturally: Men’s infidelity is often permitted while Additional variables were included in analyses to distin- women’s is not (see, e.g., the Kapauku of Indonesia [Pospisil guish between early and late childhood effects (Barry et al. 1958]). Other societies forbid adultery for both sexes, but 1976) on cultural outcomes as described above. If there is an only women are punished (see, e.g., Aymara of Bolivia [La early sensitive period for development of cultural models, Barre 1948]). Elsewhere, extramarital sex is unacceptable for then inclusion of late childhood variables in a multiple re- men and women. Among the Garo of India, for example, gression model should not mediate the effects of early childhood parenting. If the sensitive period is complete, then the Post-marital fidelity of the husband and wife is very rigor- effects in late childhood should either be negligible or fully ously observed. Any violation of such fidelity is socially ostracized and heavily penalized. . . . This is a social of- mediated by early childhood variables. First, “Inculcation of fence for which the mothers of the boys and the girls Sexual Restraint in Late Childhood” was included to control are to pay fines . . . and . . . provide a feast to the villagers. for later socialization effects in an analysis of “Extramarital [Sinha 1966:97] Sex.” Second, “Inculcation of Aggression in Late Childhood for Boys” was included in an additional analysis of “Ag- We used multivariate analysis to better understand regression.” Third, “Inculcation of Trust” was included in an lations between parenting and cross-cultural variation in analysis of “Prevalence of Witchcraft” because witchcraft is attitudes toward infidelity. We included all four early child- based in mistrust of community members. Finally, “Incul- hood parenting predictors in this analysis of attitudes to- cation of Obedience in Late Childhood” was included in ward extramarital sex. The risk-response model predicts the analysis of “Theft.” less acceptance of extramarital sex in societies in which parents are available and responsive to their children. Fa- ther’s sleeping proximity to the infant and parental re- RESULTS sponsiveness to infant crying were significant indepen- Acceptance of Extramarital Sex dent predictors of extramarital sex (Table 2, models 1 and Extramarital sex gauges the strength of conjugal bonds and 2). Extramarital sex tended to be less acceptable in soci- attitudes toward multiple sex partners. There is considerable eties in which fathers sleep in the same room or same cross-cultural variation in attitudes toward infidelity. Some bed with their infants (B =−0.45, p = 0.001), and in which societies allow extramarital sex for both genders. Among parents respond quickly to infant crying (B =−0.39, p = the Chukchee of Siberia, for example, infidelity leads to “an 0.002). Father involvement in early childhood showed a amicable adjustment” through group marriage including “curvilinear” relationship to acceptance of extramarital sex the two who initiated the affair and their spouses (Bogoraz- (Figure 2). Extramarital sex was more acceptable in cul- Tan 1904:401). The Maasai of Kenya have especially open tures in which fathers tended to be distant, and acceptance attitudes: “Husbands do not mind if their wives have in- of infidelity decreased as fathers were in closer proxim- tercourse with other men provided that they do not do ity to young children; however, the effect “tapered off” in so actually at a time when their husbands want them” societies in which fathers were regularly close (B for father 170 American Anthropologist • Vol. 109, No. 1 • March 2007 involvement =−1.59, p = 0.002; B for father involvement this kind, so it is done for them by the villagers, and the squared = 0.22, p = 0.005; Figure 2). villagers believe that this places the Mbuti in a dependent In an additional analysis, we compared effects of par- position. In fact of course it does not, because the Mbuti be- enting in early childhood with socialization in later child- lieve in natural causes rather than supernatural” (Turnbull hood. We predicted that later socialization would have 1965:60). weaker effects than does early childhood parenting. If so, Some societies restrict attribution of witchcraft to par- then adding later socialization should have little effect on ticular community members, as with the Azande, for whom the association between early childhood parenting and ex- only sons of male witches and daughters of female witches tramarital sex. Model 2 (Table 2) shows that inculcation may inherit witchcraft abilities, which are “psychic” (requir- of sexual restraint in late childhood was not a significant ing no rituals or outward actions; see Evans-Pritchard 1937). predictor of extramarital sex (B =−0.01, p = 0.387), nor Every Azande is nevertheless an “authority on witchcraft,” did it influence the association between early childhood for witchcraft information in so pervasive. E. E. Evans- parenting and acceptance of extramarital sex. This result Pritchard writes that managu (witchcraft) was one of the is consistent with the prediction that parenting in early first words he heard in Zandeland, and he heard it uttered childhood has the strongest effects on later outcomes. The daily (1937). model accounted for between 32 and 34 percent of the In the second set of analyses, we examined cross- cross-cultural variance in the acceptance of extramarital cultural associations between importance of witchcraft and sex. the four parenting variables. Again, the risk-response model Table 2 also includes results of an ordinal logit analysis predicts that witchcraft is less important in societies in to test the adequacy of the linear model, which uses an which parents are available and responsive to their chil- outcome variable with four levels (see appendix). Results dren. Fathers’ sleeping proximity to infants (B =−0.39, p = are presented as log odds of being in the next higher level 0.000) and parental responsiveness to infant crying (B = of the outcome variable. In sum, the ordinal logit model −0.44, p = 0.009) were negatively and significantly associ- gave results very similar to the linear model, supporting the ated with prevalence of witchcraft (Table 3, model 1). In so- linear analysis (see appendix for more details). cieties in which fathers slept close to infants and in which parents were more responsive to infants, witchcraft tended to be less important, indicating less-hostile social cogni- Importance of Witchcraft tion in societies with responsive parenting. Similar to re- Witchcraft is a culture-level expression of “hostile social sults for extramarital sex, father involvement in early child- cognition,” which ascribes misfortune to the malevolent hood showed a “curvilinear” relationship to importance of and covert actions of others. Witchcraft is a serious matter, witchcraft (see Figure 2). Witchcraft was more important in so much so that in some cultures believers fear talking about cultures in which fathers tended to be distant, and preva- it openly (Quinlan 2004). In other cultures, witchcraft may lence of witchcraft decreased as fathers were in closer prox- be a perennial topic of conversation (Evans-Pritchard 1937). imity to young children. Again the effect “tapered off” in Cross-cultural examples illustrate the variation and ex- societies in which fathers were regularly close (B for father tent of insecurity associated with witchcraft beliefs. Among involvement =−1.46, p = 0.020; B for father involvement Basques of Spain, who have a high score for witchcraft, the squared = 0.22, p = 0.015; see Figure 2). topic is so sensitive that ethnographers have encountered We included inculcation of trust as measure of so- difficulty studying Basque witchcraft legends: “Not that the cialization in later childhood in an additional analysis of belief in witchcraft is extinct among the Basques, but be- witchcraft. Our intention was to test the relative effects cause it is so rife” (Webster 1877:60). Traditional Basques of parenting in early and late childhood. Inculcation of suspect witchcraft even by ostensibly trustworthy commu- trust was not a significant predictor of witchcraft indepen- nity members (Caro Baroja 1958; Webster 1877; Zulaika dent of early childhood parenting (B =−0.01, p = 0.379, 1988). There are accounts of bewitched “priests who sud- Table 3, model 2). This result is consistent with the pre- denly changed into dogs” (Zulaika 1988:261). Apparently, diction that parenting in early childhood has stronger Basques even fear their animals as witches: “Dogs, cats, effects on development than does parenting later in donkeys, goats, cows, [and] oxen are turned projectively childhood. into beasts of mythical aggression” (Zulaika 1988:261). Such We also included presence of writing as control vari- notions reflect a culture in which few relationships seem able (see Appendix). Witchcraft tended to be more impor- secure. tant in societies with writing (B = 0.89, p = 0.000, Table 3); Conversely, the Ituri Forest Mbuti (hunter-gatherers) however, writing did not substantially mediate the effect of have the lowest witchcraft score. Mbuti reject notions of parenting practices on the importance of witchcraft cross- malevolent supernatural powers, despite exposure to such culturally. Considered simultaneously all of the variables ac- ideas from neighbors, Bantu and Sudanic villagers. Such vil- counted for between 40 and 56 percent of the cross-cultural lagers attribute most deaths to malevolent magic, and they variance in the importance of witchcraft. An ordinal logit attempt to discover the witch to prevent more harm. In model supported the adequacy of the linear analysis of four contrast, “The Mbuti have no knowledge of divination of levels of witchcraft (Table 3). Quinlan and Quinlan • Parenting and Cultures of Risk 171

