SPR346421
21/12/09
3:42 am
Page 1
Modeling the hierarchical structure of personality-attachment associations: Domain diffusion versus domain differentiation Chris G. Sibley & Nickola C. Overall University of Auckland
ABSTRACT
This study examined the extent to which personality traits and attachment orientations within distinct relationship domains (romantic, familial, and friendship) exhibited diffuse versus differentiated associations with attachments toward specific persons (e.g., current romantic partner, mother, best friend). The results were consistent with a hierarchically organized model of the attachment representational network containing global (personality) and specific (attachment) components. Autonomous and sociotropic personalities (controlling for the Big Five) were diffusely associated with attachments toward specific persons across domains, whereas domain-specific attachments were associated with specific attachments only within corresponding domains. Domain-specific attachment representations mediated the links between autonomy and sociotropy and relationship-specific attachments within corresponding domains. Autonomy and sociotropy capture global summaries of the same regularities in responding that underlie the attachment system. KEY WORDS: attachment • autonomy • Big Five • hierarchical linear
modeling • personality • sociotropy
Both authors contributed equally to this manuscript. All correspondence concerning this article should be addressed to Chris Sibley, Department of Psychology, The University of Auckland, Private Bag 92019, Auckland 1142, New Zealand [e-mail:
[email protected]]. ??????? was Action Editor on this article. Journal of Social and Personal Relationships © The Author(s), 2010. Reprints and permissions: http://www.sagepub.co.uk/journalsPermissions.nav, Vol. 27(1): 1–24. DOI: 10.1177/0265407509346421
SPR346421
21/12/09
2
3:42 am
Page 2
Journal of Social and Personal Relationships 27(1)
A key premise of attachment theory is that people develop multiple working models of their different attachment partners (including romantic partners, parents, siblings, and close friends) (Bowlby, 1969/1982). These cognitive representations specify the likelihood of attachment figure availability and the viability of proximity seeking in times of distress and underlie regularities in the way individuals respond within close relationship contexts. Two attachment dimensions capture tendencies to use hyperactivating strategies indexed by attachment anxiety (such as increasing proximity seeking and eliciting attention from an attachment figure) and deactivating strategies indexed by attachment avoidance (such as withdrawing and suppressing proximity-seeking motivations) in order to regulate attachment insecurity (Mikulincer & Shaver, 2003). Self-report measures of attachment anxiety and avoidance designed to tap these cognitive representations therefore reflect individual differences in the specific if . . . then . . . behavioral contingencies (as discussed by Mischel & Shoda, 1995, 1999) that regulate behavior in attachment-relevant contexts. Collins and Read (1994) were the first to systematically outline how representations of multiple attachment figures are represented and organized, detailing three levels of a hierarchically organized attachment network. Representations summarizing the behavioral contingencies most effective in regulating attachment within relationships with specific persons, such as a current romantic partner (relationship-specific representations), are hypothesized to be nested under more global and abstract representations summarizing the effectiveness of regulatory strategies in different relationship domains, such as romantic relationships in general (domain-specific representations) (Collins & Read, 1994; Overall, Fletcher, & Friesen, 2003). Domain-specific representations are, in turn, nested under more global summaries of the contingencies most likely to apply across all relationship domains (including romantic, familial, and friendship domains). As Collins and Read (1994) suggested, the most global level of the attachment network should resemble a dispositional or trait-like way of responding that is consistent across a wide range of relationships and domains (see also Mikulincer & Shaver, 2003 for a discussion of the development of these multiple representations). However, despite impressive advances in our understanding of the processes underlying attachment system functioning within specific relationship contexts, empirical research examining the ways in which the attachment representational system is cognitively organized and integrated within a wider personality system remains relatively limited. Addressing this lacuna, we have recently presented a model that integrates research on attachment with research regarding the personality dimensions of autonomy and sociotropy identified within the literature on the personality diatheses of depression (Sibley, 2007; Sibley & Overall, 2007, 2008a, 2008b). This framework operationalizes autonomy and sociotropy as broadbandwidth (personality-level) measures that assess global summaries in two distinct patterns of if . . . then . . . regularities in interpersonal behavior that relate to the (more specific) dimensions of attachment avoidance and anxiety identified in the attachment literature. This study tests key ideas of
SPR346421
21/12/09
3:42 am
Page 3
Sibley & Overall: Modeling attachment representations
3
this account by using Hierarchical Linear Modeling (HLM) to assess the unique associations between autonomy and sociotropy and the lower levels (relationship-specific and domain-specific) of the attachment representational network. Assessing the attachment network Recent empirical findings support the key premises derived from Collins and Read’s (1994) outline of the structure and organization of attachment representations. Firstly, supporting the existence of a trait-like orientation, Pierce and Lydon (2001) provided evidence that a global and non-specific measure of attachment (how participants “felt in their close relationships, including their relationships to parents, siblings, close friends, relatives and their romantic partner”) predicted significant variance in diary ratings of social interaction quality across individuals’ interactions with close others, including their mother, father, closest friend, and romantic partner. Secondly, however, Pierce and Lydon (2001) reported that relationship-specific representations of close others (e.g., attachment toward the mother) predicted notably higher proportions of variance in the overall quality and intimacy of social interactions with that close other (e.g., interaction with the mother) than did more global models. Similarly, Cozzarelli, Hoekstra, and Bylsma (2000) found that attachment ratings of one’s current romantic partner were more strongly predictive of outcomes within participants’ romantic relationships than more global ratings referring to others in general. Finally, Sibley and Overall (2007) reported that domain-specific romantic attachment (assessing both previous and current romantic relationship experiences) was more strongly predictive of relationship-specific attachment ratings of particular romantic partners than attachment toward family members and friends. Supporting Collins and Read’s (1994) conceptualization, these and related findings (see Overall et al., 2003; Ross & Spinner, 2001) suggest that people possess multiple attachment representations that differ in specificity, including representations specific to particular relationships (relationship specific), as well as those that are broader in bandwidth and reflect regularities in attachment orientation across relationships, including global representations that describe attachment orientations across a range of relational contexts (e.g., Pierce & Lydon, 2001) and those that describe attachment orientations in particular relationship domains (domain-specific representations; e.g., Ross & Spinner, 2001; Sibley & Overall, 2007). Findings across these studies also suggest, however, that more specific relationship representations (e.g., romantic representations) will be most predictive of interpersonal behaviors, cognitions, and emotions within congruent contexts (e.g., with a romantic partner) than global representations (e.g., across relationship domains) (Sibley & Liu, 2006; Sibley & Overall, 2008b). This pattern is also consistent with Collins and Read’s (1994) theorizing regarding when different aspects (or levels) of the attachment hierarchy
SPR346421
21/12/09
4
3:42 am
Page 4
Journal of Social and Personal Relationships 27(1)
should most likely govern interpersonal behavior. To paraphrase Collins and Read (1994, p. 58), more specific representations should be activated in and guide behavior and emotion in relational contexts congruent with that model, whereas more global and abstract representations at the top of the hierarchy are theorized to function as “default models that encompass both close and non-close social relations and hence apply to a wide range of situations although they may not accurately fit any given situation particularly well.” A hierarchically organized attachment representational network such as this provides the flexibility to operate in functional ways across relationship contexts, such that the most accurate and context-specific representations are more strongly activated to direct relational responding (Collins & Read, 1994; Overall et al., 2003). A hierarchically organized network should also produce a specific pattern of differential associations across measures of domain-specific and relationship-specific attachment. Attachments within similar types of relationships should be more closely associated because they are more likely to encompass similar interpersonal experiences and govern attachment system functioning in similar contexts. Accordingly, associations between domain-specific attachment orientations (e.g., how one regulates responding in romantic relationships) should be strongly associated with attachment orientations toward specific relationships within that same domain (e.g., attachment toward specific past and current romantic partners), but should be more weakly associated with ratings of relationship-specific attachment in other domains (e.g., attachment toward specific family members or friends). Using HLM, Sibley and Overall (2008a) found solid support for this pattern of differential associations across two independent studies. Considering the romantic relationship context, for example, shared (withinperson) consistencies in attachments toward past and current romantic partners were predicted by a general domain-specific romantic attachment orientation, but not by domain-specific orientations referring to friendship or familial domains. In contrast to domain-specific representations, the most global level of the hierarchical representational network should summarize regularities in responding across all relationships, and therefore should be relatively evenly associated with attachment toward specific people across domains. Furthermore, if the representational network is structured in the hierarchical manner proposed, the associations between representations at the most global level with relationship-specific attachment at the lowest level should occur via intermediary-level models of attachment within the corresponding domain, because these more specific models contain more detailed information that map directly, and therefore more accurately predict, responding within that context.This study evaluates these predictions by testing whether the personality dimensions of autonomy and sociotropy demonstrate the specific set of associations that the trait-like global level of the hierarchy should (as described above).
