The Parent-Child Relationship: A Potential Source of Resilience for . models of the self and of the other were combined by Bartholomew . RSQ-M) and Adolescent Relationship Scales Questionnaire Dad (A-RSQ-D). relationships, the child's attachment system and the parent's .. McEwan (); RQ = Relationship Questionnaire (Bartholomew & Horowitz. Items 19 - 36 quality, it can be seen that parent-child relationships can affect the child's . In a study conducted by Bartholomew and Horowitz (), four categories .. I~ the Attachment Style Questionnaire, the Adult Separation Anxiety.
Originally, the theory was aimed at explaining child and adult psychopathology in terms of nonoptimal relationships between children and their caregivers, or "at- tachment figures. Theoretically, these representations influence a person's expectations, emotions, defenses, and relational behavior in all close relationships.
Although the theory does not assume or require that internal working models persist without ,:: In one line of research, Main and her colleagues focused on the possibility that adult "states of mind with respect to attachment" i.
Members of Main's researchgroup interviewed parents about their child- hood family relationships and then searched for scorable features of the interview transcripts that could "postdict" their infants: Infants classified as "avoidant" in the Strange Situation had primary caregivers who themselves were dismiss- ing of attachment-related memories and feelings; infants classified as "anxious" had primary caregivers who were anxiously preoccupied with attachment-related issues; and infants classified as "secure" had caregivers who were "free and autonomous" with respect to attachment.
In subsequent work, a fourth infant pattern, "disorganized," was found to be associated with caregivers who were "unresolved" with respect to losses and traumas in their attachment history. In the second, completely independent line of research, Hazan and Shaverwho had been studying adolescent and adult loneliness, followed up Weiss's idea that chronic loneliness is associated with insecure attachment.
Reasoning that most chronically lonely young adults were unsuccessfully seeking a secure romantic attachment, and that orientations to romantic relationships might be an outgrowth of previous attachment experiences, Hazan and Shaver devised a simple self-report questionnaire for adults based on Ainsworth's three patterns of childhoOd attachment: The measure asked people to think back across their most important romantic relationships and decide which of the three types was most self-descriptive.
In subsequentstudies, this measure and several variants of it have been related to a host of theoretically relevant personality ': Although a few studies have correlated this measure with retrospective reports of childhood experiences with parents, the bulk of research in this tradition has focused on the influence of attachment patterns on personal adjustment and adult relationships. Bowlby was primarily a child psychia- trist, and Ainsworth was a child clinical and developmental psychologist.
Many of the current attachment researchersin the first "subculture" e.
Researchersin this group tend to think psychodynami- cally;: Hazan and Shaver were persoriality I social psychologists, and their work was quickly assimilated by other such psychologists, who tend to think in tenns of persoriality traits and social interactions, be interested in normal subject populations, prefer simple questionnaire measures, study relatively large samples, and focus on adult social relationships, including friendships, dating relationships, and marriages.
Because both lines of research are grounded in Bowlby's and Ainsworth's attachment theory, and both focus on individual differences and classify people into categories parallel to Ainsworth's infant attach- ment typology, it was inevitable that some researcherswould assume that the two adult classification systems, Main's and Hazan and Shaver's, must be highly related.
In other words, since both the AAI and Hazan and Shaver's questionnaire place people into categories roughly desig- nated as secure, avoidant, and anxious or preoccupied, it is often assumed that the two assessmentprocedures are more or less interchangeable. In ,Bartholomew reviewed the adult attachment researchin both traditions and came to the conclusion that the two approaches to assessing attachment differed in a number of ways.
First, she noted that the dismissing-avoidant individuals identified by the AAI denied experienc- ing subjective distress and downplayed the importance of attachment needs, whereas avoidant subjects identified by Hazan and Shaver's self- report measure reported relatively high levels of subjective distress and fears of becoming close to others.
She argued that two distinct forms of avoidance were evident, one pattern motivated by a defensive mainte- nance of self-sufficiency labeled "dismissing" and the other motivated by a conscious fear of anticipated rejection by others labeled "fearful". Second,she noted that the two approaches fli.: The AAI focuses on dynamics of internal working models that are revealed indirectly by the way a person talkS about childhood relationships; the measure is not based on the assumption that people are conscious of these cynamics.
In contrast, the self-report measure focuses on feelings and behaviors in close relation- ships of which a person is aware and which the person Candescribe fairly accurately. Building on both traditions, Bartholomew proposed an ex- panded model of adult attadunent that included two forms of avoidance. Problems in Comparing Measures from the Tw'o Traditions In recent years, papers have begun to appear e.
It is easy for interview researchersto add a simple self-report measure to their studies, but difficult for question- naire researchers to learn to conduct and code what are essentially intensive clinical interviews.
Thus, comparisons between different kinds of attachment measures have been made largely by researchers who concentrate on parent-child relationships and take Ainsworth's Strange Situation and the AAI as benchmarks. To the extent that conclusions about attachment measures affect researchers' understanding of attachment processesor lead them to have little confidence in particular bodies of research,the question of measure convergence is important.
Some of the authors who have found little convergencebetween self-report and interview measures have concluded that self-report measures are especially prone to measurement error and unlikely to be related to behavior.
They fail to consider studies such as those by Shaver and BrennanFeeney and Noller ,Kirkpat- rick and Davis ,Kobak and Hazan ,Mikulincer and Nachshon ,and Simpson, Rholes, and Nelligan ,which show that self- report measuresof adult attachment patterns do relate significantly to the Methods of Asssessing Adult Attachment I 29 ways in which a person discusses close relationships, to observations of marital communication, to relationship breakups, to patterns of self- disclosure, and to seeking and providing social support under stressful conditions.
These inciude insights and research ideas that arIse when an investigator hears what people actually say when interviewed in depth about important relationships. Many researchers relying on self-report measuresof attachment have also failed to seriously consider the possibility that there are aspects of attachment patterns that are inaccessibleto conscious awareness and, therefore, cannot be assessed by self-report methods cf. Given the significance of the measure convergence issue, it is essen- tial that comparisons between measures be thoughtful and statistically appropriate.
These differences make it unlikely that the two kinds of mea- sures will converge strongly. Some authors who compare attachment measures also overlook or misinterpret the domain differences between the AAI, which focuses on adults' characterizations of their childhood relationships with parents, and the Hazan-Shaver measure, which focuses on experiences in roman- tic relationships. Although the AAI has sometimes been conceptualized as assessing generalized attachment representations, the Hazan-Shaver measurewas specifically designed to meaSL: Of course, even if the two measures had been intended to measure precisely the same construct, method variance would be expected to reduce the degree of "associationbetween them.
Self-report measures focus on conscious, potentially inaccurate summaries by a person of his or her own experiences and behaviors. The AAI focuses primarily on the way a person talks about childhood attach- ment experiences, with the major distinctions having to do with what might be called defensive style e.
These differences in communica- tional behavior and defensive style are not necessarily noticed or ac- knowledged by the people who exhibit them. Given all of these differ- ences, a moderate association, at best, would be expected between the AAI and Hazan and Shaver's self-report measure.
Also problematic is the fact ;;hatcomparers of measures often fail to conduct power analyses Cohen, before performing statistical tests. Insufficient power a reflection of sample size and expected effect size is especially disconcerting when investigators are inclined to accept the null hypothesis, and sufficient power is especially difficult to attain when categorical variables are under investigation.
Researchers have also failed to consider that unreliability in both measures will always attenuate the observed degree of correspondence. Single-item measures are particularly likely to be unreliable. Given the systematic differences between the AAI and self-report romantic attachment measures, the low power of the tests used to test the associations between them, and the relatively low reliability of some attachment measures, it is hardly sur- prising that previous studies have failed to show convergence.