TABLE 3. Multiple Regression Analysis Showing Effects of Parenting on Importance of Witchcraft

Unstandardized Log Log odds Model Stats Model Variables Beta 1-tail p Odds 1-tail p R2 Np

1 Intercept 6.38 .001 Father’s sleeping proximity to infant −0.39 .006 −1.10 .000 Father involvement in early childhood −1.46 .017 −4.46 .020 0.41 50 .000 Father involvement in early childhood squared 0.22 .012 0.67 .015 Parental responsiveness to crying −0.44 .034 −1.35 .009 Age at weaning −0.06 .102 −0.15 .155 Writing 0.75 .002 2.27 .000 2* Intercept 7.53 .000 Father’s sleeping proximity to infant −0.32 .086 −0.07 .127 Father involvement in early childhood −2.01 .000 −9.19 .004 Father involvement in early childhood squared 0.34 .000 1.43 .001 0.56 40 .000 Parental responsiveness to crying −0.53 .027 −1.58 .015 Writing 0.93 .005 3.34 .000 Inculcation of trust −0.08 .039 −0.42 .001 3∗∗ Intercept 6.50 .000 Father involvement in early childhood −2.70 .000 −8.59 .000 Father involvement in early childhood squared 0.40 .001 1.28 .000 0.40 55 .000 Parental responsiveness to crying −0.23 .123 −0.76 .089 Writing 0.89 .001 2.74 .000 Inculcation of trust −0.01 .442 −0.03 .379

Note. ∗Age at weaning was removed because n/k < 10 (see Appendix). ∗∗Father’s sleeping proximity was removed because n/k < 10. Log odds estimates refer to an ordinal logit model used because the outcome variable has only four levels (see Appendix).

Prevalence of Aggression (Homicide and Assault) later socialization for aggression was, however, a significant We also examined cross-cultural associations between predictor of aggression independent of early childhood par- = . , = . prevalence of aggression and early childhood parenting. enting practices (B 1 11 p 0 006). This finding is con- The risk-response hypothesis predicts less aggression in sistent with previous studies (Ember and Ember 2002). The societies with relatively more responsive parenting. Fa- result also suggests that in some domains of behavior both thers’ sleeping proximity to infants (B =−2.69, p = 0.007) early childhood parenting practices and later socialization and age at weaning (Beat =−0.23, p = 0.003) were signif- are important. icantly and negatively associated with aggression. Aggres- sion tended to be less prevalent in societies in which fa- Prevalence of Theft thers slept nearer to infants and in which weaning was rela- Finally we examined cross-cultural associations be- tively late. The model accounted for 32 percent of the cross- tween the prevalence of theft and parenting in early cultural variance in aggression. childhood, predicting that theft would be less common Inculcation of aggression in late childhood did not ap- in societies with relatively intensive parenting. In the first preciably alter the effects of father’s sleeping proximity and model, age at weaning (B =−0.52, p = 0.019) and father age at weaning on aggression (Table 4, models 3 and 4). involvement (B =−0.78, p = 0.032) were associated with In contrast to socialization for sexual restraint and trust, theft (Table 5, model 1). Theft tended to be less prevalent

TABLE 4. Multiple Regression Analysis Showing Effects of Parenting on Aggression (Homicide and Assault)

Unstandardized Model Variables Beta 1-tail p Model R2 Stats N p

1 Intercept 22.98 .008 Father’s sleeping proximity to infant −3.34 .016 Father involvement in early childhood −0.56 .158 0.32 37 .002 Parental responsiveness to crying 0.06 .471 Age at weaning −0.23 .003 2* Intercept 12.98 .019 Father’s sleeping proximity to infant −2.69 .007 0.34 44 .003 Father involvement in early childhood 0.03 .475 Age at weaning −1.37 .004 Socialization for aggression 1.11 .006

Note. *Parental responsiveness to crying was removed because n/k<10 (see Appendix). Log odds estimates are unnecessary because the outcome has 16 levels and is a “quasi-interval” variable. 172 American Anthropologist • Vol. 109, No. 1 • March 2007

TABLE 5. Multiple Regression Analysis Showing Effects of Parenting on Frequency of Theft

Unstandardized Model Variables Beta 1-tail p Model R2 Stats N p

1 Intercept 12.71 .009 Father’s sleeping proximity to infant −0.12 .370 Father involvement in early childhood −0.78 .032 0.14 51 .018 Parental responsiveness to crying −0.93 .115 Age at weaning −0.52 .019 2* Intercept 14.62 .000 Father involvement in early childhood −0.60 .031 Parental responsiveness to crying −1.43 .013 0.13 52 .004 Age at weaning −0.19 .152 Inculcation of obedience in late childhood −0.24 .000

