Ïîèñê ïî ñàéòó




Ïèøèòå íàì: info@ethology.ru

Follow etholog on Twitter

Ñèñòåìà Orphus

Íîâîñòè
Áèáëèîòåêà
Âèäåî
Ðàçíîå
Êðîññ-êóëüòóðíûé ìåòîä
Ñòàðûå ôîðóìû
Ðåêîìåíäóåì
Íå â òåìó

ñïèñîê ñòàòåé


What is the relevance of attachment and life history to political values?

Randy Thornhill, Corey L. Fincher

1. Introduction

2. Methods

2.1. Participants and demographics

2.2. Other questionnaire measures

3. Results

3.1. Political values scales

3.2. Time preference

3.3. Personality

3.4. Childhood stresses

3.5. Attachment

3.6. Multivariate analyses

4. Discussion

Acknowledgment

References

Copyright

1. Introduction

As measured in questionnaire-based research by political scientists, conservatism–liberalism is a dimension of individual variation in which the more liberal one is, the less conservative such individual is and vice versa (reviewed in Knight, 1993). Moreover, conservatism–liberalism questionnaire scale scores coincide with people's political involvement and party voting preferences (e.g., Altemeyer, 1996, Knight, 1993). Liberals tend to be: against, skeptical of, or cynical about familiar and traditional ideology; open to new experiences; individualistic and uncompromising, pursuing a place in the world on personal terms; private; disobedient, even rebellious rulebreakers; sensation seekers and pleasure seekers, including in the frequency and diversity of sexual experiences; socially and economically egalitarian; and risk prone; furthermore, they value diversity, imagination, intellectualism, logic, and scientific progress. Conservatives exhibit the reverse in all these domains. Moreover, the felt need for order, structure, closure, family and national security, salvation, sexual restraint, and self-control, in general, as well as the effort devoted to avoidance of change, novelty, unpredictability, ambiguity, and complexity, is a well-established characteristic of conservatives. Jost, Glaser, Kruglanski, and Sulloway (2003) show that the ideology of conservatism is for the establishment and maintenance of security or safety and that this is accomplished by fear and associated management of uncertainty and threats to security. The labels “conservatives” and “liberals” are used widely across cultures and correspond to the differences between the two ideologies we have mentioned (see meta-analysis by Jost et al., 2003 involving 12 countries, 88 samples, and 23,000 people; see also Feather, 1979, Forabosco & Ruch, 1994, Knight, 1993).

Although the differences in values between conservatives and liberals have been thoroughly described across decades of political science research, there is no evolutionary theory for this variation. An open question then is: “Which psychological adaptation(s) yields individual differences in political ideology?” A related question is: “Are the feelings and behaviors associated with politics incidental byproducts of responsible psychological adaptation, or are they, at least to some degree, functional (i.e., the reasons why the underlying psychology was favored directly by natural selection)?” We hypothesize that, proximately, individual differences in political values are manifestations of species-typical psychological adaptation of attachment, which in turn ontogenetically arises from experiences of early childhood stressors. Specifically, we propose that conservative ideology is caused by relatively low levels of childhood stress and associated secure attachment, whereas liberal ideology is caused by higher childhood stress and associated avoidant attachment.

In ultimate causal terms, we hypothesize that past selection favored major aspects of attachment because of their adaptive (ancestrally) associated political values and behavior, with conservative values providing advantage in familial and other in-group social relations and with liberal values providing advantage in out-group relations. Hence, we suggest that individual differences in conservatism–liberalism are ontogenetically condition-dependent social tactics that functioned historically in in-group or out-group behavior and that salient ancestral cues affecting individual differences include the degree of childhood stress experienced.

There are three attachment styles described briefly: avoidant persons restrict intimacy and closeness in relationships, avoiding strong emotional connections to others; ambivalent–anxious persons strive to merge with relationship partners and fear loss of closeness in relationships through partner's divestment or abandonment; and secure-attachment persons also strongly value close and intimate relationships but do not fear abandonment. A person's attachment style arises at an early age (measured by caretaker–infant interaction) and has significant but imperfect stability across the life span in romantic relationships in adulthood (see reviews in Kirkpatrick, 2005, Simpson et al., 1996). Prior research on attachment styles has tied them empirically to the degree of early childhood stresses. Avoidant and ambivalent–anxious attachment styles (often combined by researchers into a variable called “insecure attachment”) are associated with higher childhood stress than is secure attachment (Chisholm, 1999, Chisholm et al., 2005).