In the remainder of this chapter we compare various measures of attachment based on Bartholomew's typology. BecauseBartholomew has created both interview measures of parental and peer attachment and a self-report measure of peer attachment; similar in many respects to the Hazan-Shaver measure but including two avoidant categories, it is pos- sible to compare assessmentmethods without confounding them with different conceptual schemes.
In addition, continuous prototype ratings of the four attachment patterns allow for adequate power to test for moderate associations in relatively small samples. In two separate sam- ples, the associations between three measures of attachment were as- sessed-a self-report measure of general orientation to close relationships, an interview measurefocusing on early family relationships, and a second interview focusing on peer relationships.
Four prototypical attachment patterns are defined in terms of two dimensions: Two-dimensional four-category model of adult attachment.
The self model is therefore associated with the degree of anxiety and dependency on other's approval in close relationships. The positivity of the other model indicates the degree to which others are generally expected to be available and supportive.Are You Ready to Have Children? -- Challenging Questionnaire For Prospective Parents
The other model is therefore associated with the tendency to seek out or avoid closenessin relationships. Secureadult attachment is characterized by the combination of a positive self model and a positive model of others. Secure individuals have an internalized sense of self-worth and are comfortable with inti- macy in close relationships. Preoccupiedattachment is characterized by a negative self model and a positive model of others.
Preoccupied individu- als anxiously seekto gain acceptanceand validation from others, seeming to persist in the belief that they could attain safety, or security, if they could only get others to respond properly toward them.
Fearful attach- ment is characterized by negative self and other models. Fearful individu- als, like the preoccupied, are highly dependent on others' acceptanceand affirmation; however, because of their negative expectations, they avoid intimacy to avert the pain of loss or rejection. Dismissing attachment is characterized by a positive se;: Dismissing individuals also avoid closenessbecauseof negative expecta- tions; however, they maintain a sense of self-worth by defensively deny- ing the value of close relationships.
As indicated previously; three of these patterns-secure, preoccupied, and dismissing-are conceptually similar to the correspondin,i? And three-secure, preoccupied, and fearful-are similar to Hazan and Shaver's secure, anxious-ambivalent, and avoidant categories. A Test of Correspondence among Bartholomew's Attachment Measures The convergence of different approaches to assessing adult attachment was tested in two samples.
Participants completed three measures: Two independent raters coded each interview for interviewee's degree of fit to a prototype for each of the four attachment patterns. Final interview ratings were based on an average of the two coders' ratings. Thus, each participant received a profile of ratings on the four attachment patterns.
The highest of the four ratings was also used to define a best fitting categorization for each method. For each combination of methods, correlations were computed be- tween the continuous ratings of corresponding attachment patterns see Table 2. For both samples, each of the associationsbetween correspond- ing peer and family interview ratings was significant.
In addition, the associations between corresponding ratings '",ere stronger than those between noncorresponding ratings, and none of the noncorresponding ratings were significantly positively associated with each other not shown in the tablesuggesting both convergent and discriminant validity. Associations between corresponding ratings on the peer interview and self-report measure were also significant, and the pattern of correlations again suggested both convergent and discriminant validity.
In contrast, associationsbetween corresponding ratings on the family interview and self-report measure were weaker and more variable. In both samples, security and dismissing ratings were positively correlated. However, corresponding fearful ratings were associated only in Sample 1, and preoccupied ratings were not significantly correlated in either sample. In a sample of assault- ive men, correlations between self-report and family interview measures ranged from.
View and self-report measure Sample 1. Because of inadequate power, it was not possible to test the corre- spondence between classifications derived from the three different meth- ods, but there was sufficient power to compare the secure and insecure categories across methods.
With one exception, the proportion of agree- ment for the secure-insecure distinction ranged from. However, the proportion of agreement between the family mterview and self-report measure in Sample 2 was. As can be seen in the figures, the three methods exhibit substa.