Note. *Father’s sleeping proximity was removed because n/k < 10 (see Appendix). Log odds estimates are unnecessary because the outcome has nine levels and is a “quasi-interval” variable. in societies in which weaning was relatively late and in fluenced by models that respond to environmental condi- which fathers were more involved. tions. The relationships among environment, culture, and Effects of early-parenting practices on frequency of behavior are perhaps best illustrated with a hypothetical theft were somewhat mediated by later socialization. Incul- example of culture change. Let us suppose there is a popu- cation of Obedience in Late Childhood was a significant predic- lation that has been exposed to substantial extrinsic risk for tor of theft (B =−0.24, p = 0.000), and it appeared to medi- quite some time so that a model for unresponsive or “low- ate the effect of age at weaning (B =−0.19, p = 0.152 after intensity” parenting is a stable cultural adaptation. Imagine controlling for Inculcation of Obedience). Adding inculcation that the source of extrinsic risk is a complex social system of obedience also made parental responsiveness to crying sig- encouraging strong discrimination against the members of nificant (B =−1.43, p = 0.013, Table 5, model 2). Again, re- our hypothetical population. In short, the environment is sults suggest that both early childhood parenting practices such that no matter how people care for their children (be- and later socialization are equally important for some beyond meeting basic needs), their children’s prospects are haviors. Note, however, that these variables account for a quite limited and they are exposed to many extrinsic haz- small proportion (13 to 14 percent) of the variance in the ards. Now suppose that the discriminatory ideology and be- frequency of theft. havior, for whatever reason, begins to erode in the wider society. The pressures from extrinsic risk slacken, and some DISCUSSION members of our hypothetical population begin to perceive the change in prospects for their children. This perception These results are generally consistent with evolutionary life then translates into new parenting behavior and the chil- history theory and specific predictions of the risk response dren of the perceptive parents begin to succeed in the al- hypothesis. Extramarital sex tended to be less acceptable tered environment. Through social learning (see Flinn 1997; and witchcraft tended to less important in societies with Richerson and Boyd 2005), other members of the popula- responsive parenting. Similarly aggression and theft tended tion begin to imitate the behavior and take on the values to be less prevalent in societies with responsive parents. So- of the successful “innovators” and a new model takes hold. cialization in later childhood did not “mediate” or weaken As long as the “new” environment remains relatively sta- the effects of early childhood parenting. Only in the cases ble, then the new model will influence the parenting prac- of aggression and theft were socialization practices in late tices of subsequent generations. Cultural models for par- childhood significant predictors along with early childhood enting themselves, however, are no less of an ecological parenting variables. These findings suggest a sensitive pe- response. riod for effects of parenting on development (see also Belsky The above scenario suggests two “modes of cultural et al. 1991; Quinlan 2003). Finally, the results suggest a so- transmission” (Cavalli-Sforza and Feldman 1981): Pressure cioecological approach for understanding the development from cultural models suggests a kind of “horizontal trans- and maintenance of cultural models related to risk taking. mission,” by which individuals learn from those around The risk-response model might have broad applications; for them possibly through a “conformist learning bias,” that example, it may help account for associations between un- is, they conform to behavior they see around them be- predictable resources, socialization practices, and warfare cause it presumably works (Richerson and Boyd 2005). The (Ember and Ember 1992b, 1994). risk-response hypothesis offers a fairly detailed theory of “vertical transmission” of culture in which children Evolutionary Ecology of Cultural Models “learn” from their parents (see also Freedman and Gorman Parenting behaviors themselves might be the product of 1993). Culture models for parenting may facilitate verti- cultural models. We suggest that parenting is strongly in- cal transmission: It may be easier to be a responsive or Quinlan and Quinlan • Parenting and Cultures of Risk 173 unresponsve parent when models indicate that such behav- child development may reflect the lack of coherent the- ior is proper. ory and robust predictive models. A life history approach From an evolutionary standpoint, parenting is a high offers a unified theory of parental motivation and ex- stakes endeavor. Failure to adopt the locally appropriate par- pected child outcomes that may resolve some long-standing enting strategy could result in dire fitness consequences. controversies. Hence, we may be evolved to carefully evaluate prevailing Life history strategies might be encoded, through expe- models of parenting, particularly when there is evidence for rience, as personality traits targeting adaptive motivations. changes in extrinsic risk. Perhaps such evaluation under- Locus of control (LOC) is a psychological measure of an in- lies the conflicted emotions parents often experience when dividual’s perception of their control over events. People considering what is best for their children. with an external locus of control believe that they cannot A further note is important here: Correlations among influence outcomes through their own efforts: They per- high extrinsic risk, parenting, and child outcomes are un- ceive extrinsic risk. One dimension of external LOC is the likely because of genetic differences between populations. belief in “powerful others”—the notion that other people Human environments change on too short a time scale control events in one’s life (Levenson 1981). This “power- for genetic adaptation to fix behavioral strategies. Rather, ful others” belief seems inherent in witchcraft. Conversely natural selection acted on the cognitive-emotional internal locus of control reflects a belief that one can in- decision-making structures that generate flexible response fluence outcomes. Studies in Western cultures show that to local conditions (e.g., Tooby and Cosmides 1992). external LOC was associated with teen pregnancy (Young Furthermore, the genetic evidence for human behavioral et al. 2004; Young et al. 2001), “risky” sexual behavior dispositions is equivocal, to say the very least. Heritability (Loue et al. 2004), externalizing behaviors (Jackson et al. estimates (proportion of variance in expressed traits be- 2000), and positive attitudes toward promiscuous sexual- cause of genetic variance) for human behavior are modest, ity (Werner-Wilson 1998), traits indicative of mating effort and they may be inflated by shortcomings of analytical strategies. Interestingly—and perhaps not coincidentally— approaches: Twin studies tend to overestimate heritability external LOC is an “individual level” characteristic of Oscar when other family members are excluded (Feldman and Lewis’s classic “culture of poverty” (Lewis 1966). Otto 1997), which is usually the case. Human heritabil- Life history strategies apparently respond to environ- ity estimates frequently have very large standard errors mental risk with at least two related consequences: First, leading to statistically insignificant results—so much so individuals decide whether to reproduce now or delay un- that statistical significance is not reported in many studies til the future when they are more likely to succeed after (Kamin and Goldberger 2002). And the typical practice accruing resources and skills. Second, individuals can tend of lumping gene-environment correlations and interac- toward either risk-prone or risk-averse behavior depending tions with “residual error” may seriously bias heritability on the predictability of outcomes. estimates (Dickens and Flynn 2001; Ehrlich and Feldman Time perspective has been linked with reproductive 2003; Maccoby 2000). Similarly, “shared” and “nonshared strategies. Differences in time perspective are predicted to environment” are rarely measured (Maccoby 2000). Finally, target potential payoffs to current versus future reproduc- we are skeptical of substantial heritability (>.5 for example) tion (Chisholm 1999b). “Present orientation”—a focus on in human behavior on theoretical grounds: “Phenotypic current rewards rather than delayed gratification—is asso- plasticity” is the hallmark of human adaptation. From a ciated with current reproduction and a mating effort strat- life history standpoint, we might ask why humans would egy. Conversely, “future orientation”—a focus on delayed invest heavily in a brain designed to live in myriad environ- gratification—is associated with delayed reproduction and ments, if motivational dispositions (personality traits) are a parenting effort strategy (Chisholm 1999a). Present orien- locked-in genetically or even substantially heritable? We tation has been associated with impulsivity and risk-taking argue that natural selection “designed” the developmental behaviors (Robbins and Bryan 2004). Present orientation is processes indicated in this study to deal with environmen- also associated with individual level “culture of poverty” tal fluctuations from one generation to the next. Child characteristics (Lewis 1966). development, hence, may be the underpinning of cultural Impulsivity is also correlated with aspects of mating- adaptation. effort strategies in humans (Comings et al. 2002; Donohew et al. 2000; Hoyle et al. 2000; Zuckerman and Kuhlman 2000) and rhesus monkeys (Gerald et al. 2002). Impulsivity Implications for Personality Development may underlie such behavioral strategies, which often in- Some recent trends in behavior genetics and developmen- clude “present orientation,” or a lack willingness to delay tal theory suggest that family environment has little effect gratification (Chisholm 1999a), and externalizing behav- on children’s personality development. These suggestions ior such as delinquency (Belsky et al. 1991). Impulsivity, in are based in part on the often weak or inconsistent asso- fact, has been associated with multiple risk-taking behaviors ciations between parenting practices and child outcomes (Aklin et al. 2005). Development of impulsivity has been (see Harris 1995). We present an alternative interpretation: linked to unstable family environments during childhood Weak or inconsistent associations between parenting and (Elder et al. 1986). 174 American Anthropologist • Vol. 109, No. 1 • March 2007