We used anonymous responses on questionnaires to determine the relationships among individual differences in conservatism–liberalism, attachment styles, and childhood stresses, allowing the testing of predictions that high conservatism is associated with high secure and low avoidant attachment and low childhood stress, whereas liberalism correlates positively with avoidant attachment and childhood stress and negatively with secure attachment. We also collected participants' scores on right-wing authoritarianism (RWA) and social dominance orientation (SDO)—two dimensions of political values related to conservatism–liberalism. In addition, we collected data on personality in order to control its effects on analyses of relationships between political ideology and other key variables.

Attachment style is related to time preference, a central variable in life history theory (Chisholm, 1999, Chisholm et al., 2005). We examine the relationship between time preference, and hence life history theory, and political values. Our time preference questionnaires determined participants' attitudes about past, present, and future time frames. Attitude about the past addresses childhood stress, and present-time and future-time preference address life history theory. According to life history theory, rearing stresses, when predictive of reduced adult life span, cause individuals to adopt a present-time preference rather than a future-time orientation. Present-time preference, compared to future-time preference, is associated with allocation of less somatic effort and more reproductive effort and risk taking (e.g., Charnov, 1993, Chisholm, 1999). We collected data on participants' expected life span as this also is central to life history and time preference.

2. Methods

2.1. Participants and demographics

The 123 participants were enrolled in a nonmajor science course at a US university. Participants filled out an anonymous confidential questionnaire at their desk during a class period. They reported their sex (males, 38; females, 85) and age (mean±S.D.=20.06±3.40 years; range, 18–46 years) and their “religious/spiritual affiliation” as “no religion, non-Catholic Christian, Catholic Christian, or other religion.” In analyses, we used a religion variable, with 1=religion (n=91) and 0=no religion (n=32). The participants indicated socioeconomic status by marking a rung on a ladder scale corresponding to the interpretation of their standing in their self-described community (Singh-Manoux, Adler, & Marmot, 2003). Ladder Rung 1 is the highest in one's community, and Ladder Rung 10 is the lowest. The mean ladder rung selected was 5.2 (S.D.=1.77, n=113). Participants also reported their life expectancy (Chisholm et al., 2005): expected longevity (mean±S.D.)=80.96±11.23 years (n=120).

2.2. Other questionnaire measures

We measured conservatism–liberalism with the 28-item C-scale (Eaves et al., 1997). The C-scale assesses numerous political values: attitude about death penalty, abortion, minorities, immigration, racial segregation, censorship, gay's and women's rights, X-rated movies, military draft, aesthetics, pacifism, nuclear power, and so on across most domains that separate reliably the two political wings according to several decades of research. This measure is calculated such that higher scores align with greater conservatism and less liberalism. Cronbach's α for our sample was .86. We measured RWA by a 30-item scale (Altemeyer, 1996). Cronbach's α for our sample was .94. Prior research shows that RWA and conservatism, measured by the C-scale or similar conservatism scales, highly and positively covary, but that RWA and conservatism are not identical measures (e.g., Altemeyer, 1996). The third political value measure used was the 14-item SDO questionnaire (Sidanius, Pratto, & Bobo, 1994). Cronbach's α for our sample was .88. Scores on SDO correlate positively, but only slightly to moderately, with those on RWA or on conservatism. People high on RWA want to be dominated by authorities, and people high on SDO want to become the dominating authorities themselves (Altemeyer, 1996).

We calculated participants' attachment styles from their responses on the Adult Attachment Questionnaire (AAQ), a 17-item measure of romantic relationship attachment (Simpson et al., 1996). This scale contains two major factors: avoidant attachment and ambivalent–anxious attachment. We extracted these two factors by the principal components method from participants' responses on the AAQ (Bartlett's χ2=602.50, p<.0001; avoidant: eigenvalue=3.98, 23.4% variance; ambivalent–anxious: eigenvalue=2.42, 14.2% variance). These two factors negatively correlated (r=−.26, p=.004, n=121). For each participant, we generated an oblique factor score for each attachment factor and used these factor scores in analyses. Secure attachment for a participant was the sum of one's two factor scores multiplied by −1. Low scorers on both factors are high in secure attachment (Simpson et al., 1996). Secure attachment correlated highly and negatively with each of the two factors (in each case, r=−.61, p<.0001).

We assessed upbringing stresses with the two sets of questions used by Chisholm et al. (2005). The questions referred to “while you were a child, up to about 10 years old.” The four questions in the first set were answered with yes or no and deal with violence between one's parents, parental divorce or separation, and the presence/absence of father/mother. Responses were scored 1=yes (stress present) and 0=no. The second question set (five questions) asked about the participant's relationship with one's mother or father, each parent's personality, and the parents' relationship with each other. For each question of this set, participants responded to a list of “positive,” “negative,” and “mixed” adjectives by indicating those that apply. As in Chisholm et al., we combined the negative and mixed into the category “not positive.” The number of positive adjectives minus the number of not-positive adjectives was the participant's score on that question, and this sum was multiplied by −1 to give a stress score. Finally, sums across both sets of the questions were added to estimate overall childhood stress.