The findings indicate a moderate degree of convergence across the three approaches. As might be expected, the correlational results were weakest when both the method interview vs. The results of the factor analyses also indicate convergence. This conclusion agrees with findings obtained by BellgO'Hearn and Davisand Saunders But it would be easy to miss the convergence if one compared different conceptualizations of attachment such as three vs. The interviews focused on the partici- pants' relationship with the deceasedand their responsesto the loss; they were coded by independent sets of AAI and Bartholomew coders.
Both coding systems contain the secure, preoccupied, and dismissing catego- ries. The AAI usually includes an "unresolved loss and trauma" U category as well, but in the bereavement study it could not be used because the interviewees were all mourning a loss that had occurred within the previous year.
As explained earlier, Bartholomew's coding system includes a fourth, "fearful," category not included in the AM. Perhaps a more appro- priate test of the association between the two measures is one that leaves out the seven interviewees who were classified as fearful in Bar- tholomew's system. Of eight people judged dismissing by the AAI coders, seven were judged dismissing by the Bartholomew coders. All five disagreements involved the secure category and were mainly a result of the different secure-categorybase rates for the two sets of coders.
The AAI coders more readily labeled people secure. The seven people who were coded fearful in Bartholomew's system and omitted from the 3 x 3 analysis were distributed as follows in the AAI system: As a more powerful test of the association,we conducted analyses of variance ANOVAs in which the AAI categories served as the inde- pendent variable and the four continuous prototype ratings from Bartholomew's scoring system served as dependent variables.
Considering that the interview being coded was not primarily an attachment interview i. The results suggest that strong evidence for convergence to the extent that the coding systems are parallel would be obtained from a study of a larger, more representative sample based on appropriate attachment interviews.
Participants placed themselves into one of Hazan and Shaver's three categories secure, anxious, and avoidant and rated huw self-descriptive each of the three prototypes was. They also placed themselves into one of Bartholomew's four categories secure, preoccupied, fearful, and dismissing and rated how self-descriptive each of the four prototypes was. As suggested earlier, there is no category on the Hazan-Shaver measure that is strongly parallel to dismissing, so most dismissing subjects are forced to choose fearfuL which acknowledges their avoidant tendencies, or secure, which emphasizes their autonomy and self-esteem.
Becauseparticipants in the Brennan et al. The correlations for the parallel ratings secure with secure, etc. In each case, these correlations were higher than any of the correlations among nonparallel ratings.
Frequently asked questions - Psychology Members' Site
The dismissing rating was not strongly correlated with any of the Hazan-Shaver ratings, but its correla- tion with the avoidant rating-the most lvgical quasi-parallel category- was a highly significant. You need only reference the measures appropriately. You are also welcome to revise or update the measures as you see fit, as long as you clearly describe the changes in your method section.
These measures were developed for research purposes only and are not appropriate for use in individual assessments. They have not been validated for this purpose and there are not adequate norms for the measures to allow for a confident interpretation of individual results. That being said, some practitioners have found that self-report attachment measures are helpful as a basis of informal self-exploration and discussion.
You cannot use these measures for commercial purposes. May I translate the attachment measures into another language? However, many researchers have asked for permission to translate these measures over the years and there is a reasonable chance that a translation is already available in your language of interest.
This search can be done with PsycInfo. A general Google search may also be helpful. Translations may be available in other languages as well e.
I also suggest that you consider Fraley's ECR-Revised see Fraley's websitewhich has been translated into a number of languages. We recommend you consult this article for normative data on the RQ. All of the studies conducted in our lab have been done with samples that are smaller than is necessary for establishing norms. If norms and reliability etc. The four-prototype, two-dimensional attachment model has been validated and applied most extensively in samples of young adults and community members including samples of gay men.
We recommend that when the characteristics of the target population differ from the population on which the model was validated researchers undertake a pilot study to validate the model and measures with the new population.