CONCLUSION Evolutionary life history theory provides useful predictions Robert J. Quinlan Department of Anthropology, Washing- for relations between environmental risk, parenting prac- ton State University, Pullman, WA 99164–4910 tices, and later development. The risk-response model, de- Marsha B. Quinlan Department of Anthropology, Wash- rived from life history thinking, accounts for significant ington State University, Pullman, WA 99164–4910 cross-cultural variation in sexual attitudes, aggression, and witchcraft, but substantial unexplained variance remains. NOTES Some unexplained variance is likely because of the often- Acknowledgments. coarse variable coding in the SCCS (Taylor et al. 2006). New Thank you, William Divale, for help with the Standard Cross-Cultural Sample. We also thank Barry Hewlett, parenting and life history variables may improve model Courtney Meehan, and Jeannette Mageo for helpful comments on fit. Additionally, forms of risk response, such as witchcraft, early drafts of this article. may tend to emerge under specific socioecological regimes. 1. Cultural success refers to locally valued, socially salient “curren- Identifying sociocultural antecedents or catalysts for par- cies” such as status, wealth, or knowledge (Irons 1979). ticular modes of risk response may help account for ad- 2. Robert LeVine et al. (1994) also suggest that infant and child mortality may shape parenting practices; however, in contrast to life ditional cross-cultural variance. Finally, the complexity of history theory, they argue that infant mortality is positively associ- human development and tremendous environmental vari- ated with parental responsiveness. Based on a qualitative compari- ation suggests that we may never account for the total son between Gusii and Euro-American parents, their model posits “pediatric” or “pedagogic” parenting “goals,” depending in part variance in sociocultural risk response (see Hewlett et al. on local mortality risk (LeVine et al. 1994:249–256). More recent 2000b). Robust predictive theory coupled with multivari- studies suggest that the pediatric–pedagogic distinction offers an ate analysis including many cultures from multiple world incomplete picture of parental motivation (Hewlett et al. 2000b; Quinlan in press). Life history theory makes a crucial distinction regions might offer the best chance of understanding such between conditional mortality risk, which is sensitive to parenting, complexity. and extrinsic risk, which is not influenced by parental care (Borg- Following trends in developmental biology, our find- erhoff Mulder 1992; Harpending et al. 1990; Quinlan in press). A recent study shows that distinctions in sources of environmental ings suggest that early influences are particularly strong risk may account for cross-cultural trends in parental responsive- (West-Eberhard 2003). However, as B. Whiting (1980) notes, ness (Quinlan in press). Furthermore, parental trade-offs between although early environment is important, we should not direct (childcare) and indirect parental investment (wealth and status) may better explain some contrasts between parenting in sub- expect people to be inflexibly locked into developmental sistence level and Euro-American society. That is, in some environ- trajectories established early on. At least with respect to ments parents may perceive that direct childcare is less crucial for aggression, for example, later socialization does appear to children’s long-term success than is attainment of wealth and status, including activities (e.g., a 50-hour work week) that are often have an independent additive effect: Despite early parent- mutually exclusive of direct parenting. Unfortunately, more com- ing practices, later socialization might still reduce aggressive plete development of the latter hypothesis is beyond the scope of behavior. Results presented here and in other studies sug- the present analysis. gest potential application. For example, violence preven- tion programs may simultaneously target later childhood REFERENCES CITED socialization and parenting practices. Ainsworth, Mary D. Salter, Mary C. Blehar, Everett Waters, and Sally If all humans share the same basic biological adapta- Wall tions including neurological structures, then how do we ac- 1978 Patterns of Attachment. Hillsdale, NJ: Erlbaum. count for such tremendous cross-cultural variability? This Aklin, W. M., C. W. Lejuez, M. J. Zvolensky, C. W. Kahler, and M. Gwadz central question has motivated cultural anthropologists for 2005 Evalutation of Behavioral Measures of Risk Taking Propen- the last century. Approaches integrating evolutionary ecol- sity with Inner City Adolescents. Behaviour Research and Ther- ogy, psychology, and cultural anthropology have emerged apy 43:215–228. in the last few decades, but there are few empirical stud- Amato, Paul R. 1996 Explaining the Intergenerational Transmission of Divorce. ies using such frameworks specifically to understand cul- Journal of Marriage and the Family 58:628–640. tural variation. This study used one set of evolutionary pre- 1999 Children of Divorced Parents as Young Adults. In Coping dictions for links between parenting practices and cultural with Divorce, Single Parenting, and Remarriage: A Risk and Resiliency Perspective. E. Mavis Hetherington, ed. Pp. 147–163. models to understand relations between psychological and Mahwah, NJ: Lawrence Erlbaum. cultural variation in traditional societies. We hope to see Barber, Nigel similar empirical studies in the near future. In sum, life 1998 Sex Differences in Disposition toward Kin, Security of At- tachment, and Sociosexuality as a Function of Parental Di- history thinking offers an important tool to understand vorce. Evolution and Human Behavior 19:125–132. cross-cultural trends. Such an approach might be use- Barry, Herbert, III, Lili Josephson, Edith Lauer, and Catherine Mar- fully integrated with other evolutionary culture theo- shall 1976 Traits Inculcated in Childhood: Cross-Cultural Codes 5. ries (e.g., Richerson and Boyd 2005). Further research Ethnology 15:83–114. into environment-parenting relationships and develop- Barry, Herbert, III, and Leonora M. Paxson mental mechanisms, including hormonal aspects of per- 1971 Infancy and Early Childhood: Cross-Cultural Codes 2. Eth- nology 10:466–508. sonality development and learning biases may prove Beaty, Lee A. fruitful. 1995 Effects of Paternal Absence on Male Adolescents’ Quinlan and Quinlan • Parenting and Cultures of Risk 175