SPR346421
21/12/09
3:42 am
Page 5
Sibley & Overall: Modeling attachment representations
5
Autonomy and sociotropy: Broad-bandwidth measures of relational responding In recent years, researchers have begun to consider the relationship between adults’ models of romantic relationships and more generalized autonomous and sociotropic dispositions identified in the literature on personality vulnerabilities to depression (e.g., Besser & Priel, 2003; Murphy & Bates, 1997; Rude & Burnham, 1995; Zuroff & Fitzpatrick, 1995). Beck (1983, p. 272) defined the personality dimension of autonomy as an index of the individual’s level of “investment in preserving his [sic] independence, mobility, and personal rights; freedom of choice, action, and expression; protection of his [sic] domain; and defining his [sic] boundaries.” Sociotropy, in contrast, reflects the individual’s level of “investment in positive interchange with other people. This cluster includes passive-receptive wishes (acceptance, intimacy, understanding, support, guidance); ‘narcissistic wishes’ (admiration, prestige, status); and feedback validation of beliefs and behavior.” A similar distinction was also proposed by Blatt, D’Afflitti, and Quinlan (1976), who used the terms self-criticism and dependency (respectively) to refer to two conceptually similar dimensions. Consistent with the dynamic interactionist model of depression (Blatt & Zuroff, 1992), research indicates that individuals high in sociotropy (or dependency) may be particularly prone to depression when interpersonal relationships and social support are threatened, whereas autonomy (or self-criticism) confers vulnerability to depression when the personal goals and achievements of the individual are threatened. Autonomy and sociotropy are remarkably similar to recent definitions of attachment anxiety and avoidance. In their comprehensive summary of the attachment behavioral system, Shaver and Mikulincer (2002, pp. 135–136), for example, defined attachment anxiety as the predisposition for an “intense need to be close, accepted, supported, and reassured” by attachment figures, whereas attachment avoidance reflects a tendency to be “uncomfortable with closeness, self-disclosure, feelings and expressions of vulnerability, and dependency” in attachment relationships. However, consistent with the distinction between global and specific attachment representations, the latter two constructs are more broadly defined and tap global regularities in relational responding that describe behavior across a range of different contexts, whereas attachment anxiety and avoidance refer to patterns of relational responding within particular relationships or domains (Sibley, 2007). Crossovers across the two bodies of work also indicate similarities between attachment and depressive personality. Recent research, for example, has demonstrated that greater attachment anxiety and avoidance heighten vulnerability to depression, particularly with regard to stressful events that involve the individual’s relationships with attachment figures (Simpson, Rholes, Campbell, Tran, & Wilson, 2003). Moreover, individual differences in autonomy and sociotropy predict similar patterns of responding in romantic relationships. Like individuals high in attachment anxiety (e.g., Campbell, Simpson, Boldry, & Kashy, 2005; Collins, 1996), highly sociotropic individuals typically perceive their romantic partners as withdrawing
SPR346421
21/12/09
6
3:42 am
Page 6
Journal of Social and Personal Relationships 27(1)
and describe their own behavior in romantic relationships as demanding (Lynch, Robins, & Morse, 2001). In contrast, like those high in attachment avoidance (e.g., Simpson, Rholes, & Phillips, 1996), people high in autonomy generally perceive their partners as demanding and their own behaviors as withdrawing (Lynch et al., 2001), and are more likely to express hostility and withdrawal during conflict interactions (Mongrain, Vettese, Shuster, & Kendal, 1998; Zuroff & Duncan, 1999). Finally, the conceptual and empirical similarities described above are also supported by the statistical associations between these measures. In a meta-analysis of previous research, Sibley (2007) provided evidence that although sociotropy and autonomy were psychometrically distinct from attachment anxiety and avoidance, sociotropy and autonomy were nevertheless moderately positively associated with domain-specific ratings of anxiety and avoidance (respectively) within the romantic domain. Taken together, this prior work provides good evidence that autonomy and sociotropy index global regularities in relational responding akin to the highest level of the attachment hierarchy proposed by Collins and Read (1994). Based on the discussion above, if autonomy and sociotropy do reflect more global (broad-bandwidth) indices of attachment avoidance and anxiety, then autonomy and sociotropy should (i) exhibit diffuse and comparable significant associations with measures of relationship-specific attachment across multiple-relational domains (romantic, familial, and friendship), but (ii) the links between autonomy and sociotropy and relationship-specific attachment should occur via domain-specific representations within the corresponding domain, which more accurately describe and regulate responding within that context. Sibley and Overall (2007) provided preliminary evidence for this hypothesized pattern, reporting that (i) autonomy and sociotropy predicted variation in levels of attachment avoidance and anxiety toward a range of undifferentiated others (including romantic partner, friends, and family), but (ii) domain-specific romantic attachment mediated these effects when predicting relationship-specific attachment with romantic, but not nonromantic, partners. However, Sibley and Overall (2007) did not test whether the three domain-specific representations hypothesized to form the attachment network (family, friendship, and romantic domains) mediated the links between autonomy and sociotropy and relationship-specific representations, and importantly, if they only mediated these links within each corresponding domain. The present research addresses this issue, and also assesses whether the personality dimensions of autonomy and sociotropy uniquely predict variation in attachment at different levels of the hierarchy above and beyond other core personality dimensions. Big Five personality and the attachment representational network Prior research has not tested whether other central personality dimensions, such as the Big Five, capture regularities in attachment across relationships and relationship domains. The “Big Five” refers to broad-bandwidth dimensions of personality or dispositional tendencies, including: Extraversion,
SPR346421
21/12/09
3:42 am
Page 7
Sibley & Overall: Modeling attachment representations
7
Agreeableness, Conscientious, Neuroticism, and Openness to Experience. Research indicates that autonomy and sociotropy, and attachment avoidance and anxiety, are distinct from the Big Five and predict unique variance in a range of outcomes (such as relationship quality) above and beyond variation in the (presumably) more broad-bandwidth generalized aspects of personality (Dunkley, Blankstein, & Flett, 1997; Dunkley, Blankstein, Zuroff, Lecce, & Hui, 2006; Noftle & Shaver, 2006; Shaver & Brennan, 1992; Zuroff, 1994). Considering the relationship between the Big Five and attachment, Noftle and Shaver (2006), for example, reported that broad-bandwidth indicators of the Big Five failed to predict unique variance in relationship quality when controlling for attachment anxiety and avoidance, whereas anxiety and avoidance retained significant predictive ability controlling for the Big Five. However, extant research does indicate that autonomy and sociotropy, and avoidance and anxiety, are systematically related to different dimensions of the Big Five at moderate-to-high levels. Dunkley et al. (1997, 2006) reported that dependency (a construct similar to sociotropy) and selfcriticism (a construct similar to autonomy) tended to correlate with the Neuroticism range at around .50–.60. Attachment also tends to show lower, more moderate correlations with the Big Five. In a sample of more than 8000 people, Noftle and Shaver (2006) reported that attachment anxiety correlated with Neuroticism at r = .42, whereas avoidance correlated with Neuroticism at r = .14. Anxiety and avoidance were also weakly negatively correlated with the other four dimensions of the Big Five (around –.20) for their association with Agreeableness. Impressively, Donnellan, Burt, Levendosky, and Klump (2008) have further reported that a substantial majority of the covariance in attachment and personality appears to be due to shared genetic influences. Thus, although research indicates that personality and attachment are distinct and predict unique variance in outcomes such as relationship quality (Noftle & Shaver, 2006), there is nevertheless a strong shared genetic component to both sets of constructs (Donnellan et al., 2008). These observations raise the possibility that individual differences in sociotropy and autonomy, and attachment anxiety and avoidance, may overlap with Big Five personality factors, such as Neuroticism or Agreeableness. Thus, to test our hypothesis that autonomy and sociotropy uniquely represent personality-level regularities in the attachment network, we considered it important to also include the Big Five when testing a hierarchical model of the three-tier attachment network, including global interpersonal tendencies, domain-specific representations, and attachment orientations in relationships with specific others. This allowed us to control for the possibility that associations between sociotropy and attachment anxiety, and autonomy and attachment avoidance, may occur merely because of the shared associations of these constructs with the Big Five, rather than representing any meaningful pattern of associations beyond that explained by variation in broad-bandwidth indicators of Big Five personality.