We examined the participants' time preferences using two questionnaires. The 56-item Zimbardo Time Perspective Inventory (ZTPI; Zimbardo & Boyd, 1999) identifies five distinct components of a person's time perspective: future, past–negative, past–positive, present–hedonistic, and present–fatalistic; each component is measured by a different factor. In analyses of present-versus-future tradeoff, we combined the two present components of ZTPI to measure present-time preference, and we used the future component of ZTPI to measure future-time preference. We also examined time preference with the 14-item Consideration of Future Consequences Scale (CFCS) (Strathman, Gleicher, Boninger, & Edwards, 1994). The CFCS has not been studied in relation to present–future tradeoffs, but we reasoned that it is as relevant as the ZTPI in measuring future (high scores) and present (low scores) time preference. The CFCS measures the importance of consideration of future outcomes and willingness to sacrifice current benefits (Strathman et al., 1994).

Our final measure was the 44-item personality scale of Benet-Martínez and John (1998). It measures the “Big Five” personality factors: agreeableness, conscientiousness, extraversion, neuroticism, and openness.

All significance tests were two tailed.

3. Results

3.1. Political values scales

Conservatism–liberalism (C-scale) and RWA correlated positively with involvement in some type of religion (C-scale: r=.31, p=.0007, n=116; RWA: r=.38, p<.0001, n=116); SDO did not (r=−.01, p=.94, n=116). Males and females did not differ in RWA or C-scale scores (ps>.4, ns=114). As often found in studies of SDO, males had higher scores (t=2.62. p=.01, df=114; mean±S.D.: males, 38.6±15.71; females, 31.34±12.84). Socioeconomic status (1=highest) negatively correlated with each of the three political scales (ps=.06 –.10), providing some evidence that high scorers on each political scale were relatively of high status. Given this and the widespread assumption that socioeconomic status affects behavior, socioeconomic status was included in some multiple regression analyses. Age was not correlated with any of the political ideology scales (ps≥.20).

3.2. Time preference

Scores on the CFCS and the future component of the ZTPI (ZTPI future) were highly positively related (r=.55, p<.0001, n=121). Each of these future-time orientation measures correlated negatively with the sum of the two ZTPI present components (n=121; CFCS, r=−.40, p<.0001; ZTPI future, r=−.45, p<.0001) and positively with ZTPI past–positive (n=121; CFCS, r=.23, p=.01; ZTPI future, r=.21, p=.02). Each of the two future-time orientations correlated negatively with ZTPI past–negative (n=121; CFCS, r=−.23, p=.02; ZTPI future, r=−.14, p=.13). ZTPI past–positive and past–negative were negatively related (n=120: r=−.39, p<.0001). Time preference in relation to attachment styles is discussed below (see Section 3.5).

We found no evidence that C-scale or RWA scores were correlated with CFCS, life expectancy, and present and future components of ZTPI. The relationships among scores on each of the three political scales and CFCS, life expectancy, and the components of ZTPI (present components combined) were analyzed in a correlation matrix (n=107). SDO correlated negatively with CFCS (r=−.25, p=.008). Hence, high scorers on SDO reported a relatively present-time orientation. RWA and C-scale correlated negatively with ZTPI past–negative (RWA, r=−.26, p=.008; C-scale, r=−.19, p=.05) and positively with ZTPI past–positive (RWA, r=.31, p=.001; C-scale, r=.24, p=.03). Thus, high RWA and C-scale participants reported feelings of a positive and less negative past, with liberals viewing the past in more negative and less positive ways. There were no correlations among any of the political values scales and life expectancy (ps≥.40) or the present or future components of ZTPI. SDO and ZTPI future correlated somewhat (r=−.17, p=.09), but otherwise, the political scales and the present and future components of ZTPI show little covariation (ps≥.30).

3.3. Personality

C-scale scores and RWA correlated similarly across the five factors of personality (n=107). Conservatism correlated with openness (r=−.22, p=.02), neuroticism (r=−.21, p=.03), conscientiousness (r=.19, p=.05), extraversion (r=.23, p=.02), and agreeableness (r=.18, p=.07). Liberals, in comparison to conservatives, were more open and neurotic, and less conscientious, extroverted, and agreeable. RWA correlated with agreeableness (r=.24, p=.01), openness (r=−.20, p=.04), neuroticism (r=−.19, p=.05), conscientiousness (r=.20, p=.04), and extraversion (r=.25, p=.01). SDO showed a different overall covariation with the five personality factors (n=107). Only the correlation with openness was substantial (r=−.32, p=.001); the other four correlations have ps≥.2.