Peer Relations and Self-Image. Adolescence 30(120):873– havior: An Anthropological Perspective. L. Cronk, N. Chagnon, 880. and W. Irons, eds. Pp. 203–221. New York: Aldine de Gruyter. Belsky, J. Deater-Deckard, K., N. Atzaba-Poria, and A. Pike 1997 Attachment, Mating, and Parenting: An Evolutionary In- 2004 Mother- and Father-Child Mutuality in Anglo and Indian terpretation. Human Nature 8:361–381. British Families: A Link with Lower Externalizing Behaviors. 1998 Paternal Influence and Children’s Well-Being: Limits of, Journal of Abnormal Child Psychology 32(6):609–620. and New Directions for, Understanding. In Men in Families: Dickens, W. T., and J. R. Flynn When Do They Get Involved? What Difference Does It Make? 2001 Heritability Estimates versus Large Environmental Effects: Alan Booth and Ann C. Crouter, eds. Pp. 279–290. Mahwah, The IQ Paradox Resolved. Psychological Review 108:346–369. NJ: Lawrence Erlbaum. Divale, W. Belsky, J., L. Steinberg, and P. Draper 2004 Codebook of Variables for the Standard Cross-Cultural 1991 Childhood Experience, Interpersonal Development, and Sample. World Cultures: The Journal of Cross-Cultural and Reproductive Strategy: An Evolutionary Theory of Socializa- Comparative Research 14(2):1–347. tion. Child Development 62:647–670. Donohew, L., R. Zimmerman, P. S. Cupp, S. Novak, S. Colon, and Bereczkei, T., and A. Csanaky R. Abell 1996 Evolutionary Pathway of Child Development: Lifestyles of 2000 Sensation Seeking, Impulsive Decision-Making, and Risky Adolescents and Adults from Father-Absent Families. Human Sex: Implications for Risk-Taking and Design of Interventions. Nature 7:257–280. Personality and Individual Differences 68(6):1079–1091. Betzig, Laura Dornbusch, S., J. Carlsmith, S. Bushwal, P. Ritter, H. Leiderman, A. 1989 Causes of Conjugal Dissolution: A Cross-Cultural Study. Hastorf, and R. Gross Current Anthropology 30:654–676. 1985 Single Parents, Extended Households and the Control of Bogoraz-Tan, Vladimir G. (sometimes published as “Waldemar Bo- Adolescents. Child Development 56:326–341. goras”) Dow, M. M. 1904 The Chukchee: Material Culture, New York: E. J. Brill. Doc. 1987 Subsample Replication in Cross-Cultural Surveys. Ameri- Number 1, eHRAF Collection of Ethnography. can Anthropologist 89:125–133. Borgerhoff Mulder, M. 1991 Statistical Inference in Comparative Research: New Direc- 1992 Reproductive Decisions. In Evolutionary Ecology and Hu- tions. Behavior Science Research 25:235–257. man Behavior. E. A. Smith and B. Winterhalder, eds. Pp. 339– Draper, Patricia, and Henry Harpending 374. New York: Aldine de Gruyter. 1982 Father Absence and Reproductive Strategy: An Evolution- 2001 Using Phylogenetically Based Comparisons in Anthropol- ary Perspective. Journal of Anthropological Research 38:255– ogy: More Questions Than Answers. Evolutionary Anthropol- 279. ogy 10:99–111. Ehrlich, Paul, and Marcus Feldman Borooah, V. K. 2003 Genes and Cultures: What Creates Our Behavioral Phe- 2002 Logit and Probit: Ordered and Multinominal Models. nome? Current Anthropology 44(1):87–107 Thousand Oaks, CA: Sage. Elder, G. H., A. Caspi, and G. Downey Bowlby, John 1986 Problem Behavior and Family Relationships: Life Course 1969 Attachment and Loss. New York: Basic. and Intergenerational Themes. In Human Development and Broude, Gwen J., and Sarah J. Greene the Life Course: Multidisciplinary Perspectives. Aage B. 1976 Cross-Cultural Codes on Twenty Sexual Attitudes and Prac- Sørensen, Franz E. Weinert, and Lonnie R. Sherrod, eds. Pp. tices. Ethnology 15:409–429. 293–340. Hillsdale, NJ: Erlbaum. Campbell, B. C. Ellis, B. J., J. E. Bates, K. A. Dodge, D. M. Fergusson, L. J. Horwood, In press Adrenarche and the Evolution of Human Life History. G. S. Pettit, and L. Woodward American Journal of Human Biology. 2003 Does Father Absence Place Daughters at Special Risk for Caro Baroja, Julio Early Sexual Activity and Teenage Pregnancy? Child Develop- 1958 Los Vascos. 2nd edition. Madrid: Ediciones Minotauro. ment 74:801–821. Doc. Number 20, eHRAF Collection of Ethnography. Ellis, B. J., S. McFadyen-Ketchum, K. Dodge, G. Pettit, and J. Bates Cavalli-Sforza, L. L., and M. W. Feldman 1999 Quality of Early Family Relationships and Individual Dif- 1981 Cultural Transmission and Evolution. Princeton: Prince- ferences in the Timing of Pubertal Maturation in Girls: A Lon- ton University Press. gitudinal Test of an Evolutionary Model. Journal of Personality Chisholm, James S. and Social Psychology 77(2):387–401. 1996 The Evolutionary Ecology of Attachment Organization. Ember, Carol R., and Melvin Ember Human Nature 7(1):1–37. 1992a Codebook for “Warfare, Aggression, and Resource Prob- 1999a Attachment and Time Preference: Relations between lems: Cross-Cultural Codes.” Behavior Science Research Early Stress and Sexual Behavior in a Sample of American Uni- 26:169–186. versity Women. Human Nature 10:51–83. 1992b Resource Unpredictability, Mistrust and War: A Cross- 1999b Death, Hope and Sex: Steps to an Evolutionary Ecology Cultural Study. Journal of Conflict Resolution 36(2):242– of Mind and Morality. New York: Cambridge. 262. Chisholm, James S., and V. K. Burbank 1994 War, Socialization and Interpersonal Violence: A Cross- 2001 Evolution and Inequality. International Journal of Epi- Cultural Study. Journal of Conflict Resolution 38(4):620–646. demiology 30:206–211. 2001 Cross-Cultural Research Methods. Walnut Creek, CA: Chisholm, James S., J. A. Quinlivan, R. W. Petersen, and D. A. Coall AltaMira. 2005 Early Stress Predicts Age at Menarche and First Birth, Adult 2002 Father Absence and Male Aggression: A Re-Examination of Attachment, and Expected Lifespan. Human Nature 16(3):233– the Comparative Evidence. Ethos 29(3):296–314. 265. Evans-Pritchard, Edward Evan Clutton-Brock, T. H. 1937 Witchcraft, Oracles and Magic among the Azande. Ox- 1991 The Evolution of Parental Care. Princeton: Princeton Uni- ford: Clarendon Press. Doc. Number 71, eHRAF Collection of versity Press. Ethnography. Comings, D. E., D. Muhleman, J. P. Johnson, and J. P. MacMurray Feldman, M. W., and S. Otto 2002 Parent-Daughter Transmission of the Androgen Recep- 1997 Twin Studies, Heritability, and Intelligence. Science tor Gene as an Explanation of the Effect of Father Ab- 278:1383–1384. sence on Age of Menarche. Child Development 73:1046– Flinn, Mark V. 1051. 1997 Culture and the Evolution of Social Learning. Evolution Cronk, Lee and Human Behavior 18(1):23–67. 2000 Female-Biased Parental Investment and Growth Perfor- 2004 Culture and Developmental Plasticity: Evolution of the So- mance among the Mukogodo. In Adaptation and Human Be- cial Brain. In Evolutionary Perspectives on Child Development. 176 American Anthropologist • Vol. 109, No. 1 • March 2007

K. MacDonald and R. L. Burgess, eds. Pp. 73–98. Thousand and Human Social Behavior: An Anthropological Perspective. Oaks, CA: Sage. N. A. Chagnon and W. Irons, eds. Pp. 257–272. North Scituate, Freedman, D. G., and J. Gorman MA: Duxbury Press. 1993 Attachment and the Transmission of Culture: An Evolu- Jackson, Y., P. Frick, and J. Dravage-Bush tionary Perspective. Journal of Social and Evolutionary Systems 2000 Perceptions of Control in Children with Externalizing and 16(3):297–329. Mixed Behavior Disorders. Child Psychiatry and Human De- Gant, L., K. M. Heath, and G. G. Ejikeme velopment 31(1):43–58. In press Early Motherhood, High Mortality and HIV/AIDS Rates Jaeger, Richard M. in Sub-Saharan Africa. Journal of Health and Social Policy. 1990 Statistics: A Spectator Sport. 2nd edition. Newbury Park, Geary, David C. CA: Sage. 2002 Sexual Selection and Human Life History. Advances in Kamin, L. J., and A. S. Goldberger Child Development and Behavior 30:41–101. 2002 Twin Studies in Behavioral Research: A Skeptical View. The- Gerald, M. S., S. Higley, I. D. Lussier, G. C. Westergaard, S. J. Suomi, oretical Population Biology 61:83–95. and J. D. Higley Kaplan, Hillard S., Kim Hill, Jane B. Lancaster, and A. Magdalena 2002 Variation in Reproductive Outcomes for Captive Male Hurtado Rhesus Macaques (Macac mulatta) Differing in CSF 5- 2000 A Theory of Human Life History Evolution: Diet, Intel- Hydroxyindoleascetic Acid Concentrations. Brain, Behavior ligence, and Longevity. Evolutionary Anthropology 9:156– and Evolution 60:117–124. 185. Goody, Jack Kaplan, Hillard S., and Jane B. Lancaster 1969 Comparative Studies in Kinship. Stanford: Stanford Uni- 2003 An Evolutionary and Ecological Analysis of Human Fer- versity Press. tility, Mating Patterns, and Parental Investment. In Offspring: Gough, E. Kathleen Human Fertility Behavior in Biodemographic Perspective. K. 1959 The Nayars and the Deﬁnition of Marriage. Journal of the W. Wachter and R. A. Bulatao, eds. Pp. 170–223. Washington, Royal Anthropological Institute 89:23–34. DC: National Academy. Guyll, M., and S. Madon La Barre, Weston 2003 Trait Hostility: The Breadth and Speciﬁcity of Schema 1948 The Aymara Indians of the Lake Titicaca Plateau. Menasha, Effects. Personality and Individual Differences 34:681– WI: American Anthropological Association. Doc. Number 3, 693. eHRAF Collection of Ethnography. Handwerker, W. Penn Leakey, Louis S. B. 2003 Traumatic Stress, Ecological Contingency, and Sexual Be- 1930 Some Notes on the Masai of Kenya Colony. Journal of the havior: Antecedents and Effects of Sexual Precociousnees, Sex- Royal Anthropological Institute of Great Britain and Ireland ual Mobility, and Adolescent Childrearing in Antiqua. Ethos 60:185–209. Doc. Number 11, eHRAF Collection of Ethnogra- 31(3):385–411. phy. Harpending, Henry., Patricia Draper, and Renee Pennington Levenson, H. 1990 Cultural Evolution, Parental Care and Mortality. In Disease 1981 Differentiation among Internality, Powerful Others, and in Populations in Transition: Anthropological and Epidemio- Chance. In Research with the Locus of Control Construct, vol. logical Perspectives. A. C. Swedlund and G. J. Armelagos, eds. 1. H. M. Lefcourt, ed. Pp. 15–63. New York: Academic Press. Pp. 251–265. New York: Bergin and Garvey. LeVine, Robert A., Suzanne Dixon, Sarah LeVine, Amy Richman, P. Harris, J. R. Herbert Leiderman, Constance H. Keefer, and T. Berry Brazelton 1995 Where Is the Child’s Environment? A Group Socialization 1994 Child Care and Culture: Lessons from Africa. Cambridge: Theory of Development. Psychological Review 102(3):458– Cambridge University Press. 489. Lewis, Oscar Hawkes, K., J. F. O’Connell, N. G. Blurton Jones, H. Alvarez, and E. 1966 La Vida: A Puerto Rican Family in the Culture of Poverty. L. Charnov New York. New York: Random House. 1998 Grandmothering, Menopause, and the Evolution of Hu- Liao, Tim F. man Life Histories. Proceedings of the National Academy of 1994 Interpreting Probability Models: Logit, Probit and Other Science 95:1336–1339. Generalized Linear Models. Thousand Oaks, CA: Sage. Hewlett, Barry S. Loue, S., M. Cooper, F. Traore, and J. Fiedler 1991 Intimate Fathers: The Nature and Context of Aka Pygmy 2004 Locus of Control and HIV Risk among a Sample of Mex- Paternal Infant Care. Ann Arbor: University of Michigan Press. ican and Puerto Rican Women. Journal of Immigrant Health Hewlett, Barry S., M. Lamb, B. Leyendecker, and A. Scholmerich 6(4):155–165. 2000a Internal Working Models, Trust and Sharing Maccoby, E. E. among Foragers. Current Anthropology 41(2):287– 2000 Parenting and Its Effects on Children: On Reading and 297. Misreading Behavior Genetics. Annual Review of Psychology 2000b Parental Investment Strategies among Aka Foragers, 51:1–27. Ngandu Farmers, and Euro-American Urban Industrialists. In Mace, Ruth Adaptation and Human Behavior: An Anthropological Perspec- 2000a An Adaptive Model of Human Reproductive Rate Where tive. L. Cronk, N. Chagnon, and W. Irons, eds. Pp. 155–178. Wealth Is Inherited: Why People Have Small Families. In Adap- New York: Aldine de Gruyter. tation and Human Behavior: An Anthropological Perspective. Hill, Kim, and A. Magdalena Hurtado L. Cronk, N. Chagnon, and W. Irons, eds. Pp. 261–282. New 1996 Ache Life History: The Ecology and Demography of a For- York: Aldine de Gruyter. aging People. New York: Aldine de Gruyter. 2000b Evolutionary Ecology of Human Life History. Animal Be- Hoyle, R. H., M. C. Fejfar, and J. D. Miller haviour 59:1–10. 2000 Personality and Sexual Risk Taking: A Quantitative Review. Matsueda, R., and K. Heimer Journal of Personality 68(6):1203–1231. 1987 Race, Family Structure, and Delinquency: A Test of Dif- Hughes, M. M., M. Blom, R. P. Rohner, and P. A. Britner ferential Association and Social Control Theories. American 2005 Bridging Parental Acceptance-Rejection Theory and At- Sociological Review 52(6):826–840. tachment Theory in the Preschool Strange Situation. Ethos McDade, Thomas W. 33(3):378–401. 2003 Life History Theory and the Immune System: Steps toward Hurtado, A. Magdalena, Carol A. Lambourne, Kim R. Hill, and a Human Ecological Immunology. Yearbook of Physical An- Karen Kessler thropology 46:100–125. 2006 The Public Health Implications of Maternal Care Trade- Middleton, John Offs. Human Nature 17(2):129–154. 1967 The Concept of “Bewitching” in Lugbara. In Magic, Irons, William Witchcraft and Curing. J. Middleton, ed. Pp. 55–68. Garden 1979 Cultural and Biological Success. In Evolutionary Biology City, NY: Natural History Press. Quinlan and Quinlan • Parenting and Cultures of Risk 177

Murdock, George P., and Douglas R. White Rothbaum, F., J. Weisz, M. Pott, K. Miyake, and G. Morelli 1969 Standard Cross-Cultural Sample. Ethnology 8:329–369. 2000 Attachment and Culture. American Psychologist Murdock, George P., Suzanne F. Wilson, and Violetta Frederick 55(10):1093–1104. 1978 World Distributions of Theories of Illness. Ethnology Scheper-Hughes, Nancy 17:449–470. 1992 Death without Weeping: The Violence of Everyday Life in Nash, Manning Brazil. Berkeley: University of California Press. 1967 Witchcraft as Social Process in a Tzeltal Community. In Simons, Kevin J., Carl E. Paternite, and Cecilia Shore Magic, Witchcraft and Curing. J. Middleton, ed. Pp. 127–134. 2001 Quality of Parent/Adolescent Attachment and Aggression Garden City, NY: Natural History Press. in Young Adolescents. Journal of Early Adolescence 21:182– O’Connell, Ann A. 203. 2006 Logistic Regression Models for Ordinal Response Variables. Sinha, Tarunchandra Thousand Oaks, CA: Sage. 1966 The Psyche of the Garos. Calcutta: Anthropological Sur- Pennington, Renee, and Henry Harpending vey of India, Government of India. Doc. Number 18, eHRAF 1993 Structure of African Pastoralist Community: The Demog- Collection of Ethnography. raphy, History, and Ecology of the Ngamiland Herero. Oxford: Stearns, Stephen C. Oxford University Press. 1992 The Evolution of Life Histories. Oxford: Oxford University Pietromonaco, P. R., and L. F. Barrett Press. 2000 The Internal Working Models Concept: What Do We Re- Sugiyama, Lawrence S. ally Know about the Self in Relation to Others? Review of Gen- 2004 Illness, Injury and Disability among Shiwiar Forager- eral Psychology 4(2):155–175. Horticulturalists: Implications of Health-risk Buffering for the Pospisil, Leopold J. Evolution of Human Life Histories. American Journal of Phys- 1958 Kapauku Papuans and Their Law. New Haven, CT: Depart- ical Anthropology 124:371–389. ment of Anthropology, Yale University. Number 1, eHRAF Col- Taylor, A. B., S. G. West, and L. S. Aiken lection of Ethnography. 2006 Loss of Power in Logistic, Ordinal Logistic, and Probit Re- Quinlan, Marsha B. gression When an Outcome Variable Is Coarsely Categorized. 2004 From the Bush: The Front Line of Health Care in a Educational and Psychological Measurement 66(7):228–239. Caribbean Village. Belmont, CA: Wadsworth. Thornton, Arland, and Donald Camburn Quinlan, Robert J. 1987 The Influence of the Family on Premarital Sexual Attitudes 2003 Father-Absence, Parental Care and Female Reproductive and Behavior. Demography 24:323–340. Development. Evolution and Human Behavior 24:367–390. Tooby, John, and Leda Cosmides 2006 Gender and Risk in Matrifocal Caribbean Community. 1992 The Psychological Foundations of Culture. In The Adapted American Anthropologist 108(3):464–479. Mind: Evolutionary Psychology and the Generation of Culture. In press Human Parental Effort and Environmental Risk. J. H. Barkow, L. Cosmides, and J. Tooby, eds. Pp. 19–136. New Proceedings of the Royal Society B: Biological Sciences York: Oxford University Press. 274(1606):121–125. Trivers, Robert Quinlan, Robert J., and Mark V. Flinn 1972 Parental Investment and Sexual Selection. In Sexual Selec- 2003 Intergenerational Transmission of Conjugal Stability in tion and the Descent of Man. B. Campbell, ed. Pp. 136–179. a Caribbean Village. Journal of Comparative Family Studies Chicago: Aldine de Gruyter. 34:569–583. 1985 Social Evolution. Menlo Park, CA: Benjamin–Cummings. Quinlan, Robert J., and Marsha B. Quinlan Turnbull, Colin M. In press Human Lactation, Pair-Bonds and Alloparents: A Cross- 1965 Wayward Servants: The Two Worlds of the African Pyg- Cultural Analysis. Human Nature. mies. Garden City, NY: Natural History Press. Doc. Number 2, Quinlan, Robert J., Marsha B. Quinlan, and Mark V. Flinn eHRAF Collection of Ethnography. 2003 Parental Investment and Age at Weaning in a Caribbean Walker, R., M. Gurven, K. Hill, A. Migliano, N. Chagnon, R. De Village. Evolution and Human Behavior 24:1–16. Souza, G. Djurovic, R. Hames, A. M. Hurtado, H. Kaplan, K. 2005 Local Resource Enhancement and Sex-Biased Breastfeed- Kramer, W. J. Oliver, C. Valeggia, and T. Yamauchi ing in a Caribbean Village. Current Anthropology 46(3):471– 2006 Growth Rates and Life Histories in Twenty-Two Small-Scale 480. Societies. American Journal of Human Biology 18:295–311. Richerson, Peter J., and Robert Boyd Waynforth, David, A. Magdalena Hurtado, and Kim Hill 2005 Not by Genes Alone: How Culture Transformed Human 1998 Environmentally Contingent Reproductive Strategies in Evolution. Chicago: University of Chicago Press. Mayan and Ache Males. Evolution and Human Behavior Robbins, Reuben N., and Angela Bryan 19:369–385. 2004 Relationships between Future Orientation, Impulsive Sen- Webster, Wentworth sation Seeking, and Risk Behavior among Adjudicated Adoles- 1877 Basque Legends: Collected Chiefly in the Labourd. Lon- cents. Journal of Adolescent Research 1994:428–445. don: Griffith and Farron. Doc. Number 7, eHRAF Collection of Roff, Derek A. Ethnography. 2002 Life History Evolution. Sunderland, MA: Sinauer. Weinstein, B. S., and D. Ciszek Rogers, William 2002 The Reserve Capacity Hypothesis: Evolutionary Origins 1993 Regression Standard Errors in Clustered Samples. Stata and Modern Implications of the Trade-Off between Tumor- Technical Bulletin 13:12–23. Suppression and Tissue-Repair. Experimental Gerontology 37: Rohner, Ronald Preston 615–627. 1975 They Love Me, They Love Me Not: A Worldwide Study of Werner-Wilson, R. J. the Effects of Parental Acceptance and Rejection. New Haven, 1998 Gender Differences in Adolescent Sexual Attitudes: The CT: Human Resource Area Files. Influence of Individual and Family Factors. Adolescence Rohner, Ronald Preston, Abdul Khaleque, and David E. Cournoyer 33(131):519–31. 2005 Parental Acceptance-Rejection: Theory, Method, Cross- West-Eberhard, Mary Jane Cultural Evidence, and Implications. Ethos 33(3):299– 2003 Developmental Plasticity and Evolution. Oxford: Oxford 334. University Press. Rothbaum, F., and G. Morelli Whiting, Beatrice Blyth 2005 Attachment and Culture: Bridging Relativism and Uni- 1980 Culture and Social Behavior: A Model for the Development versalism. In Culture and Human Development: The Impor- of Social Behavior. Ethos 8(2):95–116. tance of Cross-Cultural Research to the Social Sciences. W. Whiting, John W. M., and Irvin L. Child Friedlmeier, P. Chakkarath, and B. Schwarz, eds. Pp. 99–124. 1953 Child Training and Personality: A Cross-Cultural Study. Lisse, the Netherlands: Swets and Zeitlinger. New Haven, CT: Yale University Press. 178 American Anthropologist • Vol. 109, No. 1 • March 2007