SPR346421
21/12/09
8
3:42 am
Page 8
Journal of Social and Personal Relationships 27(1)
The present research The present research extended our earlier work (Overall et al., 2003; Sibley, 2007; Sibley & Overall, 2007, 2008a, 2008b) by testing associations across all components of the attachment representational hierarchy within a single integrated study. To do this we measured autonomy and sociotropy (global level of hierarchy), domain-specific representations of general attachment within family, friendship, and romantic relationship domains (mid-level of hierarchy), and relationship-specific attachment toward a variety of persons within each of the three relationship domains (lowest level of hierarchy). We tested the following three hypotheses. Hypothesis 1: Diffuse personality effects. We predicted that if autonomy and sociotropy represent global summaries in the two patterns of if . . . then . . . regularities that relate to the more specific dimensions of attachment avoidance and anxiety, then we should see evidence of domain diffusion. Specifically, autonomy and sociotropy should exhibit diffuse and comparable significant associations with respective measures of relationship-specific attachment anxiety and avoidance across all three relational domains (romantic, familial, and friendship). Importantly, we also extended previous research by testing (and controlling for) the extent to which the Big Five dimensions of personality were systematically associated with attachment across domains. We expected that all predicted patterns of association between autonomy, sociotropic, and attachment orientations at different levels of the representational network would remain consistent when controlling for broad-bandwidth markers of the Big Five. Hypothesis 2: Domain differentiated attachment effects. Secondly, replicating Sibley and Overall (2008a), we predicted that domain-specific attachment representations located at the mid-level tier of the network would be strongly associated with attachment toward specific relationships within the corresponding domain but – unlike autonomy and sociotropy – would not be associated with ratings of relationship-specific attachment in other domains. Hypothesis 3: Domain-congruent indirect effects. Finally, integrating the above hypotheses, we predicted that domain-specific attachment representations would mediate the associations between autonomy and sociotropy with relationship-specific attachments, but do so only within corresponding domains. Autonomy, for example, should be associated with relationshipspecific attachment avoidance in interactions with specific romantic partners, family members, and friends. However, the diffuse effects of autonomy on more specific attachment orientations should be mediated by mid-level attachment representations that more accurately reflect the contingencies that govern how the individual responds within the specific relational domain. Moreover, these domain-congruent indirect effects should be evidently controlling for the Big Five dimensions of personality.
SPR346421
21/12/09
3:42 am
Page 9
Sibley & Overall: Modeling attachment representations
9
Method Participants The participants were 179 university students (76 men, 103 women; Mage = 22.92, SDage = 6.37). The participants had been previously involved in one or more serious romantic relationships and the majority were romantically involved (91%). The participants were recruited via posted advertisements and paid NZ$10. Measurement Personality measures. Sociotropy and autonomy were measured using the 24-item scales developed by Robins et al. (1994). Example items include “I tend to keep other people at a distance” (autonomy) and “I get uncomfortable when I’m not sure whether or not someone likes me” (sociotropy). The participants also completed the 50-item International Personality Item Pool (IPIP) inventory (Goldberg, 1999), which includes 10 items assessing each of the Big Five personality dimensions: Extraversion (e.g., “Am the life of the party”), Agreeableness (e.g., “Am interested in people”), Conscientiousness (e.g., “Pay attention to details”), Neuroticism (e.g., “Get stressed out easily”), and Openness to Experience (e.g., “Enjoy thinking about things”). Attachment measures. Global attachment orientation within each relationship domain (family, friends, and romantic partners) was assessed using the Adult Attachment Questionnaire (AAQ; Simpson et al., 1996). The participants completed three versions of the 17-item scale, which assessed attachment anxiety and avoidance toward their “close family relationships in general”, their “close platonic (non-sexual) friendships in general”, and their “close romantic relationships in general.” Consistent with prior research (Overall et al., 2006; Sibley & Overall, 2008a), in order to assess relationship-specific attachments the participants identified their three most important current and/or past relationships within each domain: family (excluding children), friends, and romantic relationships (including current romantic partner). For each of the nine nominated relationships, participants rated six modified AAQ items in terms of how well each item described their feelings and experiences within that specific relationship. Three items assessed attachment anxiety, including (worded for specific family members) “I often worry that my family member doesn’t really love me,” “I’m confident that my family member loves me just as much as I love them” (reverse coded), and “I usually want more closeness and intimacy than my family member does.” Three items assessed attachment avoidance including (worded for specific friends) “I find it relatively easy to get close to my friend” (reverse coded), “I don’t like my friend getting too close to me,” and “My friend often wants me to be more intimate than I feel comfortable being.”