3.4. Childhood stresses

The childhood stress item “mother absent” was not reported by any participant. Our childhood stress variable, the sum of the remaining eight categories of childhood stresses, covaried negatively with participants' (n=109) socioeconomic position (r=−.22, p=.02) and life expectancy (r=−.22, p=.02). Also, stresses negatively correlated (n=117) with CFCS (r=−.23, p=.01) and ZTPI past–positive (r=−.40, p<0001). Moreover, stresses positively correlated with ZTPI past–negative (r=.34, p=.0001). Stresses did not correlate with present or future ZTPI (ps≥.5, n=117). Thus, childhood stress was higher in (a) low-socioeconomic participants than in high-socioeconomic participants; (b) those who anticipate a shorter life span than in those with greater expected longevity; (c) those with low future orientation compared to those with high future orientation (CFCS); and (d) those who interpret their past in less positive and more negative terms than in those who knew their past as more positive and less negative.

Childhood stress negatively correlated with conservatism and RWA (C-scale: r=−.30, p=.001, n=113; RWA: r=−.24, p=.01, n=112). SDO showed a near-zero correlation with stress (r=−.06, p=.5, n=112). Each item of stress correlated negatively with C-scale scores, but rs ranged from −.10 (father absent, relationship with mother) to −.36 (father's personality) (n=113).

3.5. Attachment

As shown in Table 1, childhood stress correlated positively with avoidant attachment (r=.20, p=.04, n=117); the relationship with ambivalent–anxious attachment was of similar effect size (r=.15, p=.11, ns, n=117). Secure attachment was negatively related to childhood stress (r=−.28, p=.002). Thus, high avoidant attachment and possibly high ambivalent–anxious attachment were associated with relatively stressful childhood experiences, and high secure attachment corresponded to relatively fewer childhood stressors. These patterns were similar to those found by Chisholm et al. (2005) in multiple samples of adults. However, they used the related but not identical attachment measure of Hazan and Shaver (1987) and combined avoidant and ambivalent–anxious styles into one variable, “insecure attachment” (Chisholm et al., 2005).

Table 1.

Attachment styles in relation to childhood stresses and political values scales

Attachment
Avoidant Ambivalent Secure
Childhood stressesa .20 (p=.04) .15 (p=.11) −.28 (p=.002)
C-scaleb −.33 (p=.0003) .01 (p=.92) .25 (p=.007)
RWAb −.13 (p=.17) −.11 (p=.26) .20 (p=.04)
SDOb −.12 (p=.20) .15 (p=.12) −.02 (p=.83)
a

n=117.

b

n=115.

Results on relationships between past-time preference and attachment styles also link attachment and childhood stress. ZTPI past–negative positively correlated with ambivalent–anxious attachment (r=.44, p<.0001, n=120) and negatively with secure attachment (r=−.41, p<.0001, n=120); the relationship of ZTPI past–negative and avoidant attachment was near zero (r=.06). ZTPI past–positive was negatively related to ambivalent–anxious attachment (r=−.24, p=.009, n=120) and avoidant attachment (r=−.16, p=.09, n=120). ZTPI past–positive was positively correlated with secure attachment (r=.32, p=.0003, n=120). ZTPI future was negatively related to ambivalent–anxious attachment (r=−.28, p=.002, n=120), but positively related to avoidant attachment (r=.11, p=.10, n=120).

Table 1 shows the relationships between attachment and political values. Conservatism negatively correlated with avoidant attachment (r=−.33, p=.0003, n=115) and positively correlated with secure attachment (r=.25, p=.007, n=115). The C-scale's relationship with ambivalent–anxious attachment was near zero (r=.01, p=.92). Hence, liberals were more avoidant and less secure in attachment than conservatives. RWA also had a positive relationship with secure attachment (r=.20, p=.04, n=115), but had no relationship with the other two attachment styles (ps>.16). SDO was not related to attachment types (ps≥.12).

3.6. Multivariate analyses

We followed up with multiple regression analyses to examine the covariates of political values while statistically controlling potential confounds identified by the bivariate analyses above. Personality, according to multiple regression, did not confound our finding of the negative relationships between conservatism and avoidant attachment and between conservatism and childhood stress. In one multiple regression, the five personality factors and avoidant attachment were independent variables, with C-scale scores as the dependent variable. The regression is robust (F=3.93, p=.001, R=.43, n=114), and only the variables openness and avoidant attachment were predictors (openness: β=−.24, t=−2.60, p=.01; avoidant attachment: β=−.23, t=−2.37, p=.02). Hence, high scorers on the C-scale (conservatives) were less open and less avoidant than low scorers (liberals). A second multiple regression was identical, except that childhood stress was substituted for avoidant attachment. This regression was robust (F=4.58, p=.0004, R=.46, n=112), and only openness and childhood stress were predictors (openness: β=−.30, t=−3.30, p=.001; stress: β=−.19, t=−1.99, p=.05).

