Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Dec 1.
Published in final edited form as: J Child Psychol Psychiatry. 2019 Jun 19;60(12):1309–1322. doi: 10.1111/jcpp.13086

Antecedents of attachment states of mind in normative-risk and high-risk caregiving: cross-race and cross-sex generalizability in two longitudinal studies

John D Haltigan 1, Glenn I Roisman 2, Ashley M Groh 3, Ashley S Holland 4, Cathryn Booth-LaForce 5, Fred A Rogosch 6, Dante Cicchetti 2,6
PMCID: PMC6856397  NIHMSID: NIHMS1032563  PMID: 31215651

Abstract

Background:

Longitudinal investigations of relatively large typical-risk (e.g., Booth-LaForce & Roisman, 2014) and higher-risk samples (e.g., Raby et al., 2017; Roisman et al., 2017) have produced evidence consistent with the claim that attachment states of mind in adolescence and young adulthood, as measured by the Adult Attachment Interview (AAI), are associated with the quality of caregiving experienced during childhood. None of these studies, however, has examined whether such associations are consistent across sex and/or race, as would be expected in light of the sensitivity hypothesis of attachment theory.

Methods:

We examine whether sex or race moderates previously reported links between caregiving and AAI states of mind in two longitudinal studies (pooled N = 1,058) in which caregiving was measured either within (i.e., observed [in]sensitive care) or outside (i.e., childhood maltreatment) of the normative range of caregiving experiences.

Results:

Hierarchical moderated regression analyses in both longitudinal cohorts provided evidence that maternal insensitivity and experiences of maltreatment were prospectively associated with dismissing and preoccupied states of mind in adolescence, as hypothesized. Moreover, these associations were generally comparable in magnitude for African American and White/non-Hispanic participants and were not conditional on participants’ biological sex.

Conclusions:

Both maternal insensitivity and the experience of maltreatment increased risk for insecure attachment states of mind in adolescence. Moreover, our analyses provided little evidence that either participant race or participant sex assigned at birth moderated these non-trivial associations between measures of the quality of experienced caregiving and insecure attachment states of mind in adolescence. These findings provide support for the sensitivity hypothesis of attachment theory and inform the cultural universality hypothesis of attachment processes.

Keywords: Adult Attachment Interview, attachment, attachment states of mind, cultural differences, maltreatment, sensitivity

Introduction

Across five decades, attachment theory (Bowlby, 1969, 1973, 1981) has endured as an organizing framework from which to understand human social, behavioral, and emotional functioning across the lifespan, with a wealth of empirical findings supporting its core hypotheses (Cassidy & Shaver, 2016; Sroufe, Egeland, Carlson, & Collins, 2005). This is especially true of its ethological foundations, which emphasize that infants possess an innate, evolutionary-shaped propensity to become attached to caregivers, regardless of their specific cultural niche or biological sex (Mesman, Van IJzendoorn, & Sagi-Schwartz, 2016; Van IJzendoorn & Sagi-Schwartz, 2008). Although Bowlby made such claims based in part on his synthesis of ethological accounts of animal behavior and Darwin’s theory of natural selection, it was Ainsworth’s pioneering observational studies of infant-caregiver attachment in rural Uganda and later Baltimore, Maryland that provided some of the first cross-cultural evidence in support of these universalist assumptions (Ainsworth, 1967; Ainsworth, Blehar, Waters, & Wall, 1978).

The primary measures used by developmental psychologists to assess attachment-related individual differences in infancy (e.g., the Strange Situation Procedure [SSP]; Ainsworth et al., 1978) and adolescence/adulthood (e.g., the Adult Attachment Interview [AAI]; George, Kaplan, & Main, 1984–1996; Main, Kaplan, & Cassidy, 1985) were developed and validated against core attachment constructs within the theory’s nomological network (Cronbach & Meehl, 1955), particularly those indexing the quality of early care experienced with primary caregivers. Indeed, perhaps the most well replicated and empirically supported associations are those between caregiver sensitivity and the development of infant attachment security (e.g., De Wolff & Van IJzendoorn, 1997; NICHD ECCRN, 1997). Quantitative meta-analytic (Van IJzendoorn, 1995) and more recent empirical work (Haltigan et al., 2014; Whipple, Bernier, & Mageau, 2011), have also documented non-trivial associations between AAI states of mind and the degree to which parents are observed to sensitively interact with their own children.

Longitudinal studies examining the degree to which (in)sensitive caregiving in childhood predicts AAI states of mind in adolescence and adulthood are in contrast limited, largely because such studies take decades and a great deal of human and financial capital to complete successfully. Nonetheless, a growing body of prospective, longitudinal work examining the antecedents of attachment states of mind has provided evidence for moderate associations between insensitive caregiving in childhood and the development of insecure attachment states of mind in late adolescence and young adulthood (e.g., Beckwith, Cohen, & Hamilton, 1999; Haydon, Collins, Salvatore, Simpson, & Roisman, 2012; Haydon, Roisman, Owen, Booth-LaForce, & Cox, 2014). Paralleling these findings about the antecedents of AAI states of mind in earlier parenting within the normative range, childhood abuse and neglect have also been shown to be associated with increased risk for insecure AAI states of mind, in some instances years later (Raby, Labella, Martin, Carlson, & Roisman, 2017; Roisman et al., 2017).

There has been some evidence that observations of insensitive caregiving are uniquely associated with dismissing AAI states of mind (Haydon et al., 2014) and that the experience of childhood maltreatment is uniquely associated with preoccupied states of mind (Raby et al., 2017; Roisman et al., 2017). Such findings, while by no means unequivocal and in need of further replication, are generally consonant with both prior empirical studies (e.g., Beckwith et al., 1999; Beijersbergen, Juffer, Bakermans-Kranenburg, & Van IJzendoorn, 2012) and more recent theorizing about relations between childhood caregiving experiences and the development of AAI states of mind (e.g., Haydon & Roisman, 2014; Roisman et al., 2017; Whipple et al., 2011). More specifically, the basic premise of such work is that, insensitive care within the normative range represents a rejection of attachment-related needs, which should predispose individuals to defensively exclude (via dismissing strategies of idealization and normalization) unloving behavior from awareness. In contrast, trauma and threatening experiences, such as maltreatment (i.e., physical and sexual abuse), heighten attachment-related distress, thus conferring a vulnerability for preoccupied states of mind. Of note, whether parental insensitivity outside of the normative range (i.e., extreme parental insensitivity), which may reflect a boundary area between insensitive care and frank maltreatment, also confers a vulnerability for preoccupied states of mind is an open question.

Despite the early returns on this work demonstrating relations between childhood caregiving and later AAI states of mind, direct examinations of the cross-race and cross-sex generalizability of these associations have not yet been undertaken. Indeed, notwithstanding previous work in attachment research claiming similar rates of both infant and adult attachment security across cultures and/or sex (e.g., Bakermans-Kranenburg & Van IJzendoorn, 2009; Van IJzendoorn & Bakermans-Kranenburg, 2010; Van IJzendoorn & Kroonenberg, 1988), little research has directly examined the cross-group generalizability of developmental associations between adult attachment states of mind and other constructs within the nomological network of attachment theory, particularly caregiver sensitivity given its privileged status as the core antecedent of attachment security (but see Bakermans-Kranenburg, Van IJzendoorn, & Kroonenberg, 2004 and Mesman, Van IJzendoorn, & Bakermans-Kranenburg, 2012 for relevant work with respect to infant attachment security).

Only two studies to our knowledge have explicitly tested the cross-group generalizability of the predictive significance of AAI states of mind. In the first, Roisman and colleagues (2004) examined associations between AAI states of mind and autonomic nervous system responses during the AAI. Replicating seminal work by Dozier and Kobak (1992), dismissing states of mind were associated with increases in skin conductance levels during the AAI above resting levels. Notable in the current context, the associations reported by Roisman et al. (2004) were not moderated by participant sex and were roughly the same magnitude for Chinese/Chinese American and White participants. In a second study, Haltigan et al. (2014) similarly demonstrated that associations between observed maternal insensitivity and dismissing states of mind were comparable for African American and White/non-Hispanic mothers, consistent with the sensitivity hypothesis of attachment theory (Van IJzendoorn & Sagi, 1999; Van IJzendoorn & Sagi-Schwartz, 2008). Importantly, however, neither of these prior studies focused on antecedents of adult states of mind and whether such associations vary as a function of sex or race.

This paucity of work is attributable in part to two factors. First, given the predominant interest in the intergenerational transmission of attachment patterns, the vast majority of research on the AAI has focused on primary caregivers, who are overwhelmingly female. Second, most studies of the AAI have been conducted using Western samples of mostly White/non-Hispanic participants, thus precluding direct tests of cross-race or cross-ethnic generalizability. Indeed, the disproportionate use of Western samples in studies examining attachment theory has led some scholars (e.g., Keller, 2018; Rothbaum, Weisz, Pott, Miyake, & Morelli, 2000) to suggest that such work might not be applicable to cultures with differing value systems and that attachment processes may be culturally specific. Nonetheless, tests of the generalizability of associations between adult attachment states of mind and attachment constructs indexing the quality of childhood caregiving (e.g., maternal sensitivity) are important as they directly inform the universalist assumption of the sensitivity hypothesis in attachment theory (Van IJzendoorn & Sagi, 1999; Van IJzendoorn & Sagi-Schwartz, 2008), which claims that attachment security will be found linked to maternal sensitivity.