SPR346421
21/12/09
10
3:42 am
Page 10
Journal of Social and Personal Relationships 27(1)
All items were rated on scales ranging from 1 (strongly disagree) to 7 (strongly agree), and the order of (i) personality, domain-specific and relationship-specific measures and (ii) domain presentation (e.g., family, friends, romantic partners) was randomized. Cronbach’s alphas for the three item measures of relationship-specific anxiety for each of the nine reported relationships (three for each of the three domains) ranged from .63 to .76. Cronbach’s alphas for the three item measures of relationship-specific avoidance for each of the nine reported relationships were slightly lower with ratings for these relationships ranging from .55 to .64. There was, however, one exception, with ratings of anxiety toward the third listed friend displaying an alpha of .45. Given that all analyses remained consistent regardless of whether or not this third rating was included, we therefore opted to report all analyses with the measures included. As reported in Table 1, measures of the Big Five, sociotropy, autonomy and domain-specific attachment anxiety, and avoidance also all demonstrated acceptable internal reliability with Cronbach’s alphas >.70. Results Descriptive statistics and correlations between the personality and domainspecific attachment measures are presented in Table 1. Autonomy was positively associated with domain-specific avoidance (average r = .43) and sociotropy was positively correlated with domain-specific attachment anxiety (average r = .30) across domains. Domain-specific measures of attachment avoidance and anxiety were also positively correlated (average withindomain r = .29 and .36). After controlling for these shared associations, we predicted that: (i) autonomy and sociotropy would be associated with attachments to specific persons across all domains (diffuse personality effects); (ii) global ratings of attachment avoidance and anxiety in the romantic, familial, and friendship domains would predict attachment toward specific relationship partners in congruent domains only (domain-differentiated attachment effects); and (iii) the effects of autonomy and sociotropy would be mediated by individuals’ attachment orientation within each domain (domain-congruent indirect effects). The measurement of attachment ratings referring to multiple specific relationships (within persons) and personality and global attachment orientations referring to different domains (between person) form a hierarchical data structure. The lower level (Level 1) comprise the multiple relationshipspecific ratings recorded by each participant, which are in turn nested within personality and domain-specific ratings at the higher level (Level 2). HLM is uniquely equipped to test predictions regarding the magnitude of associations within hierarchical data structures (see Raudenbush & Bryk, 2002). We first used the HLM statistical package to examine within-person consistencies in attachment toward specific persons by decomposing the variation in relationship-specific attachment into two components: that occurring at Level 2 (systematic variation indicative of differences that occurred at the
6.37
SD
*p < .05
22.92
.20* .08 .21* –.14 .14 –.11 –.15* –.04 –.07 –.04 –.04 –.01 .06
M
–.12 .10 –.07 –.12 –.28* .04 –.15* .03 –.04 –.04 –.06 .02 –.10 –.33*
2
.85
1.10
4.40
.22* .08 –.27 .19* –.13 –.14 –.19* –.09 –.13 –.17* –.26* –.06
3
.77
.82
5.28
.14 –.09 .25* .33* –.25* –.16* –.02 –.14 –.18* –.24* –.09
4
6
.78
.97
4.56
.81
1.25
4.12
.03 .19* –.07 –.08 .43* .01 .46* .02 .25* –.01 .28* –.14 .34* .00 .32* –.02 .30* –.07 .18*
5
.78
.84
5.17
–.16 .01 –.15* –.14 –.28* –.07 –.06 –.22*
7
.86
.83
4.58
.29* .22* .35* .34* .23* .29* .22*
8
.82
.77
4.13
.28* .22* .19* .34* .49* .45*
9
.75
1.01
2.28
.45* .21* .55* .32* .12
10
.85
.96
2.86
.42* .32* .39* .23*
11
.73
1.14
3.17
.20* .15* .25*
12
.85
1.30
2.94
.33* .12
13
.80
1.04
3.06
.40*
14
.82
1.18
3.21
15
3:42 am
1. Gender 2. Age 3. Extraversion 4. Agreeableness 5. Conscientiousness 6. Neuroticism 7. Openness 8. Sociotropy 9. Autonomy 10. Domain-specific familial anxiety 11. Domain-specific friendship anxiety 12. Domain-specific romantic anxiety 13. Domain-specific familial avoidance 14. Domain-specific friendship avoidance 15. Domain-specific romantic avoidance
1
21/12/09
TABLE 1 Descriptive statistics and correlations between personality, sociotropy, autonomy, and domain-specific attachment anxiety and avoidance
SPR346421 Page 11
Sibley & Overall: Modeling attachment representations 11
SPR346421
21/12/09
12
3:42 am
Page 12
Journal of Social and Personal Relationships 27(1)
between-person level), and variation occurring at Level 1 (remaining withinperson variation indicative of idiosyncratic differences in orientation toward multiple relationship partners that was unexplained by differences across individuals and includes unidentified measurement error). Separate analyses were conducted for relationship-specific ratings of attachment avoidance and anxiety. The baseline intercept model, yij = 0j + rij , showed that 21.2% of the variance in ratings of relationship-specific avoidance, and 21.5% of the variance in relationship-specific anxiety, occurred at the between-person level. This suggests that a substantial portion of the variance in relationshipspecific attachment ratings was systematic across a variety of relationships with specific others. This systematic variation indicates regularities in attachment orientations across relationships – regularities that are potentially attributable to more global individual differences, such as domain-specific attachment and/or personality. We next used HLM to examine how domain-specific attachment and personality measures predicted the between-person variation in ratings of relationship-specific attachment across relational domains.We did this in two stages. In the first model, we assessed the links between personality variables and attachments to specific persons across all domains (Hypothesis 1: Diffuse personality effects). In the second model, we entered global ratings of attachment anxiety and avoidance within each domain into the analyses to test whether domain-specific attachment would predict relationshipspecific attachment within congruent domains only (Hypothesis 2: Domaindifferentiated attachment effects), and whether the effects of sociotropy and autonomy were mediated by domain-specific attachment within congruent domains (Hypothesis 3: Domain-congruent indirect effects). Separate sets of analyses were conducted for attachment avoidance and anxiety. Model 1: Assessing the diffuse effects of personality. Firstly, to test the proposed diffuse effect of personality on attachment within specific relationships, we ran models in which autonomy, sociotropy, and the Big Five dimensions of personality predicted the multiple relationship-specific ratings of avoidance (or anxiety). Relationship domains were represented by three dummy coded (0,1) variables indicating whether relationship-specific attachment ratings referred to relationships with people categorized as family members, friends, or romantic partners. yij = 1j (familial relationships) + 2j (friendships) + 2j (romantic relationships) + rij
(1)
In this equation yij represents ratings of attachment avoidance (or anxiety) toward specific relationship partners (subscripted i) made by each participant (subscripted j), 2j represents average attachment ratings within familial relationships, 2j represents attachment ratings within friendships, 3j represents attachment within romantic relationships, and rij represents the Level 1 error. Note that this equation did not include an intercept, thus the coefficients for the relationship domain dummy codes represent scores for
SPR346421
21/12/09
3:42 am
Page 13
Sibley & Overall: Modeling attachment representations
13
relationship-specific avoidance (or anxiety) for each participant’s (j) rating of their attachment orientation within that domain (i). At Level 2, demographics (gender, age), Big Five measures of personality, and autonomy and sociotropy were entered as predictors of the Level 1 relationship-specific measures of avoidance (and anxiety) within relationships with family members, friends, and romantic partners. Shown in Equation (2), this Level 2 model tested simultaneous but separate regression equations within each relational domain. Familial relationships
1j = γ 10 + γ 11(gender) + γ 12(age) + γ 13(extraversion) + γ 14(agreeableness) + γ 15(conscientiousness) + γ 16(neuroticism) + γ 17(openness) + γ 18(sociotropy) + γ 19(autonomy) + u1j
Friendships
1j = γ 20 + γ 21(gender) + γ 22(age) + γ 23(extraversion) + γ 24(agreeableness) + γ 25(conscientiousness) + γ 26(neuroticism) + γ 27(openness) + γ 28(sociotropy) + γ 29(autonomy) + u2j
Romantic relationships
1j = γ 30 + γ 11(gender) + γ 32(age) + γ 33(extraversion) + γ 34(agreeableness) + γ 35(conscientiousness) + γ 36(neuroticism) + γ 37(openness) + γ 38(sociotropy) + γ 39(autonomy) + u3j
(2)