We tested the independent effects of childhood stress and avoidant attachment on conservatism by multiple regression. The regression was robust (F=12.14, p<.0001, R=.43, n=112), and each independent variable had a negative effect on C-scale scores (childhood stress: β=−.22, t=−2.49, p=.01; avoidant attachment: β=−.32, t=−3.53, p=.0006). Hence, conservatives reported fewer childhood stresses, as well as reduced avoidant attachment compared to liberals.

This pattern of effects of stress and avoidant attachment was not seen with RWA and SDO, as expected given that the C-scale measures more encompassing conservatism–liberalism than the other two scales. A multiple regression with RWA as the dependent variable and with stress and avoidant attachment as independent variables revealed an effect of stress (β=−.21, t=−2.23, p=.03) but not avoidant attachment (p=.20). Adding C-scale scores to this regression (n=107) eliminated the importance of stress (p=.90) and reduced that of avoidant attachment (p=.10). This regression, of course, was robust because of the strong effect of conservatism on RWA (β=.86, t=13.45, p<.0001). A multiple regression involving SDO as a dependent variable, and stress and avoidant attachment as predictors revealed no relationship (F=1.05, p=.35, n=111).

Another multiple regression examined the predictiveness of the variables that, as we report above, showed relationships with conservatism–liberalism in our prior analyses (Table 2 lists the variables). The regression was robust (F=40.37, p<.0001, R=.87, n=98). RWA was the strongest predictor (p<.0001), and SDO had an independent effect (p=.01). Avoidant attachment (p=.003) and stress (p=.03) each had predictive power. Religion, socioeconomic standing, and the personality factor openness were not predictors (ps=.37–.94). Given that these three variables were not predictors, a follow-up multiple regression excluded them. It was robust (F=70.32, p<.0001, R=.86, n=104), and all independent variables were predictors (RWA: β=.71, t=12.54, p<.0001; SDO: β=.13, t=2.45, p=.02; avoidant attachment: β=−.18, t=−3.27, p=.002; stress: β=−.11, t=−2.02, p=.05).

Table 2.

Multiple regression results with conservatism (C-scale) as the dependent variable and seven predictor variables that are described in Methods (n=98)

Predictor variable β t p
RWA .71 11.54 <.0001
Avoidant attachment −.17 −3.07 .003
SDO .15 2.65 .01
Stress −.13 −2.19 .03
Socioeconomic status .05 .91 .37
Religion .03 .57 .57
Openness −.004 −.08 .94

4. Discussion

We investigated in a sample of 123 young adults individual differences in political values, as measured by the conservatism scale (C-scale), SDO scale, and RWA scale, as well as the relationship of political values to attachment styles. These three political scales were interrelated positively, but RWA and C-scale had the most overlap. The C-scale was our most important measure of political ideology because it measures a person's values on a comprehensive conservatism–liberalism dimension. Our investigation provides a test of the hypothesis that political values arise from attachment adaptation such that liberals will score high and conservatives will score low in avoidant attachment, and the related hypothesis that liberals score low and conservatives score high on secure attachment. The findings supported the hypotheses. C-scale scores were associated negatively with avoidant attachment and positively with secure attachment. These relationships remained when potential confounds were controlled statistically in multiple regression. Ambivalent–anxious attachment did not correlate with any of the three political values scales.

A major difference found by political scientists is that liberals, in comparison to conservatives, are significantly more individualistic, behaviorally and psychologically independent, autonomous, and self-reliant (see Introduction). Therefore, it may seem no surprise that we found that liberals' attachment style is avoidant, which has been described as the style of independence and autonomy (Rholes, Simpson, Campbell, & Grich, 2001). However, multiple authors suggest that conservatives, not liberals, have insecure attachment (avoidant and ambivalent–anxious) (Holmes, 1996, Roccato, in press). These authors reason that because conservatives are fearful or insecure about status quo upsets, they will be insecurely attached. In what was apparently the first study to test the relationship between a conservatism measure (the RWA scale) and attachment, Roccato found the reverse of this pattern, a result consistent with our findings. Also consistent with our findings is evidence that religious commitment, a positive covariate of conservatism, is positively associated with secure attachment (Kirkpatrick, 2005).