The universalist assumption concerning developmental relations between maternal sensitivity and attachment security is typically tested with two rival hypotheses. According to the ‘no group difference’ hypothesis (Bakermans-Kranenburg, 2004; Van IJzendoorn et al., 2006) few if any racial (or sex) differences in developmental processes (as evidenced by patterns of covariation among relevant variables, such as maternal sensitivity and measures of attachment security) will be found and racial- (or sex-) specific experiences do not alter the associations among developmental variables. This hypothesis allows for cultural and/or sex differences in variables, but assumes that the correlations among variables do not differ between groups (Van IJzendoorn et al., 2006). On the other hand, the alternative ‘group differences hypothesis’ is predicated upon culturally relative or group specific models of socialization (e.g., Garcia Coll, 1990; Ogbu, 1981, 1993) and implies that the correlations among relevant variables are different between groups, implying a lack of universality in developmental processes, such as sensitivity-security associations, that are at the heart of attachment theory.

The current study

Extending prior empirical work in this area (Bakermans-Kranenburg et al., 2004; Haltigan et al., 2014; Mesman et al., 2012; Roisman et al., 2004; Van IJzendoorn & Bakermans-Kranenburg, 2010), in the current report we investigated the cross-race and cross-sex generalizability of the antecedents of AAI states of mind in two relatively large sample studies: the normative-risk Study of Early Child Care and Youth Development cohort (SECCYD; Haydon et al., 2014) and a high-risk sample studied in Rochester, New York (The Rochester Adolescent Study; Cicchetti, Rogosch, & Sturge-Apple, 2007; Rogosch, Oshri, & Cicchetti, 2010; Roisman et al., 2017). Following earlier reports based on these datasets—and based on results of taxometric and factor analyses of AAI state of mind scales (see Booth-LaForce & Roisman, 2014)—in the current investigation we focused on the antecedents of two key, relatively independent AAI states of mind dimensions: (a) the degree to which adults freely evaluate or defensively discuss their attachment histories (i.e., dismissing states of mind) and (b) the extent to which adults become emotionally overwhelmed while discussing their early attachment experiences (i.e., preoccupied states of mind). Given that most previous work examining the generalizability of attachment findings has focused on African Americans and European Americans within the United States (e.g., Bakermans-Kranenburg et al., 2004; Haltigan et al., 2014), and because the data available to us arguably included sufficient numbers of these two racial groups, we focused on this group contrast. We similarly examined whether such associations between the quality of caregiving during childhood and adolescence and AAI states of mind were moderated by the biological sex of the participants.

Methods

Participants

SECCYD.

The National Institute of Child Health and Human Development (NICHD) Study of Early Child Care and Youth Development (SECCYD) was a long-term longitudinal study that followed 1,364 study children and their families at 10 sites across the US, from birth through age 15 years. Families were recruited for the SECCYD in 1991 from hospitals located in or near Little Rock, AR; Orange County, CA; Lawrence, KS; Boston, MA; Pittsburgh, PA; Philadelphia, PA; Charlottesville, VA; Seattle, WA; Morganton, NC; and Madison, WI. During selected 24-hour sampling periods, 8,986 women who gave birth were screened, 5,416 of whom met the eligibility criteria for the study. Families were excluded if: (a) the mother was younger than 18 years of age, (b) the family planned to move, (c) there was a multiple birth, (d) the infant had a known disability or remained in the hospital more than 7 days, (e) the mother acknowledged substance abuse, (f) the mother did not speak English, (g) the mother lived more than an hour from the laboratory site or in an extremely unsafe neighborhood, as determined by local police. Additional details about recruitment and selection procedures are available in prior publications from the study (see ECCRN, 2005) and from the study web site (https://www.nichd.nih.gov/research/supported/Pages/seccyd.aspx). Data for the current study were drawn from a subset (n = 857) of SECCYD participants (M age = 17.8 years, range 17.2 to 19.1) who were assessed at age 18 on the AAI as part of a follow-up study led by researchers at the University of Washington (see Booth-LaForce & Roisman, 2014).

Detailed information on the demographic characteristics of the AAI subsample of the SECCYD can be found in Booth-LaForce and Roisman (2014). Note that, while large, demographically diverse, and methodologically rich, the SECCYD was not designed to be a nationally representative study. Reflecting the normative-risk (relative to low or high-risk) nature of the AAI subsample, the sub-sample of participants administered the AAI in the SECCYD had an average T-score of almost exactly 50 (i.e., the population mean) on the Youth Self-Report (YSR; Achenbach, 1991) form of the Achenbach Child Behavior Checklist administered as part of the same age 18 year assessment of the SECCYD (M = 52.2, SD = 11, N = 701).

From the subset of age 18 AAI participants described above, n = 760 (n = 90, African American/non-Hispanic; n = 670 White/non-Hispanic) formed the analytic sample for this report given our objective of examining the cross-race generalizability of associations between quality of care during childhood and adolescence and AAI states of mind for African Americans and Whites as described above. We compared the present AAI analytic subsample to the full SECCYD cohort (n = 1364) on relevant covariates (i.e., participant race, sex, maternal education, and family income). There were disproportionately more White/non-Hispanic (vs. non-white; AAI: M = 88.2%; no AAI: M = 61.6%), χ2 (1, N = 1364) = 131.73, p < .01), and more females (AAI: M = 51.6%; no AAI: M = 44.2%), χ2 (1, N = 1364) = 7.33, p < .01, in the current analytic subsample relative to excluded participants from the full SECCYD cohort. Additionally, maternal education (AAI: M = 14.7, SD = 2.40; no AAI: M = 13.7, SD = 2.55), t (1361) = 7.42, p < .001) and family income levels (AAI: M = 4.21, SD = 3.14; no AAI: M = 3.20, SD = 2.80), t (1354) = 6.17, p < .001) were both significantly higher in the AAI analytic subsample than in excluded participants from the larger SECCYD cohort.

Rochester Cohort.

Participants were adolescents from low-income families, assessed as part of a larger investigation of adolescent development among youth with and without histories of child maltreatment. The sample was recruited by a Department of Human Services (DHS) project liaison, who identified families with children in the targeted age range who had experienced child maltreatment as documented in DHS records. A random sample of families was approached by the DHS liaison about possible participation in the study. If interested, the parent signed a name release form in order to provide contact information to project staff. Staff subsequently met with the parent to explain the study and obtain informed consent. Comparison adolescents from non-maltreated families were recruited from families receiving Temporary Assistance to Needy Families (TANF). The DHS liaison identified such families through DHS records and screened for the absence of maltreatment. The same procedures for maltreating families were used for recruitment of non-maltreating families.

In addition to detailed DHS records on all maltreatment experiences obtained for independent coding, project staff also interviewed mothers regarding child maltreatment experiences (see below). Information from these interviews was incorporated into maltreatment determinations. Among non-maltreating families, cases involving maternal report of child maltreatment experiences were excluded from further participation. Data for the current study were drawn from the full cohort described above (N = 354; M age = 16.7 years, SD = 1.3) which was comprised of 214 maltreated and 140 non-maltreated adolescents. Of these participants, n = 298 (n = 195, African American/non-Hispanic; n = 103 White/non-Hispanic) formed the analytic subsample for this report given study aims of examining cross-race generalizability of associations between quality of maltreatment experienced during childhood and adolescence and attachment states of mind between African Americans and Whites. We compared the current analytic subsample to the full Rochester cohort (n = 354) on relevant covariates (i.e., participant race and sex). As with the SECCYD data, there were (in this case by design) disproportionately more White/non-Hispanic participants in the analytic subsample, (analytic sample: M = 34.6%; excluded participants: 0%), χ2 (1, N = 354) = 27.30, p < .01. The Rochester analytic subsample (males = 54.4%, females = 45.6%) did not significantly differ from excluded participants (males = 64.3%, females 35.7%) from the full cohort with respect to participant sex, χ2 (1, N = 354) = 1.88, p = .17.

Measures

Adult Attachment Interview (SECCYD and Rochester cohorts).

The AAI (George, Kaplan, & Main, 1984–1996) is a semi-structured, hour-long interview in which participants are asked to describe memories related to their early childhood relationships with their primary caregivers and the effects they perceive those experiences to have had on their development. AAI coding focuses on the internal consistency of participants’ narratives as well as the extent to which participants become emotionally activated while discussing childhood memories. In the Rochester cohort, a modification of the AAI script was used (Ward & Carlson, 1995) involving minor word changes that simplified some language and clarified the nature of requests from interviewers for information.