Examining the Level 2 equation predicting attachment avoidance in familial relationships (1j), for example, γ 10 represents the intercept for attachment avoidance in relationships with specific family members, γ 11 and γ 12 are coefficients controlling for the effects of gender and age, respectively, γ 13 – γ 17 are coefficients testing the extent to which the Big Five personality predicted attachment avoidance in specific familial relationships, γ 18 and γ 19 are coefficients testing the degree to which sociotropy and autonomy predicted avoidance in familial relationships, and u1j represents the Level 2 error. Similarly, coefficients subscripted γ 2x tested the extent to which these constructs predicted attachment avoidance in relationships with specific friends, and coefficients subscripted γ 3x tested the extent to which these constructs predicted avoidance toward specific romantic partners. The results predicting relationship-specific attachment avoidance are presented in the first column of Table 2 (Model 1, predicted effects are shown in bold). Autonomy was significantly positively associated with relationshipspecific avoidance across all three relational domains. In addition, although agreeableness and openness were negatively associated with avoidance in some domains, the diffuse effect of autonomy emerged above and beyond the Big Five dimensions of personality, as well as core demographics (gender and age). Similar diffuse effects of sociotropy were evident when examining associations with relationship-specific anxiety (see Model 1, Table 3). As predicted, sociotropy was significantly positively associated with relationshipspecific anxiety across all three relational domains. Agreeableness and openness were also significantly, although relatively weakly, associated with
SPR346421
21/12/09
14
3:42 am
Page 14
Journal of Social and Personal Relationships 27(1)
TABLE 2 HLM coefficients testing the associations between personality, domain-specific attachment avoidance, and ratings of relationship-specific attachment avoidance toward specific persons within each of these three relationship domains Model 1
Model 2
t
Familial relationships Gender Age Extraversion Agreeableness Conscientiousness Neuroticism Openness Sociotropy Autonomy Domain-specific familial avoidance Domain-specific friendship avoidance Domain-specific romantic avoidance
2.88 .06 .01 –.03 –.24 –.01 .03 –.04 .34 .39
.73 .91 –.35 –2.02* –.17 .39 –.42 2.97* 3.33*
Friendships Gender Age Extraversion Agreeableness Conscientiousness Neuroticism Openness Sociotropy Autonomy Domain-specific familial avoidance Domain-specific friendship avoidance Domain-specific romantic avoidance
2.62 .03 .01 –.02 –.17 –.05 –.02 –.20 .10 .23
.44 .84 –.37 –1.48 –.70 –.30 –2.64* .95 2.19*
Romantic relationships Gender Age Extraversion Agreeableness Conscientiousness Neuroticism Openness Sociotropy Autonomy Domain-specific familial avoidance Domain-specific friendship avoidance Domain-specific romantic avoidance
3.36 –.15 –.01 .13 .15 –.08 –.09 –.34 –.01 .65
–1.94 –.02 1.75 1.31 –1.22 –1.13 –3.13* –.07 5.87*
*p < .05
t
2.87 .01 .01 .04 –.03 –.05 –.05 –.08 .14 .19 .47 .22 –.05
.06 .41 .60 –.32 –.63 –.67 –.89 1.36 1.63 8.21* 2.79* –.64
2.63 .08 .01 .04 –.03 –.05 .01 –.22 –.06 –.03 –.01 .45 .03
1.09 .16 .78 –.30 –.78 .14 –2.92* –.58 –.22 –.08 6.15* .46
3.39 .07 –.01 .13 .13 –.02 .01 –.19 –.04 .21 .01 .06 .45
1.04 –.92 1.93 1.22 –.45 .06 –1.91 –.39 1.62 .03 .82 6.05*
SPR346421
21/12/09
3:42 am
Page 15
Sibley & Overall: Modeling attachment representations
15
TABLE 3 HLM coefficients testing the associations between personality, domain-specific attachment anxiety, and ratings of relationship-specific attachment anxiety toward specific persons within each of these three relationship domains Model 1
Model 2
t
Familial relationships Gender Age Extraversion Agreeableness Conscientiousness Neuroticism Openness Sociotropy Autonomy Domain-specific familial anxiety Domain-specific friendship anxiety Domain-specific romantic anxiety
2.39 .09 .03 –.05 –.21 .04 .02 –.21 .50 .26
1.22 1.94 –.87 –2.12* .50 .31 –2.39* 3.73* 2.40*
Friendships Gender Age Extraversion Agreeableness Conscientiousness Neuroticism Openness Sociotropy Autonomy Domain-specific familial anxiety Domain-specific friendship anxiety Domain-specific romantic anxiety
2.61 .01 .23 –.02 –.25 –.11 .04 –.17 .23 .16
.90 .48 –.26 –2.38* –1.42 .58 –1.96* 2.10* 1.56
Romantic relationships Gender Age Extraversion Agreeableness Conscientiousness Neuroticism Openness Sociotropy Autonomy Domain-specific familial anxiety Domain-specific friendship anxiety Domain-specific romantic anxiety
3.25 .14 .02 –.03 –.26 –.05 .05 –.12 .54 .12
1.91 2.13* –.52 –2.80* –.81 .71 –1.52 5.14* 1.25
*p < .05
t
2.38 .08 .02 –.02 –.16 –.01 .02 –.19 .40 .17 .39 .15 –.15
1.29 2.15* –.33 –1.83 –.16 .28 –2.16* 3.78* 1.81 5.26* 1.50 –2.34*
2.60 .05 .01 –.01 –.18 –.13 –.01 –.11 .03 .11 .11 .45 .02
.73 .33 –.23 –1.99* –2.04* –.15 –1.49 .32 1.14 1.59 6.26* .26
3.24 .12 .01 –.04 –.14 –.04 –.02 –.01 .31 .09 .11 .12 .36
1.98* 1.34 –.70 –1.71 –.62 –.34 –.08 2.84* 1.20 1.53 1.55 5.80*
SPR346421
21/12/09
16
3:42 am
Page 16
Journal of Social and Personal Relationships 27(1)
relationship-specific anxiety across domains, but these associations did not reduce the significant diffuse effects of sociotropy. Model 2: Assessing the differentiated and mediating effects of domainspecific attachment. Next, we repeated the above analyses but added domainspecific measures of general attachment within familial, friendship, and romantic relationships as Level 2 predictors of attachment within specific relationships across these three domains. This allowed us to test whether domain-specific measures predicted relationship-specific attachment within corresponding domains only (Hypothesis 2: Domain-differentiated attachment effects), and whether domain-specific attachment mediated the links between more global personality differences and relationship-specific attachment within corresponding domains (Hypothesis 3: Domain-congruent indirect effects). The results predicting relationship-specific avoidance are shown in the second column of Table 2 (Model 2). Unlike the diffuse effects observed for autonomy, domain-specific measures of attachment avoidance were differentially associated with relationship-specific avoidance. As predicted, domainspecific familial avoidance was most strongly associated with ratings of avoidance toward specific family members, domain-specific friendship avoidance was only significantly associated with avoidance toward specific friends, and domain-specific romantic avoidance was only associated with avoidance within specific romantic relationships. The same differential pattern of associations was evident between domain-specific measures of attachment anxiety and ratings of anxiety within specific relationships (see Table 3, Model 2). Next, we examined whether the diffuse associations between autonomy and sociotropy with relationship-specific avoidance and anxiety across domains was differentially mediated by domain-congruent attachment orientations within each relational domain (Hypothesis 3: Domain-congruent indirect effects). Providing preliminary support for this prediction, when including domain-specific attachment within the model, the effect of autonomy on relationship-specific avoidance became non-significant across all three relationship domains (see Model 2,Table 2) and the effect of sociotropy on relationship-specific anxiety became non-significant within the romantic domain, was substantially reduced in the friendship domain, but retained a comparable effect in the familial domain (see Model 2, Table 3). To formally assess the domain-congruent indirect effects we tested for multi-level mediation. The (upper-level) multi-level mediational models took the following form: Independent Variable (Level 2) → Mediating Variable (Level 2) → Dependent Variable (Level 1) and followed the procedure described by Krull and MacKinnon (2001). For example, when testing the mediational model where autonomy (Level 2) → domain-specific romantic avoidance (Level 2) → relationship-specific romantic avoidance (Level 1), we (i) calculated the unstandardized regression coefficient (Ba) and standard error (sea) for the effect of autonomy on domain-specific romantic avoidance using standard multiple regression (controlling for
SPR346421
21/12/09
3:42 am
Page 17
Sibley & Overall: Modeling attachment representations
17
gender, age, the Big Five, and sociotropy), and then (ii) calculated the unstandardized gamma coefficient (γ b) and standard error (seb) for the effect of domain-specific romantic avoidance on the relationship-specific avoidance across the multiple romantic relationships recorded by each participant (continuing to control for all other Level 2 constructs). As outlined by Krull and MacKinnon (2001), the significance of the indirect effect was assessed using Sobel’s z, which was calculated using the first-order Taylor series expansion recommended by Krull and MacKinnon (2001): Sobel’s z = Ba × γ b /sab, where sab = √(γ b2 × sea2 + Ba2 × seb2). These results provide support for the predicted domain-congruent indirect effects. Domain-specific avoidance mediated the effect of autonomy on relationship-specific avoidance (Sobel’s z = 4.84, p < .05) and domainspecific romantic anxiety mediated the effect of sociotropy on relationshipspecific anxiety with specific romantic partners (Sobel’s z = 2.85, p < .05). Importantly, domain-specific attachments referring to other domains (i.e., familial and friendship) did not mediate the effects of autonomy and sociotropy within the romantic domain (Sobel’s z < 1.40). Similarly, domainspecific friendship avoidance mediated the effect of autonomy (Sobel’s z = 3.81, p < .05) and domain-specific friendship anxiety mediated the effect of sociotropy (Sobel’s z = 2.94, p < .05) on relationship-specific attachment ratings within the congruent domain, and domain-specific attachment in other domains did not mediate the effects of autonomy and sociotropy within the friendship domain (Sobel’s z < 1.45). However, when predicting relationship-specific attachment within the familial domain, domain-specific familial avoidance did not mediate the effect of autonomy (Sobel’s z = 1.76) and domain-specific familial anxiety did not mediate the effect of sociotropy (Sobel’s z = 1.76), although these indirect effects were marginally significant (p < .10). Discussion This study assessed three levels of the hierarchically organized attachment representational network proposed to contain both global (personality) and specific (attachment) components. Consistent with Hypothesis 1 (diffuse personality effects), autonomy was associated with relationship-specific avoidance across specific relationships with romantic partners, family members, and friends, whereas sociotropy was associated with relationship-specific anxiety across these multiple relationships. Importantly, the diffuse effects of autonomy and sociotropy emerged after controlling for the Big Five dimensions of personality. This provides strong evidence that autonomy and sociotropy assess global summaries of the regularities in relational responding that parallel those regularities represented by attachment avoidance and anxiety – regularities that are not assessed in the same detail by broad-bandwidth markers of the Big Five. Consistent with Hypothesis 2 (domain differentiated attachment effects), domain-specific measures of attachment avoidance and anxiety were strongly
SPR346421
21/12/09
18
3:42 am
Page 18
Journal of Social and Personal Relationships 27(1)
associated with the attachment ratings of specific relationships within congruent domains, but were not (in general) associated with ratings of relationship-specific attachment in other domains. These results replicate the findings of Sibley and Overall (2008a), and further indicate that domainspecific measures of attachment reliably assess consistencies in the representations individuals hold of specific relationships within a given relationship domain (e.g., how one generally responds across their romantic relationships). Again, these domain differentiated effects were observed controlling for the Big Five, and autonomy and sociotropy, revealing that the associations between measures of attachment worded to assess different levels of the attachment network are not merely attributable to more global personality constructs. Consistent with Hypothesis 3 (domain-congruent indirect effects), domainspecific attachment representations generally mediated the links between global differences in autonomous and sociotropic personality and relationship-specific attachments, but did so only within congruent domains. For example, domain-specific romantic avoidance mediated the link between autonomy and avoidance within relationships with previous and current romantic partners, and domain-specific anxiety mediated the link between sociotropy and anxiety within relationships with previous and current romantic partners. These effects were most pronounced in the romantic and friendship domains. In the familial domain, autonomy and sociotropy retained direct effects on relationship-specific avoidance and anxiety, and the indirect effects testing the mediating effect of domain-specific attachment orientations were only marginally significant (p < .10). This pattern raises the interesting possibility that in young adults, the global-dispositional components of the attachment network are more strongly associated with familial relations than they are with more recently formed attachment bonds with individual’s peers, such as attachment to romantic partners and platonic friends. This is consistent with Bowlby’s (1969/1982) and Collins and Read’s (1994) proposition that the attachment network is built on the back of repeated interactions with attachment figures in early childhood, most likely familial relationships. These familial attachment patterns, therefore, might have a stronger influence on the development of more general representations and modes of responding accounting for the consistent links between personality-level variables and relationship-specific representation even when domain-specific representations were controlled, at least in young adults, such as those sampled here. Nevertheless, in general, the domain-congruent indirect effects support the underlying functional structure of the attachment network. As argued previously, a hierarchically organized network containing both global and specific representations provides increased flexibility and functionality across different relational contexts. While attachment experiences across contexts should culminate into a more global (personality-esqué) interpersonal orientation, in order to serve their function, attachment representations must accurately guide how attachment needs and goals can be achieved. Thus, global representations should be most likely to govern
SPR346421
21/12/09
3:42 am
Page 19
Sibley & Overall: Modeling attachment representations
19
interpersonal behavior in new and unfamiliar contexts when more specific and precise representations (in terms of the accuracy of if . . . then . . . contingencies) are unavailable to guide relational responding. Within experienced contexts (i.e., with specific relationship partners and familiar domains), global representations should play a more subtle and distal role in guiding behaviors, and work through more specific representations that describe most fully regularities within that context. Finally, this study extended our previous work mapping the relations between attachment and personality by ruling out the Big Five as an alternative framework for understanding the most global aspect of the attachment representational network. The diffuse effects of autonomy and sociotropy, domain differentiated attachment effects, and domain-congruent indirect effects of autonomy and sociotropy mediated by domain-specific attachment orientations, all held when controlling for broad-bandwidth markers of the Big Five. Links between attachment and personality The distinction between individual differences in attachment orientation versus personality traits revolves around issues regarding the generality and boundary conditions of the attachment behavioral system. Are attachment dynamics limited solely to relationships with attachment figures? Or do the models governing individual differences in attachment system functioning act as a basis for the formation of more generalized interpersonal orientations that “spill over” into other interpersonal (non-attachment) domains? It is likely that for psychologically healthy individuals, the set goal of the attachment system is to seek proximity solely to attachment figures. Over time, however, the patterns learnt in these formative relationships may lead to more global behavioral tendencies that apply across different relational contexts. In their recent work outlining the psychodynamics of attachment-system functioning, Mikulincer and Shaver (2003, p. 65) emphasized this possibility when they stated that “what [begins] as representations of specific interactions with a specific partner become core personality characteristics and tend to be applied in new situations and relationships where they continue to guide the functioning of the attachment system.” Taken in concert with our prior research in this area (Sibley, 2007; Sibley & Overall, 2007, 2008a, 2008b), the current study indicates that autonomy and sociotropy reflect these core behavioral tendencies. However, while our findings support a differentiated network of attachment representations reflecting global tendencies and more attachmentspecific representations, there may exist individual differences in the extent to which global regularities (autonomy, sociotropy) and more specific attachment tendencies (avoidance, anxiety) are intertwined. For example, due to cognitive closure and reluctance to assimilate and/or accommodate new and inconsistent relationship information (e.g., Mikulincer, 1997; Mikulincer & Arad, 1999), individuals who are insecurely attached (particularly in their earlier relationships) might possess less complex attachment representation networks (i.e., lower differentiation across different types of relationships
SPR346421
21/12/09
20
3:42 am
Page 20
Journal of Social and Personal Relationships 27(1)
and relationship domains). For example, Collins and Allard (2001) reported that attributions by secure individuals varied as a function of relationship evaluations. In contrast, insecure individuals favored negative attributions, regardless of the positivity of their current relationships. These results suggest that insecurely attached individuals rely more on general representations, or have poorly elaborated and differentiated relationship-specific representations. A less differentiated network will have important implications for the operation of the attachment system. Following suggestions raised by Crittenden (1990), people who do not distinguish clearly between attachment needs and contingencies applicable to relations with platonic friends, family members, and romantic partners may be less discriminating in who they seek proximity with during times of attachment-system activation and/or are likely to respond in similar ways across contexts. This may mean that global level orientations will be more descriptive of interpersonal behavior within specific relationships and domains. Importantly, however, not encompassing key distinctions across relationship domains reduces the accuracy in the contingencies governing functioning within specific relationships and contexts; a level of accuracy that is needed in order to obtain attachment goals that vary across different domains and relationships. Moreover, as Bowlby (1988) argued, being guided by inaccurate, outdated or over-generalized representations will produce the type of inflexible responding that accounts for many relationship difficulties. Future research should test these speculations. Caveats and future research directions Recently, considerable progress has been made in mapping the hierarchical structure of personality, of which it is now thought the Big Five represent an intermediary, yet still broad or abstracted, level. For instance, Digman (1997) argued that the Big Five are subsumed under two more broad-bandwidth factors labeled as α and β (also referred to as stability and plasticity). Digman (1997) argued that the α factor contained common elements of Agreeableness, Conscientiousness, and low Neuroticism (Emotional Stability), which seem to relate to socialization processes. The β factor, in contrast, contained common elements of Extraversion and Openness, which seem to reflect personal growth. Taking a more fine-grained approach, De Young, Quilty and Peterson (2007) recently argued that each dimension of the Big Five can also be differentiated into two more specific and distinct aspects. Impressively, this view seems consistent with recent genetic studies which suggest that the dual aspects of each Big Five dimension may have distinct biological substrates (Jang, Livesley,Angleitner, Riemann, & Vernon, 2002). According to De Young et al. (2007), Neuroticism, for example, seems to separate into two correlated factors representing volatility and withdrawal, which seem similar to tendencies associated with anxiety or sociotropy and avoidance or autonomy respectively. Future research could extend our exploration of the attachment representational network by controlling for these more specific Big Five facets (particularly De Young et al.’s (2007) dual aspects of Neuroticism).
SPR346421
21/12/09
3:42 am
Page 21
Sibley & Overall: Modeling attachment representations
21
Future research could also extend our findings by testing the possible longitudinal top-down effects of personality, sociotropy, and autonomy on attachment orientations, and the corresponding bottom-up effects of specific attachments on more global regularities in behavior. In addition, an important direction for future research would be to examine the extent to which these bottom-down and top-up effects differ depending upon developmental period and related relationship contexts. Pierce and Lydon (2001) provided important initial data on these effects and reported, in a sample of young adults, that both relationship-specific and domain-specific attachments reciprocally influenced one another over a four-month period. Amongst young adults, then, these different levels of the attachment network appear to shape one another over relatively short time periods – perhaps because in this population the people are rapidly developing new (presumably romantic and friendship) relationship experiences. As originally formulated by Bowlby (1969/1982), repeated interactions with attachment figures in early childhood should shape the nature of social interactions with attachment figures in adolescence and adulthood. These carry-over effects occur because regularities associated with early experiences are summarized into representations that act as the basis for interpersonal behavior and expectations in other attachment relationships. Thus, over time, representations summarizing regularities in relational responding with specific attachment figures shape more global personality traits (Mikulincer & Shaver, 2003). Although experiences within new relationships might also produce gradual (bottom-up) normative changes in personality throughout adulthood, more global summaries of regularities in responding (i.e., personality) should be the default guide governing interpersonal behavior as the individual forms new attachment relationships (a top-down process). Our research focused primarily on young adults and used concurrent assessments of different levels of the attachment representational network. Thus, although the pattern of associations we observed speaks to how the network is cognitively structured within this population, it does not inform us on how the attachment network might develop and change over time. Future research is needed to examine the development of the different levels of the attachment representational network in greater detail. Finally, future research could extend our analyses by assessing a wider range of specific relationships within a broader number of relational domains. This would allow the evaluation of a three-level multi-level model partitioning the variation into three, rather than two components (as we did here), that map onto the three levels of the attachment representational network. Conclusions This study is the first to comprehensively map the structure and content of three levels of the attachment representational network across multiple domains. Extending prior research, our findings indicate that autonomy and sociotropy are uniquely associated with the manifestation of attachment avoidance and anxiety across a range of relationship domains and orientations within specific relationships. Measures of domain-specific romantic attachment, in contrast, constitute more specific regularities that describe
SPR346421
21/12/09
22
3:42 am
Page 22
Journal of Social and Personal Relationships 27(1)
responding within particular domains. Thus, the global personality-based component of the attachment network indirectly influences attachment toward particular persons via more accurate domain-level attachment representations. These results suggest that the attachment representational network is hierarchically structured, and that autonomy and sociotropy capture global regularities in relational responding, which underlie differences in the functioning of the attachment system.
REFERENCES Beck, A. T. (1983). Cognitive therapy of depression: New perspectives. In P. J. Clayton, & J. E. Barrett (Eds.), Treatment of depression: Old controversies and new approaches (pp. 265–290). New York: Raven Press. Besser, A., & Priel, B. (2003). A multisource approach to self-critical vulnerability to depression: The moderating role of attachment. Journal of Personality, 71, 515–555. Blatt, S. J., D’Afflitti, J. P., & Quinlan, D. M. (1976). Experiences of depression in normal young adults. Journal of Abnormal Psychology, 85, 383–389. Blatt, S. J., & Zuroff, D. C. (1992). Interpersonal relatedness and self-definition: Two prototypes for depression. Clinical Psychology Review, 12, 527–562. Bowlby, J. (1969/1982). Attachment and loss: Vol. 1. Attachment (2nd edn). New York: Basic Books. Bowlby, J. (1988). A secure base: Clinical applications of attachment theory. New York, NY: Basic Books. Campbell, L., Simpson, J. A., Boldry, J., & Kashy, D. A. (2005). Perceptions of conflict and support in romantic relationships: The role of attachment anxiety. Journal of Personality and Social Psychology, 88, 510–531. Collins, N. L. (1996). Working models of attachment: Implications for explanation, emotion, and behavior. Journal of Personality and Social Psychology, 71, 810–832. Collins, N. L., & Allard, L. A. (2001). Cognitive representations of attachment: The content and function of working models. In G. J. O. Fletcher, & M. Clark (Eds.), Blackwell handbook of social psychology: Interpersonal processes (pp. 60–85). London: Blackwell. Collins, N. L., & Read, S. J. (1994). Cognitive representations of attachment: The structure and function of working models. In K. Bartholomew, & D. Perlman (Eds.), Advances in personal relationships: Vol. 5. Attachment processes in adulthood (pp. 53–90). London: Jessica Kingsley. Cozzarelli, C., Hoekstra, S. J., & Bylsma, W. H. (2000). General versus specific mental models of attachment: Are they associated with different outcomes? Personality and Social Psychology Bulletin, 26, 605–618. Crittenden, P. M. (1990). Internal representational models of attachment relationships. Infant Mental Health Journal, 11, 259–277. De Young, C. G., Quilty, L. C., & Peterson, J. B. (2007). Between facets and domains: 10 aspects of the Big-Five. Journal of Personality and Social Psychology, 93, 880–896. Digman, J. M. (1997). Higher-order factors of the Big Five. Journal of Personality and Social Psychology, 73, 1246–1256. Donnellan, M. B., Burt, S. A., Levendosky, A. A., & Klump, K. L. (2008). Genes, personality, and attachment in adults: A multivariate behavioral genetic analysis. Personality and Social Psychology Bulletin, 34, 3–16. Dunkley, D. M., Blankstein, K. R., & Flett, G. L. (1997). Specific cognitive-personality vulnerability styles in depression and the five-factor model of personality. Personality and Individual Differences, 23, 1041–1053. Dunkley, D. M., Blankstein, K. R., Zuroff, D. C., Lecce, S., & Hui, D. (2006). Self-critical and personal standards factors of perfectionism located within the five-factor model of personality. Personality and Individual Differences, 40, 409–420.
SPR346421
21/12/09
3:42 am
Page 23
Sibley & Overall: Modeling attachment representations
23
Goldberg, L. R. (1999). A broad-bandwidth, public domain, personality inventory measuring the lower-level facets of several five-factor models. In I. Mervielde, I. Deary, F. De Fruyt, & F. Ostendorf (Eds.), Personality Psychology in Europe (Vol. 7, pp. 7–28). Tilburg, The Netherlands: Tilburg University Press. Jang, K. L., Livesley, W. J., Angleitner, A., Reimann, R., & Vernon, P. A. (2002). Genetic and environmental influences on the covariance of facets defining the domains of the five-factor model of personality. Personality and Individual Differences, 33, 83–101. Krull, J. L., & MacKinnon, D. P. (2001). Multilevel modeling of individual and group level mediated effects. Multivariate Behavioral Research, 36, 249–277. Lynch, T. R., Robins, C. J., & Morse, J. Q. (2001). Couple functioning in depression: The roles of autonomy and sociotropy. Journal of Clinical Psychology, 57, 93–103. Mikulincer, M. (1997). Adult attachment style and information processing: Individual differences in curiosity and cognitive closure. Journal of Personality and Social Psychology, 5, 1217–1230. Mikulincer, M., & Arad, D. (1999). Attachment working models and cognitive openness in close relationships: A test of chronic and temporary accessibility effects. Journal of Personality and Social Psychology, 77, 710–725. Mikulincer, M., & Shaver, P. R. (2003). The attachment behavioral system in adulthood: Activation, psychodynamics, and interpersonal processes. In M. P. Zanna, (Ed.), Advances in experimental social psychology (Vol. 35, pp. 53–152). New York: Academic Press. Mischel, W., & Shoda, Y. (1995). A cognitive-affective system theory of personality: Reconceptualizing situations, dispositions, dynamics, and invariance in personality structure. Psychological Review, 102, 246–268. Mischel, W., & Shoda, Y. (1999). Integrating dispositions and processing dynamics within a unified theory of personality: The cognitive-affective personality system. In L. A. Pervin, & O. P. John (Eds.), Handbook of personality: Theory and research. New York: Guilford Press. Mongrain, M., Vettese, L. C., Shuster, B., & Kendal, N. (1998). Perceptual biases, affect, and behavior in the relationships of dependents and self-critics. Journal of Personality and Social Psychology, 75, 230–241. Murphy, B., & Bates, G. W. (1997). Adult attachment style and vulnerability to depression. Personality and Individual Differences, 22, 835–844. Noftle, E. E., & Shaver, P. R. (2006). Attachment dimensions and the big five personality traits: Associations and comparative ability to predict relationship quality. Journal of Research in Personality, 40, 179–208. Overall, N. C., Fletcher, G. J. O., & Friesen, M. D. (2003). Mapping the intimate relationship mind: Comparisons between three models of attachment representations. Personality and Social Psychology Bulletin, 29, 1479–1493. Paunonen, S. V. (1998). Hierarchical organization of personality and prediction of behavior. Journal of Personality and Social Psychology, 74, 538–556. Pierce, T., & Lydon, J. E. (2001). Global and specific relational models in the experience of social interactions. Journal of Personality and Social Psychology, 80, 613–631. Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods (2nd edn). Thousand Oaks, California: Sage Publications. Robins, C. J., Ladd, J., Welkowitz, J., Blaney, P. H., Diaz, R., & Kutcher, G. (1994). The Personal Style Inventory: Preliminary validation studies of new measures of sociotropy and autonomy. Journal of Psychopathology and Behavioral Assessment, 16, 277–300. Ross, L. R., & Spinner, B. (2001). General and specific attachment representations in adulthood: Is there a relationship? Journal of Social and Personal Relationships, 18, 747–766. Rude, S. S., & Burnham, B. L. (1995). Connectedness and neediness: Factors of the DEQ and SAS dependency scales. Cognitive Therapy and Research, 19, 323–340. Shaver, P. R., & Brennan, K. A. (1992). Attachment styles and the “big five” personality traits: Their connections with each other and with romantic relationship outcomes. Personality and Social Psychology Bulletin, 18, 536–545. Shaver, P. R., & Mikulincer, M. (2002). Attachment-related psychodynamics. Attachment and Human Development, 4, 133–161.
SPR346421
21/12/09
24
3:42 am
Page 24
Journal of Social and Personal Relationships 27(1)
Sibley, C. G. (2007). The association between working models of attachment and personality: Toward an integrative framework operationalizing global relational models. Journal of Research in Personality, 41, 90–109. Sibley, C. G., & Liu, J. H. (2006). Working models of romantic attachment and the subjective quality of social interactions across relational contexts. Personal Relationships, 16, 243–259. Sibley, C. G., & Overall, N. C. (2007). The boundaries between attachment and personality: Associations across three levels of the attachment network. Journal of Research in Personality, 41, 960–967. Sibley, C. G., & Overall, N. C. (2008a). Modeling the hierarchical structure of attachment representations: A test of domain differentiation. Personality and Individual Differences, 44, 238–249. Sibley, C. G., & Overall, N. C. (2008b). The boundaries between attachment and personality: Localized versus generalized effects in daily social interaction. Journal of Research in Personality, 42, 1394–1407. Simpson, J. A., Rholes, W. S., Campbell, L., Tran, S., & Wilson, C. L. (2003). Adult attachment, the transition to parenthood, and depressive symptoms. Journal of Personality and Social Psychology, 84, 1172–1187. Simpson, J. A, Rholes, W. S., & Phillips, D. (1996). Conflict in close relationships: An attachment perspective. Journal of Personality and Social Psychology, 71, 899–914. Zuroff, D. C. (1994). Depressive personality styles and the Five-Factor model of personality. Journal of Personality Assessment, 63, 453–472. Zuroff, D. C., & Duncan, N. (1999). Self-criticism and conflict resolution in romantic couples. Canadian Journal of Behavioural Science, 31, 137–149. Zuroff, D. C., & Fitzpatrick, D. K. (1995). Depressive personality styles: Implications for adult attachment. Personality and Individual Differences, 18, 253–265.