We propose that the liberal mindset of relatively high independence and autonomy and associated avoidant attachment allows historically adaptive maneuver through different types of social relations than the relations for which conservatism and secure attachment are functionally designed. Consistent with this, liberals report a history of broader sexual experiences than conservatives (Feather, 1979). Relatedly, there is evidence that avoidant attachment, compared to secure attachment, is associated with relatively high sociosexual orientation (e.g., number of sex partners, number of brief sexual affairs and romantic partners, and willingness to have sex without commitment; Kirkpatrick, 2005, Rholes et al., 1997). The differences between liberals and conservatives in social behavior extend, we hypothesize, to other domains of social life besides sex and romance. Liberals, as measured on the C-scale, value diversity, out-group relations, plurality of views, and egalitarianism, and hence focus their altruistic activity importantly outside the family and local in-groups. Liberal ideology may function in social behaviors such as limitation of investment in genetic relatives and increased willingness to embrace nontraditional groups, even hostile out-groups, in alliances that may be temporary. The positive out-group attitudes of liberals and positive in-group attitudes and family values of conservatives are consistent with this hypothesis. Hence, our view of attachment psychology is a contrast to the view that attachment has been designed by natural selection to build the relationship between mother and infant primarily, and only incidentally gives rise to an individual's experience in other relationships (e.g., Bowlby, 1982). Our view is also in contrast to the view that adult attachment functions primarily or exclusively in sexual and romantic relationships (Belsky et al., 1991, Chisholm, 1999). Rather, we suggest that attachment styles are products of a psychological adaptation that functions to craft an individual's approach to all social relationships (i.e., not only the relationship between parent and offspring and romantic and sexual partners, but also between other genetic relatives, between allies, and even between enemies).

We found that C-scale scores are unrelated to some key life history variables: life expectancy and present-time and future-time preference. Hence, conservatism–liberalism may arise from psychological adaptation different from that regulating tradeoffs in current-versus-future reproductive effort.

Although the future and present dimensions of time assessment psychology are discussed often in the literature in relation to life history theory (e.g., Chisholm, 1999, Chisholm et al., 2005), personal attitude about the past has received no attention from evolution-minded researchers of human behavior. We hypothesize that ancestral humans' deduction of past–negative time perspective adaptively motivated the exploration and use of new and risky domains of cognition and behavior. Findings are consistent with this. Liberals, whom previous research has characterized as more willing than conservatives to explore the new and the different, reported a time assessment of more past–negative and less past–positive than conservatives. We anticipate that past-time assessment is a composite from a range of ancestral cues throughout juvenile life, including the duration, intensity, and intimacy of interaction with caretakers and others.

There has been considerable speculation in the literature about developmental backgrounds that give rise to conservatism versus liberalism. We test the prediction that conservatives and liberals derive from significantly different ontogenetic childhood stress levels and specifically that liberals derive from childhood backgrounds with more stress than conservatives. This hypothesis is supported. Childhood stress was negatively correlated with C-scale scores (liberals more stress), and this relationship persisted when a number of relevant potential confounds were removed by statistical control. Relatedly, we tested and supported the prediction that liberals report more negative and less positive feelings about their past than conservatives. Also, we found that viewing the past as more negative and less positive is related to relatively greater childhood stress. Our results question the view that conservatives derive from backgrounds involving divestment by fathers (for a discussion of this view applied to RWA, see Altemeyer, 1996). Instead, our results suggest that conservatives arise from families that are more consistently investing, including greater father involvement.

The literature indicates that one's attachment style is an adaptive tactic cued by the extent and nature of adversity in one's early rearing environment that reliably predicts (or did so in evolutionary historical settings) the general niche of the developing individual across the life span (e.g., Belsky et al., 1991, Chisholm, 1999, Chisholm et al., 2005). According to the literature, insecure attachment (avoidant and ambivalent–anxious styles combined) is adaptive in a niche of short-term mating relationships. Given our findings that distinguish avoidant and ambivalently–anxiously attached individuals, we agree in regard to avoidant attachment and suggest that the separation of avoidant and ambivalent–anxious attachment in future research may clarify the role of ambivalent–anxious attachment.

We hypothesize that avoidant-versus-secure attachment dimensions are related strategically to high-versus-low childhood stress, in part through ecological adversity beyond parental control (e.g., poverty, death of, or divestment by parent; Chisholm et al., 2005) and in part strategically imposed by parents to craft offspring for either a social niche of relatively high risk taking, openness to novelty, out-group tolerance, and other liberal cognitions and behavior, versus one of low-risk proneness and associated conservatism. Grossmann, Grossmann, Spangler, Suess, and Unzner, (1985), in a study of infant–mother interactions in Germany, found a high percentage of avoidant–attachment infants (49% compared to a Baltimore, USA, sample with 26%). They relate this to the cultural ideology of “independence training” by mothers in Germany. The German mothers and the Baltimore mothers showed similar amounts of contact with infants, but the contact differed (e.g., the German mothers were “less tender and affectionate,” with shorter episodes of holding their infant). They argue that since the ideal or norm in many German families is an independent and self-reliant infant, mothers accomplish this by reduced sensitivity and attendance to infant solicitations. We suggest that this kind of variation in parenting generates strategic differences in liberalism–conservatism and associated attachment tactics.