The AAI Q-Sort (Kobak, 1993; see also Kobak & Zajac, 2011; Kobak, Zajac, & Smith, 2009) was used to code adolescents’ AAI narratives in both the SECCYD and Rochester cohorts by raters centrally trained and certified as reliable on the Main, Goldwyn, & Hesse (2003–2008) AAI categorical coding system. The AAI Q-Sort contains 100 cards pertaining to attachment-related states of mind and inferred parental experience as assessed by the AAI. Cards are sorted across nine piles from most to least characteristic of the participant’s narrative in a forced, quasi-normal distribution. The distribution of participants’ cards was then correlated with prototypic sorts based on conceptually relevant attachment dimensions. Informed by both large-sample taxometric and factor-analytic evidence (Fraley & Roisman, 2014; Haltigan, Roisman, & Haydon, 2014; Haydon, Roisman, & Burt, 2012; Roisman, Fraley, & Belsky, 2007) that the primary coding systems for the AAI capture variation relevant to attachment states of mind on relatively independent dismissing and preoccupied dimensions, and paralleling prior work with the SECCYD AAI subsample evaluating the predictive significance of early caregiving for attachment states of mind (Haydon et al., 2014), we focused on the dismissing (e.g., “adjectives [describing early relationships with parents] supported by vague or shallow memories vs. adjectives supported by detailed episodic memories”) and preoccupied (e.g., “is confused or overwhelmed with information about parents vs. information about parents is adequate and well organized”) Q-sort prototypes (Kobak, 1993).

In the SECCYD cohort, reliability was calculated on approximately 21% of AAI transcripts. Cases for which reliability was .6 or higher (after Spearman-Brown correction) was 90%. Per convention, a third (and on rare occasions, fourth) coder completed another sort for unreliable cases and data from the two coders with the highest reliability above .6 were averaged and used in analyses (final M = .77, SD = .08, .60-.93). In the Rochester cohort, approximately 20% of the AAI transcripts were double sorted. Of the subset of double-coded cases, coders were reliable (≥0.6 using the Pearson–Brown prophecy formula) on 88% of cases. Where two coders were discrepant, a third coder sorted the case; the final sort was computed by averaging the two sorts (reliable ≥ 0.6) that were most highly correlated. Note that coding was completed without knowledge of maltreatment status, except as revealed within the context of the AAI itself. Further details on AAI reliability and psychometrics for the SECCYD and Rochester cohorts can be found in Booth-LaForce & Roisman (2014) and Roisman et al. (2017), respectively.

Maternal sensitivity (SECCYD Cohort).

Maternal sensitivity was assessed in the context of mother-child interactions that were videotaped during 15-min semi-structured situations at 6, 15, 24, 36 and 54 months; Grades 1, 3, and 5; and age 15. At each assessment point, at the younger ages children engaged in tasks at the zone of proximal development while primary caregivers provided assistance; at older ages (Grade 3 and older), joint tasks, including discussion tasks, were used. Tasks were designed to be developmentally appropriate. Psychometric properties for composite measures of observed maternal sensitivity at each assessment point were adequate (internal consistencies of the sensitivity composite measures for mothers averaged .79 [range .70 - .85] across assessments; for detailed information on the SECCYD sensitivity assessments see Belsky, Vandell, et al., 2007; Haltigan, Roisman, & Fraley, 2013; NICHD ECCRN, 2001, 2004, 2008). Maternal sensitivity scores were standardized and averaged to create composite measures of observed maternal sensitivity.1

Maltreatment (Rochester cohort).

The Maltreatment Classification System (MCS; Barnett, Manly, & Cicchetti, 1993) is a reliable and valid method for classifying childhood abuse and neglect (Bolger, Patterson, & Kupersmidt, 1998; English et al., 2005; Manly, 2005) that uses DHS records detailing investigations and findings involving maltreatment in identified families over time. Based on operational criteria, the MCS designates all of the subtypes of maltreatment children have experienced (i.e., neglect, emotional maltreatment, physical abuse, sexual abuse; for more detail see Roisman et al., 2017). Coding of the DHS records was conducted by trained research assistants, doctoral students, and clinical psychologists. Coders were required to meet acceptable reliability with criterion standards before coding for the study. Coders demonstrated acceptable reliability with the criterion (weighted κs with the criterion ranging from .86 to .98). κs for the presence vs. absence of maltreatment subtypes ranged from .90 to 1.00.

Adolescents in the maltreatment group all had documented histories of experiencing abuse and/or neglect according to DHS records. Among these adolescents, 82.7% had experienced neglect, 60.3% had experienced emotional maltreatment, 51.9% had experienced physical abuse, and 21.5% had experienced sexual abuse. As is typical in maltreated populations (Bolger et al., 1998; Manly et al., 1994; 2001), the majority of adolescents had experienced multiple subtypes of maltreatment. Specifically, 61.6% of the maltreated adolescents had experienced two or more maltreatment subtypes.

Results

Analytic plan

To test whether hypothesized associations between childhood and adolescent relationship experiences and AAI states of mind were moderated by participant sex or race, we conducted a series of hierarchical (moderated) regressions for each AAI Q-sort state of mind dimension (i.e., dismissing and preoccupied) in which conceptually related sets of covariates and focal interactions were entered in successive blocks. In both the SECCYD and Rochester cohorts, we coded participant sex such that 0 = male, 1 = female and participant race such that 0 = African American, 1 = White/Caucasian. To maintain consistency with prior empirical reports based on the SECCYD (Haydon, Roisman, et al., 2014), maternal education and family income were also included as covariates in the SECCYD analyses. More specifically, maternal years of education was operationalized as the number of years of education as reported by the mother at the baseline assessment of the study. Family financial resources were operationalized in terms of an income-to-needs ratio, computed separately for every assessment point in the SECCYD at which relevant data were acquired (1, 6, 15, 24, 36, 54 months; Grades 1, 3, 4, 5, 6; age 15). Scores were averaged across assessment points to create a composite measure of family financial resources (for additional detail, see Booth-LaForce & Roisman, 2014; see Table 1 for descriptive statistics and pairwise correlations among SECCYD cohort analytic variables). In contrast, we did not include indicators of SES as covariates in the analyses of the Rochester cohort because the entire sample was low income by design and previous analyses have established no main effects of such indicators on AAI states of mind in this cohort (for additional detail, see Roisman et al., 2017; Descriptive statistics and pairwise correlations among Rochester cohort analytic variables appear in Table 4). Of note, between-samples comparisons on AAI dismissing and preoccupied states of mind, as well as more detail concerning the demographic characteristics of both samples as they relate to between-samples differences on dismissing and preoccupied states of mind, have been previously reported (see Roisman et al., 2017). In short, compared to the normative-risk SECCYD, low-income status alone (equivalent to an r = .21, a medium difference) but especially maltreatment plus low-income status (r = .39, a large difference) increased risk of dismissing states of mind in the Rochester cohort. However, rates of preoccupied states of mind in the Rochester cohort were only higher than in the SECCYD among maltreated participants, and this difference was small in magnitude (r = .16).

Table 1.

Descriptive Statistics and Pairwise Bivariate Correlations (SECCYD Cohort).

1 2 3 4 5 6 7
1. Dismissing --
2. Preoccupied .09* --
3. M Sensitivity −.34** −.22** --
4. Participant Sex −.24** .12** .07 --
5. Participant Race −.16** −.19** .41** .01 --
6. M education −.25** −.13** .51** .06 .23** --
7. Family Income −.18** −.15** .42** .04 .25** .55** --
Mean −.03 −.24 .06 51.6% 88.2% 14.68 4.21
SD .40 .23 1.00 -- -- 2.40 3.14
Open in a new tab

Note. N = 760 for all correlations except those including family income (N = 759). M = Mother; F = Father. Fam. income, composite of family income-to-needs. Descriptive statistics for participant sex and race are expressed as proportions of female and White/Caucasian participants, respectively.

*

p < .05

**

p < .01

Table 4.

Descriptive Statistics and Pairwise Bivariate Correlations (Rochester Cohort).

1 2 3 4 5
1. Dismissing --
2. Preoccupied .01 --
3. Maltreatment .23** .27** --
4. Participant Sex −.21** .15* −.00 --
5. Participant Race −.06 .04 .12* −.04 --
Mean .24 −.22 .59 45.6% 34.6%
SD .35 .18 .49 -- --
Open in a new tab

Note. N = 298 for all correlations. Descriptive statistics for participant sex and race are expressed as proportions of female and White/Caucasian participants, respectively. The mean for maltreatment reflects the proportion of participants who were judged as maltreated.

*

p < .05

**

p < .01

For moderator models in each cohort, the first block tested the predictive significance of the focal quality of caregiving antecedent construct (i.e., maternal sensitivity in the SECCYD and maltreatment status in the Rochester sample) for AAI states of mind in adolescence. The second block added adolescent race or sex and its interaction with the core caregiving antecedent to examine the interaction of interest. The third block controlled for key cohort-specific covariates, as well as their interaction with the focal moderator, in order to examine the robustness of caregiving variables in predicting AAI states of mind. For both the SECCYD and Rochester models, either sex or race was entered as a covariate in this block depending on which focal moderator model was being tested (i.e., if the race moderator model was being examined, sex was entered as a main effects covariate in this block). As noted previously, for the SECCYD cohort only, maternal education, and family income were the other key covariates entered in this block. The final block of the models for each cohort tested whether any observed effects from previous blocks were robust to the inclusion of the other AAI state of mind dimension and allowed us to test for unique effects of maternal sensitivity and maltreatment on dismissing and preoccupied attachment states of mind, respectively.

SECCYD

Participant race moderator models.