Whiting, John W. M., and Beatrice Blyth Whiting (3) Prevalence of Individual Aggression (sum of vari- 1975 Aloofness and Intimacy of Husbands and Wives: A Cross- ables 1665, Homicide, and 1666, Assault [Ember and Ember Cultural Study. Ethos 3(2):183–207. Wilson, Margo, and Martin Daly 1992a]) ranges from 2 (low on both variables) to 18 (high 1997 Life Expectancy, Economic Inequality, Homicide, and Re- on both variables). productive Timing in Chicago Neighbourhoods. British Medi- (4) Prevalence of Theft (variable 1667) ranges from 1 cal Journal 314:1271–1281. Winterhalder, Bruce, and Paul W. Leslie (low) to 9 (high). 2002 Risk-Sensitive Fertility: The Variance Compensation Hy- pothesis. Evolution and Human Behavior 23:59–82. II. Predictor variables indicating parental responsive- Worthman, C. M. 1999 Epidemiology of Human Development. In Hormones, ness (Barry and Paxson 1971): (1) Father’s Sleeping Prox- Health and Behavior. C. Panter-Brick and C. M. Worthman, imity to Infant ranges from 1 (different room) to 3 (same eds. Pp. 47–94. Cambridge: Cambridge University Press. bed). This variable was recoded from Parents Sleeping Prox- Young, T. M., S. S. Martin, M. E. Young, and L. Ting 2001 Internal Poverty and Teen Pregnancy. Adolescence imity to Infant (variable 23). 36(142):289–304. (2) Role of Father in Early Childhood (variable 54) ranges Young, T., J. Turner, G. Denny, and M. Young from 1 (distant) to 5 (regularly close). 2004 Examining External and Internal Poverty as Antecedents of Teen Pregnancy. American Journal of Health Behavior (3) Parental response to Infant Crying (variable 31) ranges 28(4):361–373. from 1 (indifferent or punitive) to 5 (always speedy, nurtur- Zuckerman, Marvin, and D. Michael Kuhlman ing). 2000 Personality and Risk-Taking: Common Biosocial Factors. Journal of Personality 68(6):999–1029. (4) Age at Termination of Breastfeeding (variable 45) Zulaika, Joseba ranges from 1 (up to 12 months) to 8 (49–60 months). 1988 Basque Violence: Metaphor and Sacrament. Reno: Univer- sity of Nevada Press. Doc. Number 19, eHRAF Collection of Ethnography. III. Control variables included in the analyses of extramarital sex, aggression, theft and witchcraft: (1) Incul- cation of Sexual Restraint in Late Childhood was included as APPENDIX: STATISTICAL ANALYSIS a control variable in an additional analysis of Extramar- Variables ital Sex. Variables for girls (332) and boys (333) inculca-

I. Outcome variables: (1) Attitude Toward Extramarital Sex tion were highly correlated (rp = .79), and the two were (variable 169 in the SCCS [Broude and Greene 1976]) indi- summed to give an overall estimate for both sexes. The com- cates societal models regarding multiple sex partners and bined variable ranges from 0 (no restraint or opposite for the strength of conjugal relations. Variable 169 in the SCCS both variables) to 18 (extremely strong inculcation for both was reverse coded so that higher scores reﬂected more ac- sexes). cepting attitudes toward extramarital sex. Extramarital Sex (2) Inculcation of Aggression in Late Childhood for Boys was ranges from 1 (extramarital sex equally condemned for both included in an additional analysis of Aggression. This vari- men and women) to 4 (both men and women equally al- able (300 in the SCCS) ranges from 0 (no inculcation, or lowed to engage in extramarital sex). opposite trait) to 9 (extremely strong inculcation). Variable (2) Witchcraft as a Cause of Illness (variable 656 300 was also highly correlated with inculcation for aggres-

[Murdock et al. 1978]) indicates insecure social relation- sion in girls (variable 301, rp = .74); hence, we saw no need ships. Witchcraft ranges from 1 (absence of such a cause) to combine the two which would have reduced our sample to 4 (predominant cause recognized by society). The SCCS size without adding information. distinguishes between witchcraft (variable 656) and sorcery (3) Inculcation of Trust (variable 335) was included in (variable 655). A weak but significant negative correlation an analysis of Prevalence of Witchcraft. This variable ranges between the two is somewhat surprising (rs =−.22). The from 0 (no inculcation, or opposite trait) to 9 (extremely Murdock et al. (1978) definition and discussion of sorcery strong inculcation). (the basis of the SCCS code) emphasize the “use of mag- (4) Inculcation of Obedience in Late Childhood (variable ical techniques” (p. 455) whereas witchcraft includes the 324) was included in the analysis of Theft. Variable 324 action of “human beings believed to be endowed with a ranges from 0 (no inculcation, or opposite trait) to 9 (ex- special power and propensity for evil” (p. 456). We sug- tremely strong inculcation). gest that sorcery, as operationalized in SCCS, taps into the Other control variables may be important. After ini- general use of magic within a culture. We used witchcraft tial tests of predictive models, we conducted an extensive rather than sorcery in our analysis because the definition search for potential control variables including measures and discussion in Murdock et al. (1978) better reflect hos- of social complexity, economy, subsistence strategy, popu- tile social cognition. A multivariate model including par- lation size, division of labor by sex, community structure, enting predictors used here and control variables (“writing” weather patterns, latitude, region, acculturation, modern- and “Africa”) showed weak and nonsignificant associations ization, and contact with “Western-industrial” populations. between parenting and sorcery, which is consistent with Two potentially important controls emerged: (1) Presence our concern about the validity of sorcery (as defined by of writing was a significant predictor of the importance Murdock et al. 1978) as a general expression of hostile social of witchcraft, and its inclusion influenced the unstandard- cognition. ized betas for the predictive model—hence, presence of Quinlan and Quinlan • Parenting and Cultures of Risk 179 writing was included in Table 3; and (2) foraging subsis- Control for regional variation and phylogenetic rela- tence strategy was significantly associated with aggression tions (see Borgerhoff Mulder 2001) can be important in cross-culturally. Foragers tended to have higher prevalence cross-cultural research (Dow 1987, 1991): “Intraclass cor- of assault and homicide; however, including foraging as a relation” among societies from one region can lead to arti- control variable in the analyses did not appreciably alter the ficially low variance estimates making associations appear unstandardized betas or p-values in Table 4—hence, forag- more statistically significant than they actually are. The ing was excluded from the models. SCCS was selected from cultures separated by historical and geographic distance to avoid these problems; however, we calculated robust standard errors (Rogers 1993) based on re- Model Adequacy gionally clustered data (variable 1858) in an abundance of Multiple linear regression and correlation analyses were analytical caution. conducted in SPSS 12 and STATA 9. Partial regression and P-P Two analyses used dependant variables with only four plots did not show deviation from the assumptions of nor- levels: extramarital sex and witchcraft. Diagnostic statis- mally distributed errors or homoscedasticity of variance. All tics indicated that the models meet the linear regression standardized residuals were less than 3, indicating no out- assumptions. We were concerned, however, that using an liers. All variance inflation factors were close to 1, indicating outcome variable with only four levels might lead to ques- lack of multicolinearity. In sum, diagnostics indicated the tionable results. In a separate ordinal logit regression anal- adequacy of the analyses. ysis (shown as “log odds” in Tables 2 and 3), we examined We set the significance level at .05. We consider effects of the four parenting variables on ordinal measures marginally significant effects (p ≤ .10) worth discussing, of extramarital sex and witchcraft. Ordinal logit analysis though we suggest cautious interpretation. For all analy- models the probability of having the next higher score on ses, we used 1-tail p-values because we predict inverse rela- an outcome variable for an associated increase in a predictions between parental responsiveness and extramarital sex, tor variable (Borooah 2002; Liao 1994; O’Connell 2006). For witchcraft and aggression. example, if the log odds for a predictor variable is greater Analysts suggest a rule of thumb in which multiple re- than zero, then an increase in the predictor is associated gression becomes “unstable” when the sample size divided with an increased probability of being in a higher level of by the number of predictors is less than ten (n/k < 10). an outcome variable. If the log odds for a predictor is less We encountered this problem in two of our analyses, than zero (parental responsiveness in Table 3, for example), where we present the results of the potentially unstable then an increase in the predictor (parental responsiveness) model, and then we remove the least significant predic- is associated with reduced probability of being in higher tor and present a second model, which had n/k > 10. level of the outcome variable (importance of witchcraft). The predictor we removed was not significant and the Brant tests indicated that the models did not violate the as- results were not substantially different in the corrected sumption of proportional odds. The ordinal logit analyses model. support the adequacy of the linear models.