Birth order research findings are also consistent with these ideas. Sulloway's (1996) extensive research indicates that birth order correlates with the variables creative life, rebellious activity, risk taking, and conservative and liberal attitudes toward new ideas. Laterborns, in general, receive less parental investment than firstborns, and laterborns behave as liberals in their relatively greater independence, creative contributions, and rebellion against the status quo (Sulloway, 1996). Moreover, Salmon and Daly (1998) reported that firstborns described themselves as less “open to radical ideas” compared to laterborns. They also reported that middleborns, who generally receive less parental investment than firstborn or thirdborn children, were less attached to family and favor friends over family compared to firstborns and thirdborns. We know of no literature reports on conservatism–liberalism scale scores per se across birth order but predict that laterborns and middleborns will show relatively lower conservatism scores than firstborns. We also predict that laterborns and middleborns will show high avoidant attachment, whereas firstborns will show secure attachment; this prediction is consistent with Sulloway's (pp. 120–121) findings from measured parent–offspring relationships and conflict.

There is a large body of evidence that childhood stresses positively affect risk proneness (e.g., Chisholm, 1999, Ross & Hill, 2002). Liberals score higher than conservatives in sensation seeking and risk taking (e.g., Forabosco & Ruch, 1994). The greater risk proneness of liberals may mediate their cognitive styles comprising openness, creativity, and positive attitude toward out-groups. Such an ideology can be highly risky in a normative ideological setting of conservatism.

In sum, our conceptualization of the strategic ontogenies of conservatives and liberals is as follows: the relative magnitude of childhood stressors experienced sends conservatives down one life track and liberals down another. An individual's attachment style arises early in life (conservatives with secure attachment and liberals with avoidant attachment) and is generally retained but open to some adaptive modification if ancestrally salient cues change. Liberals further divide into two paths based on familial support and phenotypic quality. One path is risky striving outside mainstream moral values. Those on this path are delinquents and outlaws who lack the ability and resources needed for effective social navigation in the mainstream. This limitation may be from low or unpredictable family support and/or inadequate intelligence or social skills. The second path for liberals is also risky but involves social navigation while not violating the most hallowed rules of the mainstream. Given the liberal cognitive style of openness and creativity, high rewards may result. These liberals have the family support or phenotypic quality that makes the second path a feasible option. Some of those on this path generate novel culture by art, invention, and science. In contrast, conservatives are social players in that part of the mainstream involving maintenance of tradition, avoidance of the novel, adherence to rules, and respect for authority.

Acknowledgments

We thank Eric Charnov, Martin Daly, Chris Eppig, Steve Gangestad, Kenneth Letendre, Frank Sulloway, Robert Trivers, Paul Watson, and an anonymous reviewer for their useful comments on the study and/or manuscript. For help with data collection, we thank Robert Frankis, and for help with data processing, we thank Sarah Hamilton and Stacey Martinez.

References

Altemeyer, 1996 1.Altemeyer B. The authoritarian specter. Cambridge, MA: Harvard University Press; 1996;.

Belsky et al., 1991 2.Belsky J, Steinberg L, Draper P. Childhood experience, interpersonal development, and reproductive strategy: An evolutionary theory of socialization. Child Development. 1991;62:647–670. MEDLINE | CrossRef

Benet-Martínez & John, 1998 3.Benet-Martínez V, John OP. Los Cinco Grandes across cultures and ethnic groups: Multitrait multimethod analyses of the Big Five in Spanish and English. Journal of Personality and Social Psychology. 1998;75:729–750. MEDLINE | CrossRef

Bowlby, 1982 4.Bowlby J. 2nd ed.. Attachment and loss: Attachment. Vol. 1. London: The Hogarth Press; 1982;.

Charnov, 1993 5.Charnov EL. Life-history invariants: Some explorations of symmetry in evolutionary ecology. Oxford, England: Oxford University Press; 1993;.

Chisholm, 1999 6.Chisholm JS. Death, hope and sex: Steps to an evolutionary ecology of mind and morality. Cambridge: University Press; 1999;.