Our first model tested the predictive significance of maternal sensitivity and its interaction with participant race on the dismissing state of mind dimension. As previously reported in Haydon et al. (2014) we found a main effect of sensitivity on dismissing states of mind in Block 1 of this moderated regression model, a main effect that was robust to the inclusion of additional covariates, including the preoccupied state of mind dimension, in the 4th and final block of this model. Participant race did not moderate associations between maternal sensitivity and dismissing states of mind (Whites: r = −0.29, p < .01; African Americans: r = −0.39, p < .01; see Table 1 for all model coefficients).

Our second model repeated this approach for the preoccupied state of mind dimension. Once again, we found a main effect of sensitivity on preoccupied states of mind in Block 1 of this moderated regression model. That said, and unlike the model for the dismissing state of mind dimension, this main effect of sensitivity was no longer significant when controlling for key covariates. Additionally, we found evidence that participant race moderated associations between maternal sensitivity and preoccupied attachment states of mind (Whites: r = −. 19, p < .01; African Americans: r = .01, p = .96). Of note, the maternal sensitivity × participant race interaction no longer remained significant in the 4th Block of this model when controlling for participant sex, maternal education, family income, the maternal education × race and family income × race interactions, and the dismissing state of mind dimension (see Table 1 for all model coefficients).

Participant sex moderator models.

Separate models for dismissing and preoccupied state of mind dimensions were conducted in identical fashion to participant race moderator models. As would be expected based on analyses presented above, we observed main effects of maternal sensitivity on dismissing states of mind in our first model. Similarly, the main effect of maternal sensitivity on dismissing states of mind was robust to the inclusion of additional covariates and the preoccupied state of mind dimension in the 4th Block of this model. There was no evidence that participant sex moderated associations between maternal sensitivity and the dismissing state of mind dimension (males: r = −. 28, p < .01; females: r = −.36, p < .01; see Table 2 for all model coefficients).

Table 2.

Hierarchical Regression Models Predicting AAI Q-Sort Dimensions: Participant Race Models (SECCYD Cohort).

Dismissing Preoccupied
β p R2 β p R2
1. M sensitivity −.34 .00 .12 −.22 .00 .05
2. M sensitivity −.34 .00 .12 .01 .95 .06
 Participant race −.02 .65 −.17 .00
 M sensitivity × participant race .01 .93 −.18 .03
3. M sensitivity −.32 .00 .17 −.01 .89 .09
 Participant race .03 .57 −.22 .00
 M sensitivity × participant race .07 .44 −.15 .08
 Participant sex −.22 .00 .14 .00
 M education .03 .87 .15 .37
 Family income −.29 .24 .09 .72
 M education × Participant race −.12 .42 −.13 .41
 Family income × Participant race .28 .24 −.15 .54
4. M sensitivity −.32 .00 .18 .01 .96 .09
 Participant race .04 .45 −.22 .00
 M sensitivity × participant race .07 .39 −.16 .08
 Participant sex −.23 .00 .15 .00
 M education .02 .91 .03 .52
 Family income −.30 .23 .11 .68
 M education × Participant race −.12 .45 −.13 .43
 Family income × Participant race .29 .23 −.17 .50
 Dismissing — — .06 .14
 Preoccupied .05 .14 — —
Open in a new tab

Note. N = 759. M = Maternal. Participant sex coded as 0 = boys, 1 = girls. Participant race coded as 0 = African American, 1 = White/Caucasian.

In our second model examining the preoccupied state of mind dimension, we again found a main effect of maternal sensitivity on preoccupied states of mind. Additionally, and in contrast to the participant race moderator models, the main effect of maternal sensitivity on preoccupied states of mind remained significant when controlling for our additional set of covariates (participant race, income-to-needs, and maternal education), and the dismissing state of mind dimension in the 4th Block of this model. Paralleling results for the dismissing state of mind dimension, we did not find evidence that participant sex moderated associations between maternal sensitivity and the preoccupied state of mind dimension (males: r = −. 27, p < .01; females: r = −.16, p < .01; see Table 2 for all model coefficients).

Rochester cohort

Participant race moderator models.

Separate moderated regression models for the dismissing and preoccupied dimensions were conducted in parallel fashion to those above for the SECCYD cohort. As previously reported in Roisman et al. (2017), there was a main effect of maltreatment status on dismissing states of mind in the first block of the hierarchical moderated regression model predicting the dismissing state of mind dimension. Also as previously reported (Roisman et al., 2017), maltreatment status remained significantly positively associated with dismissing states of mind in the final block of this model controlling for the main effect of participant sex and the preoccupied state of mind dimension. However, there was no evidence that participant race moderated associations between maltreatment status and the dismissing state of mind dimension (Whites: r = .22, p < .05; African Americans: r = .25, p < .01; see Table 3 for all model coefficients).

Table 3.

Hierarchical Regression Models Predicting AAI Q-Sort Dimensions: Participant Sex Models (SECCYD Cohort).

Dismissing Preoccupied
β p R2 β p R2
1. M sensitivity −.34 .00 .12 −.22 .00 .05
2. M sensitivity −.28 .00 .17 −.26 .00 .07
 Participant sex −.22 .00
.13 .00
 M sensitivity × participant sex −.06 .26 .05 .32
3. M sensitivity −.23 .00 .17 −.22 .00 .08
 Participant sex −.22 .00 .13 .00
 M sensitivity × participant sex −.04 .47 .10 .12
 Participant race −.02 .51 −.11 .01
 M education −.13 .04 −.01 .93
 Family income .07 .28 .01 .94
 M education × Participant sex .04 .52 .02 .80
 Family income × Participant sex −.09 .14 −.10 .10
4. M sensitivity −.22 .00 .18 −.21 .00 .09
 Participant sex −.22 .00 .14 .00
 Sensitivity × participant sex −.05 .42 .10 .11
 Participant race −.02 .60 −.11 .01
 M education −.13 .04 .01 .99
 Family income .07 .28 .00 ..98
 M education × Participant sex .04 .53 .01 .82
 Family income × Participant sex −.08 .16 −.10 .11
 Dismissing — — .05 .19
 Preoccupied .05 .19 — —
Open in a new tab

Note N = 759. M = Maternal. Participant sex coded as 0 = boys, 1 = girls. Participant race coded as 0 = African American, 1 = White/Caucasian.

Results for the moderated regression model predicting the preoccupied state of mind dimension were similar to those obtained for the dismissing state of mind dimension. More specifically, and as previously reported (Roisman et al., 2017) there was a main effect of maltreatment status on preoccupied states of mind which remained significantly positively associated with the preoccupied state of mind dimension even when controlling for the main effect of participant sex and dismissing states of mind in the final block of this model. Once again, there was no evidence that participant race moderated associations between maltreatment status and the preoccupied state of mind dimension (Whites: r = .36, p < .01; African Americans: r = .22, p < .01; see Table 3 for all model coefficients).

Participant sex moderator models.

We next investigated whether associations between maltreatment and dismissing and preoccupied states of mind were moderated by participant sex (see Table 4). Once again, there was no evidence that participant sex moderated associations between maltreatment status and dismissing states of mind (males: r = .16, p < .05; females: r = .24, p < .01) or preoccupied states of mind (males: r = .26, p < .01; females: r = .21, p < .01). Similar to participant race models, maltreatment remained significantly positively associated with both dismissing and preoccupied states of mind in the final block of these models (see Table 4 for model coefficients).

Discussion

The present study capitalized on two well-characterized cohorts containing both African Americans and Whites and males and females to examine the cross-race and cross-sex generalizability of associations between the quality of childhood and adolescent caregiving experiences and AAI states of mind. Consistent with previously reported results based on these cohorts (Haydon et al., 2014; Roisman et. al., 2017), the current findings demonstrated that both maternal insensitivity and the experience of maltreatment (i.e., abuse and neglect) increase risk for insecure attachment states of mind. Novel to the current report, our analyses provided little evidence that either participant race (African American v. White) or participant sex assigned at birth moderated these non-trivial associations between measures of the quality of experienced caregiving and insecure attachment states of mind, and underscore the consistency of the significance and magnitude of associations between quality of care and attachment states of mind in these groups. Below we discuss these results in the context of the cultural universality hypothesis of attachment theory and the limited corpus of work to date that has directly informed this hypothesis (e.g., Bakermans-Kranenburg et al., 2004; Behrens, Hesse, & Main, 2007; Haltigan et al., 2014; Kazui, Endo, Tanaka, Sakagami, & Suganuma, 2000; Kondo-Ikemura, Behrens, Umemura, & Nakano, 2018; Roisman et al., 2004; Van IJzendoorn et al., 2010).

Tests of cross-race and cross-sex generalizability of associations among core constructs within the nomological network of attachment theory are important as they inform the cultural universality hypothesis of attachment and the cross-cultural validity of attachment measures such as the AAI. Collectively, results of our moderated regression analyses converge with prior work demonstrating that relations between attachment states of mind and theoretically relevant physiological correlates (Roisman et al., 2004) and parenting outcomes (e.g., maternal sensitivity; Haltigan et al., 2014) were not moderated by race, consistent with the culture-general socialization claims implicit in attachment theory. Additionally, these results are consistent with those found in infant attachment work (Bakermans-Kranenburg et al., 2004), demonstrating associations between maternal sensitivity and infant attachment security in African-American and White children were of comparable magnitude.

Despite this general pattern of findings concerning the cross-race and cross-sex generalizability of associations between the quality of caregiving and attachment states of mind, we did find modest evidence that participant race moderated associations between maternal sensitivity and preoccupied attachment states of mind in the SECCYD. More specifically, we found only one nominally (p < .05) significant interaction effect and evaluation of simple effects revealed that the association between maternal sensitivity and preoccupied attachment states of mind was significant only for White/non-Hispanic participants (there was essentially no association between these constructs for African Americans). If shown to replicate, this finding warrants some attention as it presents a challenge to the cultural universality hypothesis of attachment. That said, this finding was limited to the SECCYD cohort in the current report and has not been observed in other work (Haltigan et al., 2014), at least with respect to the association between mother’s AAI states of mind and their observed sensitivity toward their own children. It is also possible that this finding might reflect sampling error given the number of moderated regression analyses conducted (i.e., Table 2 reports p values uncorrected for the total number of models conducted, and the nominally significant results disappeared after controlling for interactions between sample-specific covariates [maternal education and family income] and their interaction with participant race and sex).

Despite the general lack of evidence of moderation of associations between quality of childhood and adolescent caregiving and attachment states of mind by participant race or sex, we did find consistent evidence across both cohorts that males were more likely than females to have dismissing states of mind. In contrast, females in both cohorts were more likely than males to have preoccupied states of mind. Although a previous meta-analysis of AAI classifications did not find evidence of sex differences in the use of dismissing versus preoccupied strategies (Bakermans-Kranenburg & Van IJzendoorn, 2009), the sex differences found in this investigation merit further scrutiny as we and others (Adam et al., 1996; Allen et al., 1996; Dozier, 1990; Raby et al., 2017; Riggs & Jacobvitz, 2002; Roisman et al., 2004; Rosenstein & Horowitz, 1996) have consistently observed these patterns in both clinical and nonclinical samples, particularly when dismissing and preoccupied AAI states of mind are studied as weakly correlated dimensions. To the extent they continue to be replicated in future work, these sex differences warrant explanation.

One existing explanation for mean-level sex differences in insecure attachment is based in an evolutionary, life-history theory account of interpersonal development (Belsky, Steinberg, & Draper, 1991; Del Giudice & Belsky, 2010; Del Giudice, 2009; 2018; Euler, 2019). Life-history theory emphasizes an individual’s life-history strategy, which is thought to be driven by the limited amount of resources available to all organisms, and the way in which they will allocate their resources in response to their environments to maximize their (genetic) fitness (Del Giudice, 2018). In brief, sex differences in insecure attachment are thought to emerge as a result of distinctive sex-linked behavioral strategies for promoting genetic fitness via pair-bonding and reproduction. Because males potentially gain larger reproductive benefits from casual relationships (i.e., less emotional investment, avoidance of intimacy) and less parenting investment, whereas females benefit from continued (or increased) emotional investment and more parenting investment (i.e., dependency, exaggeration of need signaling), it has been hypothesized that avoidant (e.g., dismissing) attachment orientations will be biased towards males and anxious/ambivalent attachment orientations (e.g., preoccupied) towards females under conditions of moderate ecological risk (for more detail see Del Giudice, 2009; 2019; see also David & Lyons-Ruth, 2005 for infant attachment).

Finally, although less consistent than the sex differences reported above, we also observed some evidence for elevated levels of preoccupied states of mind among African Americans relative to Whites in the SECCYD cohort. These findings converge with those of Haltigan et al. (2014), who found mean-level cultural differences in preoccupied states of mind between African American and White mothers. Given these findings, it is important to point out that the ‘no group differences’ hypothesis does not preclude direct effects of culture-specific influences on the particular manner in which attachment insecurity is manifested depending on the particular ethnic or ecological niche in which individuals develop (Van IJzendoorn & Sagi-Schwartz, 2008). For example, Haltigan et al. (2014) interpreted their findings as possibly reflective of a legacy of sociocultural-specific influences (African cultural heritage values emotional expression; White & Parham, 1990) and potentially related to proximal contextual adversities, such as economic hardship and poverty, disproportionately experienced by African Americans (Spencer et al., 2006). Further empirical work exploring potential elevations in preoccupied states of mind among African Americans (and other minority racial groups) relative to White/non-Hispanics may provide valuable insights into both the intergenerational transmission of attachment and parenting across generations in African Americans (e.g., Wakschlag et al., 1996), as well as models of risk and resilience to contextual adversity (Spencer et al., 2015).

Although there are a number of strengths to the present investigation, including the use of two large, racially non-homogenous, prospective study cohorts on which a key measure within the nomological network of attachment theory was acquired, is not without its limitations. Differences in acquired constructs and their operationalizations across both studies limited our ability to take a more thorough investigation of potential covariates in our tests of the cross-race and cross-sex generalizability of the caregiving antecedents of adult attachment states of mind. Similarly, between-group sample sizes, particularly for our racial categorizations, were not symmetrical. This was especially true for the SECCYD sample where a limited sample of less than 100 African American participants were available. In addition, our study drew on samples from the United States and therefore could not examine whether the findings generalize to non-Western countries that have a different cultural heritage.

Regarding the measurement of attachment states of mind, the Q-sort approach to scoring the AAI does not directly assess unresolved trauma and loss, and thus we are unable to address questions related to the cross-groups generalizability of relations between attachment constructs in the context of lack of resolution of trauma or loss (e.g., in particular the Rochester maltreatment sample). Related to this limitation, because parental sensitivity was not assessed in the Rochester cohort, we were not able to test our hypothesis regarding threatening, trauma-inducing experiences conferring vulnerability for preoccupied states of mind in the most comprehensive manner possible as we could not assay whether extreme parental insensitivity also predicted preoccupied states of mind. That said, all published evidence to date has suggested that AAI state of mind scales indexing preoccupied and unresolved discourse load on a common factor (i.e., preoccupation) that captures variance in—at least to some degree—trauma-inducing experiences, of which maltreatment of the kinds measured here would certainly qualify as reflecting. Future use of the Q-sort procedure would benefit from using ratings on the AAI unresolved loss (U/Loss) and unresolved trauma (U/Trauma) rating scales (Main, Goldwyn, & Hesse; 2003–2008) as an adjunct to the Q-sort (see Zajac & Kobak, 2009).

This study also does not consider genetic influences on the association between caregiving and attachment states of mind in adolescence. Although behavior genetic studies have suggested little genetic influence on infant attachment security (e.g., Bokhorst et al., 2003; O’Connor & Croft, 2001; Roisman & Fraley, 2008), it is possible that genetic variation may influence the association between the quality of caregiving and attachment security in different ways across development, particularly during adolescence when genetic variation may become increasingly influential in relation to parental attitudes and caregiving behavior (i.e., gene-environment correlation; Fearon, Shmueli-Goetz, Viding, Fonagy, & Plomin, 2014; Scarr & McCartney, 1983) and possibly the capacity to reflect upon and integrate early experiences (Main, 1996).

Conclusion

This relatively large sample, dual cohort study is the first to investigate the cross-race and cross-sex generalizability of the caregiving antecedents of adult attachment states of mind. The results of this study provided evidence consistent with the universal claims implicit in the sensitivity hypothesis of attachment theory (Van IJzendoorn & Sagi-Schwartz, 2008). One of the enduring strengths of Bowlby’s attachment theory is that it is an ethologically-grounded theory that posits an innate bias of infants to become attached, regardless of their specific cultural niche or biological sex (Van IJzendoorn & Sagi-Schwartz, 2008). Additional prospective work examining whether race and/or ethnicity may moderate associations between sensitivity and attachment states of mind, thus implying group differences in hypothesized developmental processes central to attachment theory (Bakermans-Kranenburg et al., 2004; Van IJzendoorn et al., 2006), as well as mean-level differences in attachment states of mind across race and sex, will advance our understanding of the limits of the cultural universality hypothesis of attachment. Equally important, it will contribute to our understanding of human attachment processes and psychological development in an increasingly diverse Western world (Mesman et al., in press).

Table 5.

Hierarchical Regression Models Predicting AAI Q-Sort Dimensions: Participant Race Models (Rochester Cohort).

Dismissing Preoccupied
β p R2 β p R2
1. Ever maltreated
 .23 .00 .05 .27 .00 .07
2. Ever maltreated .24 .00 .06 .21 .00 .08
 Participant race −.08 .38 −.11 .22
 Ever maltreated × participant race −.00 .97 .17 .11
3. Ever maltreated .25 .00 .10 .21 .00 .10
 Participant race −.09 .34 −.11 .23
 Ever maltreated × participant race −.01 .92 .17 .10
 Participant sex −.21 .00 .15 .01
4. Ever maltreated .25 .00 .10 .21 .00 .10
 Participant race −.10 .33 −.11 .23
 Ever maltreated × participant race −.01 .95 .17 .10
 Participant sex −.21 .00 .15 .01
 Dismissing — — −.02 .76
 Preoccupied −.02 .76 — —
Open in a new tab

Note. N = 298. Participant sex coded as 0 = boys, 1 = girls. Participant race coded as 0 = White/Caucasian, 1 =African American.