Chisholm et al., 2005 7.Chisholm JS, Quinlivan JA, Petersen RW, Coall DA. Early stress predicts age at menarche and first birth, adult attachment, and expected lifespan. Human Nature. 2005;16:233–265.

Eaves et al., 1997 8.Eaves L, Martin N, Heath A, Schieken R, Meyer J, Silberg J, et al.. Age changes in the causes of individual differences in conservatism. Behavior Genetics. 1997;27:121–124. MEDLINE | CrossRef

Feather, 1979 9.Feather NT. Value correlates of conservatism. Journal of Personality and Social Psychology. 1979;37:1617–1630. CrossRef

Forabosco & Ruch, 1994 10.Forabosco G, Ruch W. Sensation seeking, social attitudes and humor appreciation in Italy. Personality and Individual Differences. 1994;16:515–528.

Grossmann et al., 1985 11.Grossmann K, Grossmann KE, Spangler G, Suess G, Unzner L. Maternal sensitivity and newborns' orientation responses as related to quality of attachment in northern Germany. In: Monographs of the Society for Research in Child Development. Growing points of attachment theory and research. Vol. 50 (nos. 1/2):1985;p. 233–256.

Hazan & Shaver, 1987 12.Hazan C, Shaver P. Romantic love conceptualized as an attachment process. Journal of Personality and Social Psychology. 1987;52:511–524. MEDLINE | CrossRef

Holmes, 1996 13.Holmes J. Attachment theory: A secure base for policy?. In:  Kraemer S,  Roberts J editor. The politics of attachment: Toward a secure society. London: Free Association Books; 1996;p. 27–42.

Jost et al., 2003 14.Jost JT, Glaser J, Kruglanski AW, Sulloway FJ. Political conservatism as motivated social cognition. Psychological Bulletin. 2003;129:339–375. MEDLINE | CrossRef

Kirkpatrick, 2005 15.Kirkpatrick LA. Attachment, evolution, and the psychology of religion. New York: The Guilford Press; 2005;.

Knight, 1993 16.Knight K. Liberalism and conservatism. In:  Robinson JP,  Shaver PR,  Wrightsman LS editor. Measures of social psychological attitudes. Measures of political attitudes. Vol. 2:San Francisco: Sage; 1993;p. 59–158.

Rholes et al., 1997 17.Rholes WS, Simpson JA, Blakely BS, Lanigan L, Allen EA. Adult attachment styles, the desire to have children, and working models of parenthood. Journal of Personality. 1997;65:357–385. MEDLINE | CrossRef

Rholes et al., 2001 18.Rholes WS, Simpson JA, Campbell L, Grich J. Adult attachment and the transition to parenthood. Interpersonal Relations and Group Processes. 2001;81:421–435.

Roccato, in press 19.Roccato, M. (in press). Right-wing authoritarianism, social dominance orientation and attachment: An Italian study. Swiss Journal of Social Psychology.

Ross & Hill, 2002 20.Ross LT, Hill EM. Childhood unpredictability, schemas for unpredictability, and risk taking. Social Behavior and Personality. 2002;30:453–473.

Salmon & Daly, 1998 21.Salmon CA, Daly M. Birth order and familial sentiment: Middleborns are different. Evolution and Human Behavior. 1998;19:299–312.

Sidanius et al., 1994 22.Sidanius J, Pratto F, Bobo L. Social dominance orientation and the political psychology of gender: A case of invariance?. Journal of Personality and Social Psychology. 1994;67:998–1011. CrossRef

Simpson et al., 1996 23.Simpson JA, Rholes WS, Phillips D. Conflict in close relationships: An attachment perspective. Interpersonal Relations and Group Processes. 1996;71:899–914.

Singh-Manoux et al., 2003 24.Singh-Manoux A, Adler NE, Marmot MG. Subjective social status: Its determinants and its association with measures of ill-health in the Whiehall II study. Social Science & Medicine. 2003;56:1321–1333.

Strathman et al., 1994 25.Strathman A, Gleicher F, Boninger DS, Edwards CS. The consideration of future consequences: Weighing immediate and distant outcomes of behavior. Journal of Personality and Social Psychology. 1994;66:742–752. CrossRef

Sulloway, 1996 26.Sulloway FJ. Born to rebel: Birth order, family dynamics, and creative lives. New York: Pantheon; 1996;.

Zimbardo & Boyd, 1999 27.Zimbardo PG, Boyd JN. Putting time in perspective: A valid, reliable individual-differences metric. Journal of Personality and Social Psychology. 1999;77:1271–1288. CrossRef

Department of Biology, University of New Mexico, Albuquerque, NM 87131, USA

Corresponding author.

PII: S1090-5138(07)00020-7

doi:10.1016/j.evolhumbehav.2007.01.005



2007:12:16