Table 6.

Hierarchical Regression Models Predicting AAI Q-Sort Dimensions: Participant Sex Models (Rochester Cohort).

Dismissing Preoccupied
β p R2 β p R2
1. Ever maltreated
 .23 .00 .05 .27 .00 .07
2. Ever maltreated .26 .00 .10 .25 .00 .10
 Participant sex −.17 .05 .13 .14
 Ever maltreated × participant sex −.05 .62 .03 .75
3. Ever maltreated .27 .00 .11 .25 .00 .10
 Participant sex −.17 .05 .13 .14
 Ever maltreated × participant sex −.05 .60 .03 .75
 Participant race −.10 .09 .01 .85
4. Ever maltreated .27 .00 .11 .26 .00 .10
 Participant sex −.17 .05 .13 .15
 Ever maltreated × participant sex −.05 .60 .03 .75
 Participant race −.10 .09 .01 .87
 Dismissing — — −.02 .76
 Preoccupied −.02 .76 — —
Open in a new tab

Note. N = 298. Participant sex coded as 0 = boys, 1 = girls. Participant race coded as 0 = White/Caucasian, 1 =African American.

Key points.

  • Recent longitudinal investigations have provided evidence that attachment states of mind in adolescence and young adulthood are associated with the quality of caregiving experienced during childhood. No prior studies, however, have examined whether such associations are consistent across race and/or sex, as would be expected in light of the universality hypothesis of attachment theory.

  • In the current report, we examine whether participant race or sex moderates previously reported links between the quality of childhood and adolescent caregiving and AAI states of mind in two longitudinal cohorts in which caregiving was measured either within (i.e., observed [in]sensitive care) or outside (i.e., childhood maltreatment) of the normative range.

  • Results indicated that associations between indicators of caregiving and attachment states of mind were largely not conditional on participants’ biological sex, and were comparable in magnitude for African American and White/non-Hispanic participants in both cohorts, lending support to the cultural universality of attachment theory’s sensitivity-security hypothesis.

  • Results from both cohorts also indicated main effect sex differences in dismissing and preoccupied attachment states of mind that are consistent with previously. reported findings and warrant future investigation.

Acknowledgements

Research reported in this publication was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development of the National Institutes of Health under Award Numbers R01 HD054822 to C.B. and R01 HD069442 to G.I.R. Support for the Rochester adolescent study was received from the National Institute on Drug Abuse under Award Number R01 DA12903 to D.C. and F.A.R. The contents of this manuscript are the sole responsibility of the authors and do not necessarily reflect the views of the Eunice Kennedy Shriver National Institute for Child Health and Human Development or the National Institute on Drug Abuse. The authors are grateful to the Study of Early Child Care and Youth Development and Rochester participants and families for their time. The authors have declared that they have no competing or potential conflicts of interest.

Footnotes

Conflict of interest statement: No conflicts declared.

1

Results were substantively similar when using a sensitivity composite based only on maternal sensitivity ratings acquired from 6-month through the 54-month assessment.

References

  1. Achenbach TM (1991). Manual for the Youth Self-Report and 1991 profile. Burlington, VT: University of Vermont, Department of Psychiatry. [Google Scholar]
  2. Adam KS, Sheldon-Keller AE, & West M (1996). Attachment organization and history of suicidal behavior in clinical adolescents. Journal of Consulting and Clinical Psychology, 64, 264–272. doi: 10.1037/0022-006X.64.2.264 [DOI] [PubMed] [Google Scholar]
  3. Ainsworth MDS (1967). Infancy in Uganda: Infant care and the growth of love. Baltimore, MD: Johns Hopkins University Press. [Google Scholar]
  4. Ainsworth MDS, Blehar MC, Waters E, & Wall S (1978). Patterns of attachment: A psychological study of the strange situation. New York, NY: Erlbaum. [Google Scholar]
  5. Allen JP, Hauser ST, & Borman‐Spurrell E (1996). Attachment theory as a framework for understanding sequelae of severe adolescent psychopathology: An 11‐year follow‐up study. Journal of Consulting & Clinical Psychology, 64, 254–263. [DOI] [PubMed] [Google Scholar]
  6. Bakermans-Kranenburg MJ, & Van IJzendoorn MH (2009). The first 10,000 Adult Attachment Interviews: Distributions of adult attachment representations in clinical and non-clinical groups. Attachment & Human Development, 11, 223–263. doi: 10.1080/14616730902814762 [DOI] [PubMed] [Google Scholar]
  7. Bakermans-Kranenburg MJ, Van IJzendoorn MH, & Kroonenberg PM (2004). Differences in attachment security between African-American and white children: Ethnicity or socio-economic status? Infant Behavior and Development, 27, 417–433. doi: 10.1016/j.infbeh.2004.02.002 [DOI] [Google Scholar]
  8. Barnett D, Manly JT, & Cicchetti D (1993). Defining child maltreatment: The interface between policy and research In Cicchetti D & Toth SL (Eds.), Child abuse, child development, and social policy (pp. 7–74). Norwood, NJ: Ablex. [Google Scholar]
  9. Beckwith L, Cohen SE, & Hamilton CE (1999). Maternal sensitivity during infancy and subsequent life events relate to attachment representation at early adulthood. Developmental Psychology, 35, 693. doi: 10.1037/0012-1649.35.3.693 [DOI] [PubMed] [Google Scholar]
  10. Behrens KY, Hesse E, & Main M (2007). Mothers’ attachment status as determined by the Adult Attachment Interview predicts their 6-year-olds’ reunion responses: A study conducted in Japan. Developmental Psychology, 43, 1553–1567. doi: 10.1037/0012-1649.43.6.1553 [DOI] [PubMed] [Google Scholar]
  11. Beijersbergen MD, Juffer F, Bakermans-Kranenburg MJ, & Van IJzendoorn MH (2012). Remaining or becoming secure: Parental sensitive support predicts attachment continuity from infancy to adolescence in a longitudinal adoption study. Developmental Psychology, 48, 1277–1282. doi: 10.1037/a0027442 [DOI] [PubMed] [Google Scholar]
  12. Belsky J, Steinberg L, & Draper P (1991). Childhood experience, interpersonal development, and reproductive strategy: An evolutionary theory of socialization. Child Development, 62, 647–670. doi: 10.1111/j.1467-8624.1991.tb01558.x [DOI] [PubMed] [Google Scholar]
  13. Belsky J, Vandell DL, Burchinal M, Clarke‐Stewart KA, McCartney K, Owen MT, & NICHD Early Child Care Research Network. (2007). Are there long-term effects of early child care? Child Development, 78, 681–701. doi: 10.1111/j.1467-8624.2007.01021.x [DOI] [PubMed] [Google Scholar]
  14. Bernier A, Larose S, Boivin M, & Soucy N (2004). Attachment state of mind: Implications for adjustment to college. Journal of Adolescent Research, 19, 783–806. doi: 10.1177/0743558403260096 [DOI] [Google Scholar]
  15. Bolger KE, Patterson CJ, & Kupersmidt JB (1998). Peer relationships and self‐esteem among children who have been maltreated. Child Development, 69, 1171–1197. doi: 10.1111/j.1467-8624.1998.tb06166.x [DOI] [PubMed] [Google Scholar]
  16. Bokhorst CL, Bakermans-Kranenburg MJ, Fearon RMP, Van IJzendoorn MH, Fonagy P, & Schuengel C (2003). The importance of shared environment in mother-infant attachment security: A behavioral genetic study. Child Development, 74, 1769–1782. doi: 10.1046/j.1467-8624.2003.00637.x [DOI] [PubMed] [Google Scholar]
  17. Booth-LaForce C, & Roisman GI (Eds.). (2014). The Adult Attachment Interview: Psychometrics, stability and change from infancy, and developmental origins. Monographs of the Society for Research in Child Development, 79(3, Serial No. 314), 1–185. doi: 10.1111/mono.12110 [DOI] [Google Scholar]
  18. Bowlby J (1969). Attachment and loss: Vol. 1 Attachment. New York: Basic Books. [Google Scholar]
  19. Bowlby J (1973). Attachment and loss: Vol. 2 Separation: Anxiety and anger. New York: Basic Books. [Google Scholar]
  20. Bowlby J (1981). Attachment and loss: Vol. 3 Loss, sadness and depression. New York: Basic Books. [Google Scholar]
  21. Cassidy J, & Shaver PR (Eds.). (2016). Handbook of attachment: Theory, research, and clinical applications. 2nd Edition New York: Guilford Press. [Google Scholar]
  22. Cicchetti D, Rogosch FA, & Sturge-Apple ML (2007). Interactions of child maltreatment and 5-HTT and MAOA polymorphisms: Depressive symptomatology among adolescents from low socioeconomic status backgrounds. Development and Psychopathology, 19, 1161–118. doi: 10.1017/S0954579407000600 [DOI] [PubMed] [Google Scholar]
  23. Cronbach LJ, & Meehl PE (1955). Construct validity in psychological tests. Psychological Bulletin, 52, 281. [DOI] [PubMed] [Google Scholar]
  24. David DH, & Lyons‐Ruth K (2005). Differential attachment responses of male and female infants to frightening maternal behavior: Tend or befriend versus fight or flight? Infant Mental Health Journal: Official Publication of The World Association for Infant Mental Health, 26, 1–18. doi: 10.1002/imhj.20033 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Del Giudice M (2009). Sex, attachment, and the development of reproductive strategies. Behavioral and Brain Sciences, 32, 1–21. doi: 10.1017/S0140525X09000016 [DOI] [PubMed] [Google Scholar]
  26. Del Giudice M (2018). Evolutionary psychopathology: A unified approach. New York: Oxford University Press. [Google Scholar]
  27. Del Giudice M (2019). Sex differences in attachment styles. Current Opinion in Psychology, 25, 1–5. doi: 10.1016/j.copsyc.2018.02.004 [DOI] [PubMed] [Google Scholar]
  28. Del Giudice M, & Belsky J (2010). Sex differences in attachment emerge in middle childhood: An evolutionary hypothesis. Child Development Perspectives, 4, 97–105. doi: 10.1111/j.1750-8606.2010.00125.x [DOI] [Google Scholar]
  29. De Wolff MS, & Van IJzendoorn MH (1997). Sensitivity and attachment: A meta‐analysis on parental antecedents of infant attachment. Child Development, 68, 571–591. doi: 10.1111/j.1467-8624.1997.tb04218.x [DOI] [PubMed] [Google Scholar]
  30. Dozier M (1990). Attachment organization and treatment use for adults with serious psychopathological disorders. Development and Psychopathology, 2, 47–60. [Google Scholar]
  31. Dozier M, & Kobak RR (1992). Psychophysiology in attachment interviews: Converging evidence for deactivating strategies. Child Development, 63, 1473–1480. doi: 10.1111/j.1467-8624.1992.tb01708.x [DOI] [PubMed] [Google Scholar]
  32. Egeland B, & Sroufe LA (1981). Attachment and early maltreatment. Child Development, 44–52. doi: 10.2307/1129213 [DOI] [PubMed] [Google Scholar]
  33. English DJ, Upadhyaya MP, Litrownik AJ, Marshall JM, Runyan DK, Graham JC, & Dubowitz H (2005). Maltreatment’s wake: The relationship of maltreatment dimensions to child outcomes. Child Abuse & Neglect, 29, 597–619. doi: 10.1016/j.chiabu.2004.12.008 [DOI] [PubMed] [Google Scholar]
  34. Euler HA (2019). Gender differences in attachment from an evolutionary perspective: What is good for the goose may not be good for the gander. Attachment & Human Development. doi: 10.1080/14616734.2019.1589055 [DOI] [PubMed] [Google Scholar]
  35. Fearon P, Shmueli-Goetz Y, Viding E, Fonagy P, & Plomin R (2013). Genetic and environmental influences on adolescent attachment. The Journal of Child Psychology and Psychiatry, 55, 1033–1041. doi: 10.1111/jcpp.12171 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Fraley RC, & Roisman GI (2014). Categories or dimensions? A taxometric analysis of the Adult Attachment Interview. Monographs of the Society for Research in Child Development, 79, 36–50. doi: 10.1111/mono.12112 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Garcia Coll CT (1990). Developmental outcome of minority infants: A process-oriented look into our beginnings. Child Development, 61, 270–289. doi: 10.1111/j.1467-8624.1990.tb02779.x [DOI] [PubMed] [Google Scholar]
  38. George C, Kaplan N, & Main M (1984–1996). Adult attachment interview protocol. Unpublished manuscript, University of California, Berkeley. [Google Scholar]
  39. Haltigan JD, Leerkes EM, Wong MS, Fortuna K, Roisman GI, Supple AJ, … & Plamondon A (2014). Adult attachment states of mind: Measurement invariance across ethnicity and associations with maternal sensitivity. Child Development, 85, 1019–1035. doi: 10.1111/cdev.12180 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Haltigan JD, Roisman GI, & Fraley RC (2013). The predictive significance of early caregiving experiences for symptoms of psychopathology through midadolescence: Enduring or transient effects? Development and Psychopathology, 25, 209–221. doi: 10.1017/S0954579412000260 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Haltigan JD, Roisman GI, & Haydon KC (2014). The latent structure of the adult attachment interview: Exploratory and confirmatory evidence. Monographs of the Society for Research in Child Development, 79, 15–35. doi: 10.1111/cdev.12180 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Haydon KC, Collins WA, Salvatore JE, Simpson JA, & Roisman GI (2012). Shared and distinctive origins and correlates of adult attachment representations: The developmental organization of romantic functioning. Child Development, 83, 1689–1702. doi: 10.1111/j.1467-8624.2012.01801.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Haydon KC, Roisman GI, & Burt KB (2012). In search of security: The latent structure of the Adult Attachment Interview revisited. Development and Psychopathology, 24, 589–606. doi: 10.1017/S0954579412000181 [DOI] [PubMed] [Google Scholar]
  44. Haydon KC, Roisman GI, Owen MT, Booth-LaForce C, & Cox MJ (2014). Shared and distinctive antecedents of Adult Attachment Interview state-of-mind and inferred-experience dimensions. Monographs of the Society for Research in Child Development, 79, 108–125. doi: 10.11611/mono.1211 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Hinde RA, & Stevenson-Hinde J (1990). Attachment: Biological, cultural and individual desiderata. Human Development, 33, 62–72. doi: 10.1159/000276503 [DOI] [Google Scholar]
  46. Jackson JF (1993). Multiple caregiving among African Americans and infant attachment: The need for an emic approach. Human Development, 36, 87–102. doi: 10.1159/000277299 [DOI] [Google Scholar]
  47. Kazui M, Endo T, Tanaka A, Sakagami H, & Suganuma M (2000). Intergenerational transmission of attachment: Japanese mother–child dyads. Japanese Journal of Educational Psychology. doi: 10.5926/jjep1953.48.3_323 [DOI] [Google Scholar]
  48. Keller H (2018). Universality claim of attachment theory: Children’s socioemotional development across cultures. Proceedings of the National Academy of Sciences, 115, 11414–11419. doi: 10.1073/pnas.1720325115 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Kobak R (1993). The adult attachment interview Q-set. Unpublished document, University of Delaware. [Google Scholar]
  50. Kobak R & Zajac K (2011). Revaluating Adolescent States of Mind and Their Implications for Psychopathology: A Relational/Lifespan Framework In Cicchetti D & Roisman G (Eds.) The Minnesota Symposium on Child Psychology, The Origins of Adaptation and Maladaptation (Vol. 36, pp. 185–229). New York: Erlbaum. doi: 10.1002/9781118036600.ch6 [DOI] [Google Scholar]
  51. Kobak R, Zajac K, & Smith C (2009). Adolescent attachment and trajectories of hostile–impulsive behavior: Implications for the development of personality disorders. Development and Psychopathology, 21, 839–851. doi: 10.1017/S0954579409000455 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Kondo-Ikemura K, Behrens KY, Umemura T, & Nakano S (2018). Japanese mothers’ prebirth Adult Attachment Interview predicts their infants’ response to the Strange Situation Procedure: The strange situation in Japan revisited three decades later. Developmental Psychology, 54, 2007. doi: 10.1037/dev0000577 [DOI] [PubMed] [Google Scholar]
  53. Main M (1996). Introduction to the special section on attachment and psychopathology: 2. Overview of the field of attachment. Journal of Consulting and Clinical Psychology, 64, 237–243. doi: 10.1037/0022-006X.64.2.237 [DOI] [PubMed] [Google Scholar]
  54. Main M, Goldwyn R, & Hesse E (2003–2008). Adult attachment scoring and classification system. Unpublished manuscript, University of California at Berkeley. [Google Scholar]
  55. Main M, Hesse E, & Goldwyn R (2008). Studying differences in language usage in recounting attachment history: An introduction to the AAI.
  56. Main M, Kaplan N, & Cassidy J (1985). Security in infancy, childhood, and adulthood: A move to the level of representation. In Bretherton I & Waters E (Eds.), Growing points in attachment theory and research. Monographs of the Society for Research in Child Development, 50(Serial No. 209), 66–106. doi: 10.2307/3333827 [DOI] [Google Scholar]
  57. Manly JT (2005). Advances in research definitions of child maltreatment. Child Abuse & Neglect, 29, 425–439. doi: 10.1016/j.chiabu.2005.04.001 [DOI] [PubMed] [Google Scholar]
  58. Manly JT, Cicchetti D, & Barnett D (1994). The impact of subtype, frequency, chronicity, and severity of child maltreatment on social competence and behavior problems. Development and Psychopathology, 6, 121–143. doi: 10.1017/S0954579400005915 [DOI] [Google Scholar]
  59. Manly JT, Kim JE, Rogosch FA, & Cicchetti D (2001). Dimensions of child maltreatment and children’s adjustment: Contributions of developmental timing and subtype. Development and Psychopathology, 13, 759–782. doi: [PubMed] [Google Scholar]
  60. Mesman J, Minter T, Angnged A, Cissé IAH, Salali GD, Migliano AB (in press). Universality without uniformity: A culturally inclusive approach to sensitive responsiveness in infant caregiving. Child Development. doi: 10.1111/cdev.12795 [DOI] [PubMed] [Google Scholar]
  61. Mesman J, Van IJzendoorn MH, & Bakermans-Kranenburg MJ (2012). Unequal in opportunity, equal in process: Parental sensitivity promotes positive child development in ethnic minority families. Child Development Perspectives, 6, 239–250. doi: 10.1111/j.1750-8606.2011.00223.x [DOI] [Google Scholar]
  62. Mesman J, Van IJzendoorn MH, & Sagi-Schwartz A (2016). Cross-cultural patterns of attachment: Universal and contextual dimensions In Cassidy J & Shaver PR (Eds.), Handbook of attachment: Theory, research, and clinical application (3rd ed., pp. 852–877). New York, NY: Guilford Press. [Google Scholar]
  63. NICHD Early Child Care Research Network. (1997). The effects of infant child care on infant‐mother attachment security: Results of the NICHD Study of Early Child Care. Child Development, 68, 860–879. doi: 10.1111/j.1467-8624.1997.tb01967.x [DOI] [PubMed] [Google Scholar]
  64. NICHD Early Child Care Research Network. (2001). Nonmaternal care and family factors in early development: An overview of the NICHD Study of Early Child Care. Journal of Applied Developmental Psychology, 22, 457–492. [Google Scholar]
  65. NICHD Early Child Care Research Network. (2004). Multiple pathways to early academic achievement. Harvard Educational Review, 74, 1–29. [Google Scholar]
  66. O’Connor TG, & Croft CM (2001). A twin study of attachment in preschool children. Child Development, 72, 1501–1511. doi: 10.1111/1467-8624.00362 [DOI] [PubMed] [Google Scholar]
  67. Ogbu JU (1981). Origin of human competence: A cultural-ecological perspective. Child Development, 52, 413–429. doi: : https://www.jstor.org/stable/1129158 [Google Scholar]
  68. Ogbu JU (1993). Differences in cultural frame of reference. International Journal of Behavioral Development, 16, 483–506. doi: 10.1177%2F016502549301600307 [Google Scholar]
  69. Raby KL, Labella MH, Martin J, Carlson EA, & Roisman GI (2017). Childhood abuse and neglect and insecure attachment states of mind in adulthood: Prospective, longitudinal evidence from a high-risk sample. Development and Psychopathology, 29, 347–363. doi: 10.1017/S0954579417000037 [DOI] [PMC free article] [PubMed] [Google Scholar]
  70. Riggs SA, & Jacobvitz D (2002). Expectant parents’ representations of early attachment relationships: Associations with mental health and family history. Journal of Consulting and Clinical Psychology, 70, 195. doi: 0.1037//0022-006X.70.1.195 [PubMed] [Google Scholar]
  71. Rogosch FA, Oshri A, & Cicchetti D (2010). From child maltreatment to adolescent cannabis abuse and dependence: A developmental cascade model. Development and Psychopathology, 22, 883–897. doi: 10.1017/S0954579410000520 [DOI] [PMC free article] [PubMed] [Google Scholar]
  72. Roisman GI, & Cicchetti D (2017). Attachment in the context of atypical caregiving. Development and Psychopathology, 29, 331–335. doi: 10.1017/S0954579417000013 [DOI] [PubMed] [Google Scholar]
  73. Roisman GI, Fraley RC, & Belsky J (2007). A taxometric study of the Adult Attachment Interview. Developmental Psychology, 43, 675–686. doi: 10.1037/0012-1649.43.3.675 [DOI] [PubMed] [Google Scholar]
  74. Roisman GI, & Fraley RC (2008). A behavior-genetic study of parenting quality, infant attachment security, and their covariation in a nationally representative sample. Developmental Psychology, 44, 831–839. doi: 10.1111/j.l467-8624.2011.01 [DOI] [PubMed] [Google Scholar]
  75. Roisman GI, Rogosch FA, Cicchetti D, Groh AM, Haltigan JD, Haydon KC, Holland AS, & Steele RD (2017). Attachment states of mind and inferred childhood experiences in maltreated and comparison adolescents from low-income families. Development and Psychopathology, 29, 337–345. doi: 10.1017/S0954579417000025 [DOI] [PubMed] [Google Scholar]
  76. Roisman GI, Tsai JL, & Chiang K-HS (2004). The emotional integration of childhood experience: Physiological, facial expressive, and self-reported emotional response during the Adult Attachment Interview. Developmental Psychology, 40, 776–789. doi: 10.1037/0012-1649.40.5.776 [DOI] [PubMed] [Google Scholar]
  77. Rosenstein DS, & Horowitz HA (1996). Adolescent attachment and psychopathology. Journal of Consulting and Clinical Psychology, 64, 244–253 [DOI] [PubMed] [Google Scholar]
  78. Rothbaum F, Weisz J, Pott M, Miyake K, & Morelli G (2000). Attachment and culture: Security in the United States and Japan. American Psychologist, 55, 1093–1104. doi: 10.1037/0003-066X.55.10.1093 [DOI] [PubMed] [Google Scholar]
  79. Scarr S, & McCartney K (1983). How people make their own environments: A theory of genotype-environment effects. Child Development, 54, 424–435. doi: 10.2307/1129703 [DOI] [PubMed] [Google Scholar]
  80. Sroufe LA, Egeland B, Carlson E, & Collins WA (2005). The development of the person: The Minnesota Study of Risk and Adaptation From Birth to Adulthood. New York: Guilford Press. [Google Scholar]
  81. Spencer MB, Harpalani V, Cassidy E, Jacobs CY, Donde S, Goss TN, … & Wilson S (2006). Understanding vulnerability and resilience from a normative developmental perspective: Implications for racially and ethnically diverse youth In Cicchetti D & Cohen DJ (Eds.), Developmental Psychopathology: Volume One: Theory and Method (pp. 627–672). Hoboken, NJ: Wiley [Google Scholar]
  82. Steele RD, Waters TE, Bost KK, Vaughn BE, Truitt W, Waters HS, … & Roisman GI (2014). Caregiving antecedents of secure base script knowledge: A comparative analysis of young adult attachment representations. Developmental Psychology, 50, 2526. doi: 10.1080/14616734.2014.883636 [DOI] [PMC free article] [PubMed] [Google Scholar]
  83. Thompson RA (2008). Measure twice, cut once: Attachment theory and the NICHD Study of Early Child Care and Youth Development. Attachment & Human Development, 10, 287–297. doi: 10.1080/14616730802113604 [DOI] [PubMed] [Google Scholar]
  84. Van IJzendoorn MH (1995). Adult attachment representations, parental responsiveness, and infant attachment: a meta-analysis on the predictive validity of the Adult Attachment Interview. Psychological Bulletin, 117, 387–403. doi: 10.1037/0033-2909.117.3.387 [DOI] [PubMed] [Google Scholar]
  85. Van IJzendoorn MH, & Bakermans-Kranenburg MJ (2010). Invariance of adult attachment across gender, age, culture, and socioeconomic status? Journal of Social and Personal Relationships, 27, 200–208. doi: 10.1177/0265407509360908 [DOI] [Google Scholar]
  86. Van IJzendoorn MH, Bakermans-Kranenburg MJ, & Sagi-Schwartz A (2006). Attachment across diverse sociocultural contexts: The limits of universality In Rubin KH & Chung OB (Eds.) Parenting beliefs, behaviors, and parent-child relations: A cross-cultural perspective (pp. 107–42). New York: Psychology Press. [Google Scholar]
  87. Van IJzendoorn MH, & Kroonenberg PM (1988). Cross-cultural patterns of attachment: A meta-analysis of the strange situation. Child development, 59, 147–156. doi: 10.2307/1130396 [DOI] [Google Scholar]
  88. Van IJzendoorn MH, & Sagi-Schwartz A (2008). Cross-cultural patterns of attachment: Universal and contextual dimensions In Cassidy J & Shaver PR (Eds.), Handbook of attachment: Theory, research, and clinical application (2nd ed., pp. 880–905). New York, NY: Guilford Press. [Google Scholar]
  89. Wakschlag LS, Chase-Lansdale PL, & Brooks-Gunn J (1996). Not just “Ghosts in the Nursery”: Contemporaneous intergenerational relationships and parenting in young African-American families. Child Development, 67, 2131–2147. doi: 10.1111/j.1467-8624.1996.tb01848.x [DOI] [PubMed] [Google Scholar]
  90. Ward MJ, & Carlson EA (1995). Associations among adult attachment representations, maternal sensitivity, and infant‐mother attachment in a sample of adolescent mothers. Child Development, 66, 69–79. [DOI] [PubMed] [Google Scholar]
  91. Whipple N, Bernier A, & Mageau GA (2011). A dimensional approach to maternal attachment state of mind: Relations to maternal sensitivity and maternal autonomy-support. Developmental Psychology, 47, 396–403. doi: 10.1037/a0021310 [DOI] [PubMed] [Google Scholar]
  92. White JL, & Parham TA (1990). The psychology of Blacks: An African American perspective (2nd ed.). Englewood Cliffs, NJ: Prentice Hall. [Google Scholar]
  93. Zajac K, & Kobak R (2009). Caregiver unresolved loss and abuse and child behavior problems: intergenerational effects in a high-risk sample. Development and Psychopathology, 21, 173–187. doi: 10.1017/S095457940900011X [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES