Idiographic assessment of attachment relationships: Construction and validation of scales for use in narrative psychotherapy

In this study the focus was on the development and validation of scales that can be used in an attachment-based narrative form of psychotherapy/counseling. One hundred and six participants completed an online survey, in which they were asked to select one significant troublesome relationship (past or present), to describe the nature of this relationship in a couple of sentences and to characterize the relationship by means of the list of 49 attachment items. By employing exploratory and confirmatory factor analysis (N = 106), a six-factor model emerged covering four classical attachment prototypes: (I) security; (II) anxiety (preoccupation with relationships, fear of rejection); (III) dismissive avoidance (i.e., defensive separation/denial of attachment need; and (IV) fearful avoidance (avoidance/alienation, threat). In two validation studies, the interpersonal and affective meaning of these scales were examined by using instruments that accentuate agentic and communal interpersonal orientations in human contact. In a subsequent qualitative study, we used a cluster-based classification into attachment groups for the exhaustive screening of the content of (a selection of) 40 texts of 40 persons. This hermeneutic approach disclosed characteristic themes for each of the secure, preoccupied, dismissive-avoidant and fearful-avoidant prototypes which are discussed in detail. In narrative psychotherapy a thematic overview according to the attachment typology may be helpful in fine-tuning the therapeutic process, in that it may increase a client’s awareness of attachment issues and associated (dysfunctional) emotional reactions.

A fundamental idea of narrative psychotherapy is that by telling and re-telling one's experiences (to an attentive listener) beneficial changes in psychological functioning can be realized.In attachment-based forms of psychotherapy, clients are typically stimulated to explore in depth their childhood and interpersonal experiences with significant others, narrating about attachment-related issues (e.g., fear of rejection/abandonment, avoidance of intimacy, and [lack of] trust in others).Throughout the last two decades, the therapeutic significance of the narrative processing (i.e., the putting into words) of attachment related issues has been widely recognized (Dallos, 2006;Daniel, 2009Daniel, , 2011;;Levy et al., 2012;Mikulincer et al., 2013;Dallos & Vetere, 2014;Graci & Fivush, 2017).Several adult attachment-based interventions have been developed for, e.g., the treatment of depression, severe interpersonal problems in couples, and borderline personality disorder (For a review, see Levy et al., 2012).These interventions are typically aimed at overcoming disruptions in interpersonal relationships and creating security by fostering the exploration of mental representations of self and others (cf.mentalization).In addition, there is some supportive evidence that (long-term and shortterm) psychodynamic treatments may not only decrease the severity of psychiatric symptoms, but also increase the coherence of attachment-related representations of significant others, and increase the narrative coherence with which patients reflect on their childhood experiences in attachmentrelated interviews such as the Adult Attachment Interview (AAI), a method which is typically used to classify people into four attachment styles (For reviews, see Mikulincer et al., 2013;Taylor et al., 2015).This has led several authors to believe that the main goal of psychotherapy should be to increase the client's narrative coherence (e.g., Wallin, 2007).Some empirical findings suggest that (psychodynamic and other) psychotherapeutic interventions may foster changes from an (incoherent) insecure to a (coherent) secure AAI classification.Notwithstanding that a certain degree of incoherence is inevitable in psychotherapy or may sometimes even be necessary (Daniel, 2009), many psychotherapists and psychologists recognize the importance of the ability to make sense of oneself and others by telling a coherent autobiographical life story (Wallin, 2007;Levy et al. 2012;Dallos & Vetere, 2014).
We propose an attachment-based narrative form of psychotherapy/counseling, in which a client is being stimulated to investigate his/her interpersonal relationships with significant others (parents, relatives, friends, colleagues and superiors, etc.) from the past and present (Cf.Van Geel et al., 2011).The method is strongly inspired by the work of Hermans and Hermans-Jansen (1995) and Kelly (1955) who was the first to apply grid-methodology at the Correspondence to: Rolf van Geel, Department of Psychology, Open University of the Netherlands, P.O.Box 6870, 6503 GJ Nijmegen, The Netherlands.Tel.: +31-24-360-00-78.E-mail: Rolf.vanGeel@ou.nllevel of a single person (Rathod, 1982; For a review of idiographic methods, see Van Geel, 2000, Chapter 3).With their book 'Self-narratives: The construction of meaning in psychotherapy', Hermans and Hermans-Jansen (1995) advocated a form of psychotherapy in which a person is considered a 'motivated storyteller': the Self-Confrontation Method (SCM).In the SCM a person is invited to construct 'valuations' (i.e., sentences that reflect important experiences), for example, "Due to our preoccupation with our sick daughter, our son was rather neglected" (Hermans & Hermans-Jansen, 1995, p. 89) or "I consider it important to constantly learn new things, to be challenged intellectually, to be mentally on the move" (Van Geel. 2000, p. 8) (For other examples, see : Hermans & Hermans-Jansen, 1995;Van Geel, 2000;Van Geel et al., 2019).The purpose of the SCM dialogue is to capture a number of topics with which the person is structuring his/her world and, subsequently, to detect affective patterns and themes in the texts that reflect two (latent) basic motives: (a) the striving for self-enhancement, and (b) the striving for contact and union.In order to provide a concise picture of the affective side of a client's self-narrativeusually a person formulates 30-40 valuationssix types of valuations are being distinguished: (1) love and unity, (2) strength and unity, (3) success, autonomy and perseverance, (4) aggression, anger and opposition (5) powerlessness and isolation, and (6) unfulfilled longing and loss.They represent basic experiences, associated with themes on a latent level, and are derived from the affective components of a valuation: the Self, Other, Positive and Negative scales.The Self (S) scale consists of indicators that express the striving for self-enhancement (e.g., self-confidence, strength).The Other (O) scale includes feelings that reflect the striving for contact and union (e.g., love, tenderness).Feelings such as joy and happiness belong to the Positive (P) affect scale, whereas feelings such as worry and unhappiness cover the Negative (N) affective domain.For every single valuation a client is rating the intensity of these affects on a 0-5 point scale (usually with the aid of a computer program), yielding characteristic "affect profiles" (For a more elaborate description of the SCM-typology, see Van Geel et al., 2019).
Unlike the SCM, we propose a method in which a person is exclusively focusing on significant relationships by writing about his/her experiences with others, by constructing one or two sentences to portray the nature of each contact.Furthermore, by rating items from a fixed list of "attachment items" for each separate text (on a 0-5 point rating scale), characteristic attachment profiles may be found and -assuming sufficient scatter in the profileseventually lead to a classification to one of several (classical) attachment prototypes (e.g., secure, preoccupied, dismissing-avoidant, or fearful-avoidant; see next paragraph).More specifically, by scoring attachment aspects indicating security (e.g., feeling understood, to feel supported), anxiety about rejection (e.g., ruminating about the relationship, feeling rejected), avoidance of closeness (e.g., withdrawing myself, avoiding the other) and denial of attachment needs (e.g., not needing the other, solving my own problems) characteristic "attachment profiles" may be discerned.The idea is that by obtaining a complete picture of a person's social world (in past and present) an idiographic assessment of attachment relationships is possible.Based on the types of and variety in relationships a person is upholding, this procedure may reveal a dominant attachment orientation and dysfunctional aspects of the attachment system, offering clues for fine-tuning the goals in psychotherapy.
In the present study the focus is on the development and validation of attachment scales for the idiographic use in a clinical counseling context.Below, we first present a short outline of attachment theory, highlighting Bartholomew's (1990) four-category model of adult attachment styles.Subsequently, we discuss the main features of an attachment-based psychotherapeutic approach and how attachment themes and characteristic narrative styles can provide useful clues for delineating the nature and goals in psychotherapy.We then discuss the main features of the attachment-based narrative form of psychotherapy that we propose.We also review some of the principles of the Self-Confrontation Method, as this idiographic method has served as a model for ours.The main objective of this study is to identify (sub)factors that materialize from the analysis of a set of attachment items pertaining to texts about (troublesome) attachment relationships.For the purpose of construct validation, the relationships of interpersonal variables (IAS-circumplex) and SCM affect-scales (SOPN) with the derived attachment scales are examined.In addition, characteristic themes from the texts allocated to one of four attachment prototypes are studied in detail.

Attachment theory and attachment prototypes
Attachment theory explains how children, based on their experiences (positive or negative) with their parents or primary caregivers, develop mental representations of themselves and others, which, at a later age, are being manifested in affective relationships with others (Ainsworth & Bowlby, 1991;Bowlby, 1973;1980.For a historical review, see Bretherton, 1992).The theory implies that children who are lovingly raised and comforted in times of stress get the message that they are worthy of attention and love (which results in a 'positive self-image') and, at the same time, that others are trustworthy and available for help and comfort (which creates a 'positive image of others') (Bartholomew, 1990).In effect, in these circumstances, a child learns that the primary attachment strategy of seeking closeness and comfort is an effective way to deal with stress (Mikulincer & Shaver, 2007).Attachment theory further implies that children growing up in a less caring environment -where parents are scarcely or inconsistently available and children are often being neglected or abused-develop negative mental representations of themselves or others.Because the primary strategy of proximity seeking and support seeking does not work (well), neglected children use alternativeso-called secondary-strategies to regulate stress: the seeking of proximity is intensified (cf.hyperactivating strategy) or the seeking of proximity is abandoned (cf.deactivating strategy) (Mikulincer & Shaver, 2007).The hyperactivating or anxious strategy is accompanied by an increased alertness for signs of separation or rejection, an exaggerated need for affirmation, attention, and support from others.The deac-tivating or avoidant strategy involves controlling the desire for closeness by devaluing the importance of intimate relationships and acting independently; the other, as a potential source of support and comfort, has actually been given up.
An important assumption of attachment theory is that the bond a child has had with his caregivers is being manifested in affective relationships with other people in adulthood.Brennan et al. (1998) hypothesize that in intimate adult relationships, two dimensions can be distinguished that determine how a person feels in the relationship: 'anxiety about rejection and abandonment (not being accepted)' and 'avoidance of and discomfort with closeness'.Bartholomew (1990) distinguished four attachment styles that refer to the four quadrants made up by the combinations of lows and highs on these two orthogonal dimensions (anxiety and avoidance).According to Bartholomew, securely attached persons have little concern of not being accepted by others (they show low "anxiety") and easily form intimate bonds (they show low "avoidance").Individuals with a preoccupied attachment style have an insatiable need for the approval of others and a deep conviction that they don't matter.Hence, they show a strong desire for intimate relationships (low avoidance), but at the same time they are afraid of not being accepted (high anxiety).Those with a dismissiveavoidant attachment style, on the other hand, deny the need for intimacy (high avoidance) in order to maintain a sense of complacency and invulnerability (low anxiety).High avoidance is also a characteristic of those with the fearfulavoidant attachment style, but this stems from fear of being hurt by potential loss or rejection (high anxiety), without really relinquishing the desire to be accepted.Research has shown that attachment-related anxiety in particular is a risk factor for depression and anxiety disorders.Furthermore, there are indications that avoidant attachment, and in particular fearful-avoidant attachment, increases the risk of dissociative disorders (Mikulincer & Shaver, 2007) which has led some researchers to regard the fearful-avoidant attachment style as an essentially dissociative form of attachment (Liotti, 2006).Wallin (2007) has proposed an attachment-focused therapy, which he perceives as a collaborative "affective linguistic dialogue" (p.201) in which clients are being stimulated to access, articulate, and reflect upon the full range of their subjective emotional experiences.The idea is that by offering a compassionately attuned environment, a client is being encouraged to disclose and tolerate (unverbalized or dissociated) painful experiences in affect-regulating interactions, thereby (ideally) promoting a coherent account of their experience.In other words, psychotherapy can be seen as an endeavor to restore the client's inhibited capacity to reflect coherently about their experiences with others.According to Wallin, the capacity for mentalization or reflective function is crucial here, i.e., the ability to understand ourselves and others in terms of intentional mental states, such as feelings, desires, wishes, goals and attitudes.He argues that psychotherapists should make it possible for their patients to mentalize, because it makes behavior understandable and meaningful and "strengthens their ability to regulate their affects, to integrate experiences that have been dissociated, and to feel a more solid coherent [and thus secure] sense of self" (p.4).

Attachment-based psychotherapy
The capacity for coherent discourse or narrative competence has been recognized as a fundamental feature of secure attachment.In the Adult Attachment Interview (AAI, Main, 1991Main, , 1995; see also George et al., 1996;Hesse, 1999;Main et al., 2008), people are being assigned to one of four categories (secure, preoccupied, dismissing, or disorganized) based on how they express and communicate their childhood attachment experiences; to some degree the disorganized AAI prototype can be compared to Bartholomew's fearful-avoidant prototype (Mikulincer & Shaver, 2007).Crucially, the AAI classification is not so much based on the content of the story (i.e., attachment themes), but rather on the coherence with which the story is being told and the degree of collaboration with the interviewer.Securely attached individuals can communicate truthfully, coherently and clearly.They speak thoughtfully and vividly about emotional experiences and seem able to remain connected to the interviewer and seem aware of the purpose of the conversation even when experiencing strong feelings (Wallin, 2007).In addition, they are able to show understanding for parents and can forgive them without neglecting their own feelings (Nicolai, 2003).Preoccupied adults, being oversensitive to clues about abandonment, separation, and rejection, seem unable to constructively reflect on their attachment experiences during the AAI.They are easily overwhelmed when talking about intense and distressing feelings related to past attachment experiences, which often leads to a lengthy and overly detailed narrative that can be difficult to follow (Wallin, 2007).Their stories may be based on truth, but are rather chaotic and confusing (Dallos, 2006;Fivush & Graci, 2017).Because dismissive attached adults are trying to control the desire for closeness by devaluing the importance of intimate relationships, they are very uncomfortable when being interviewed about their childhood experiences (Mikulincer & Shaver, 2007).They tend to deny or devaluate the significance of their experiences with their parents and seem to remember very little, which typically results in abstract conspicuously brief AAI transcripts.Parents may be idealized (without good reasons) and experiences may be distorted, e.g., a harsh upbringing as fostering independency and determination (Nicolai, 2003).Even more than preoccupied and dismissive adults, disorganized attached adults have great difficulty to recall or reflect coherently upon their (traumatic) childhood experiences, sometimes displaying striking lapses in reasoning and discourse (Wallin, 2007).When AAI probes are being used that invite a discussion of abuse or loss, they may suddenly become emotionally overwhelmed and end up in a chaotic or trance-like state of mind (cf.dissociation).
According to Mikulincer and Shaver (2007) "(…) the identification of clients' attachment working models is an important step in understanding their transference reactions and can help therapists foresee the particular kinds of irrational and inappropriate demands and responses that can emerge during psychotherapy" (p.424).For most psychotherapists a complete standardized assessment of a person's narrative of their (problematic) childhood experiences with the AAI is not a viable option (Daniel, 2009).Nonetheless, practitioners can make use of AAI elements in a clinical interview, e.g., by using similar questions related to attachment.Importantly, by listening carefully to a client's life story and being attentive to cues related to the coherence of communication and degree of collaboration, an experienced practitioner can usually make a provisional assessment of a client's predominant attachment orientation -even in the early phase of treatment (Wallin, 2007).In addition, at the start of psychotherapy, there is no impediment for using a self-report questionnaire such as, for example, the Relationship Questionnaire (RQ, Bartholomew & Horowitz, 1991), the Experiences in Close Relationships (ECR, Brennan et al., 1998), the State Adult Attachment Measure (SAAM, Gillath et al., 2009;Xu & Shrout, 2013;Strydom, 2015) or some other self-report instrument (For reviews, see Jewell et al., 2019;Mikulincer & Shaver, 2007;Ravitz et al., 2010;Van Geel & Houtmans, 2022).One could also consider a clinical observational instrument (Westen et al., 2006).So, in addition to the clinical interpretation of a client's life story and characteristic narrative style, more standardized methods are available.
An attachment perspective on psychotherapy implies that different attachment styles can be arguments for different therapeutic intervention strategies.Preoccupied clients tend to project unrealistic hopes for being loved and cared for onto their therapist, appear overinvolved and demanding, and can become frustrated and angry if a therapist is insufficiently supportive (Mikulincer & Shaver, 2007;Daniel, 2011).Clients with this attachment style seem to benefit more from working with a therapist who adopts deactivating strategies, i.e., maintains psychological distance and encourages client's autonomy (Mikulincer & Shaver, 2007, cf. Tyrrell et al., 1999;Wallin, 2007).In psychotherapy, dismissing clients appear distant and emotionally unengaged (Daniel, 2011).They display a tendency to self-isolate in order to avoid being rejected and controlled (Wallin, 2007).Because they tend to diminish the importance of others, they are likely to devaluate the therapist and reject the intervention approach.Clients with a dismissing attachment style seem to benefit more from working with a therapist who adopts hyperactivating strategies, which means that psychotherapy should be aimed at connecting more deeply to others and the expression of intimate feelings (Wallin, 2007).Disorganized clients can be recognized by their lapses in reasoning and discourse (Wallin, 2007), they experience more emotional turmoil than preoccupied or dismissing clients do and display an incoherent mix of neediness, detachment and dissociation (Van Geel et al., 2019).According to Mikulincer and Shaver (2007), the origin of disorganized attachment lies in traumatic early childhood experiences.They argue that insecure children at times experience parents as caring and available, at other times as frightened, and at still other times as frightening.This leads to multiple incompatible working models of the self (for example as a loved child and as a victim).According to Wallin (2007), psychotherapy with disorganized patients, promoting the integration of dissociated experiences is the heart of the work.He argues that "(..) while for most patients the relationship with the therapist is a significant part of therapy, for patients who are unresolved the therapeutic relationship is the therapy" (pp. 243-244).

Idiographic attachment-based psychotherapy: An adaptation of the Self-Confrontation Method
We propose an attachment-based narrative form of psychotherapy/counseling, in which a client is being stimulated to investigate his/her interpersonal relationships with significant others.We assume that a person can exhibit attachment behavior, not only with parents or intimate partners, but also with friends or in work relationships.This means that when people are distressed they may either adopt the primary attachment strategy of seeking help and support or adopt secondary strategies to regulate stress: the seeking of proximity is intensified (cf.hyperactivating strategy) or the seeking of proximity is abandoned (cf.deactivating strategy).We hypothesize that in most long-term relationships, traces of attachment behavior can be found, where a person may be securely attached to one person, but has an insecure attachment to another person.For example, a person may relate in a fearful-avoidant manner to an intimidating superior (and exhibit dissociative behavior as a result of it), but may feel securely attached to a trustworthy colleague at work (because s/he is offering a listening ear).Mikulincer and Shaver (2007) argue that: "(…) with respect to a particular relationship and across different relationships, everyone possesses models of security attainment, hyperactivation, and deactivation, and so can sometimes think about relationships in secure terms and at other times think about them in less secure, more hyperactivating or deactivating terms "(p.24).This passage fits well within an idiographic approach, as it indicates that an attachment style should not be understood as a stable personality factor, but more as a shifting state of mind with respect to relationships (Daniel, 2009;Wallin, 2007).
As mentioned previously, the method is strongly inspired by the SCM (Hermans & Hermans-Jansen, 1995).In accordance with the SCM, we consider a person to be a coinvestigator and a storyteller, being able to self-reflect, communicate and construct meaningful texts about his/her interpersonal life.However, different from the SCM in which the questions to elicit personally relevant topics have a broad scope, we propose to exclusively focus on interpersonal relationships.
The purpose of the SCM is to stimulate self-examination inducing the client to detect affective patterns and themes in his/her story that reflect two 'latent' basic motives: (a) the (agentic) striving for self-enhancement, and (b) the (communal) striving for contact and union; the agentic and communal orientations are outlined in a variety of theories and approaches in psychology (Van Geel et al., 2019).In contrast, we envisage an idiographic method that fosters selfexamination concerning the primary and secondary attachment strategies, which are being expressed in the (attachment) components of the sentences a person formulates about his/her relationships.
An important result of the self-exploration with the SCM is that (positive and negative) themes associated with the basic motives become visible in the self-narrative that offer clues for setting and fine-tuning the goals of psychotherapy.For example, a person may discover that experiences of 'love and unity' (positive O valuations, +O) are not contained in his/her valuation system, because it is too threat-ening to his/her independence.He or she may then be stimulated to explore situations in which +O experiences have a greater chance of arising and to experiment with alternative behaviors.In general, the SCM-counselor should be attentive to the possible 'one-sidedness' of the self-narrative as a whole because this may indicate dysfunction.Hermans and Hermans-Jansen (1995) assume that for an optimal psychological functioning the fulfillment of both motivations is required, and that, correspondingly, psychopathology/dysfunction is (typically) the result of an imbalance between the desire for intimacy and that for autonomy (For a more elaborate discussion on how psychopathological aspects of SCM self-narratives can be conceptualized -supplemented with case illustrations, see Van Geel, 2000;and Van Geel et al., 2019).Likewise, for a psychotherapist adopting an attachment-based idiographic method, it is just as important to be attentive to the one-sidedness of a person's social world (i.e., an imbalance between the primary and secondary attachment strategies) as it may reveal a predominant attachment orientation and dysfunctional aspects of the attachment system.
A distinctive feature of the SCM is that the client's experiences are being recorded in separate sentences as emerging from the collaboration with the psychologist.Hence, the self-narrative as it unfolds in the initial phase and is being modified during the course of psychotherapy is, in fact, the result of a "co-construction of psychological reality" in a "dialogical symmetrical collaborative relationship" (Hermans & Hermans-Jansen, 1995).During the phase of intensified communication in which the valuations are being formulated, an experienced counselor -being aware of how mental health and psychopathological aspects of self-narratives have been conceptualized-will undoubtedly stimulate a client to formulate a coherent and complete account of experience.Hermans and Hermans-Jansen (1995) argue that "(…) most skilled counselors and therapists assert that even when they are intensely engrossed in helping clients create new life stories, they retain a metatheory and conceptual dynamics in the background of their mind" (p.29).Therefore, even in the initial phase of psychotherapy, when helping their clients to construct a coherent life story, SCM practitioners will most likely be instinctively alert for issues related to the SCM typology, i.e., the latent side of a self-narrative.When working with an attachment conceptual framework, the focus is of course exclusively on interpersonal relationships (which creates an alertness for attachment related themes, e.g., [lack of] trust in others, fear of rejection, avoidance of intimacy), but the communication between a client and a psychologist is largely comparable to that in the SCM (co-construction in a collaborate relationship) aimed at obtaining a coherent and complete account of a client's (interpersonal) experiences.

The present study
In the present study the focus is on the development of idiographic attachment scales that can be applied in a therapeutic context.Envisioning a list for idiographic use in which a person is connecting attachment items to multiple texts about a variety of relationships (say 10-20), we used Bartholomew's (1990) vignettes of the four-prototype model as a starting point and consulted various other sources (articles, handbooks and self-report questionnaires) for obtaining ideas for constructing items.We aimed at constructing 'relatively compact' items.Affects and adjectives, although compact, seem less suitable for investigating attachment relationships.Items in self-report questionnaires, on the other hand, offer a rich context and may include complex (ambivalent) content, but are generally too extensive.Therefore, we opted for an intermediate approach, in which interpersonal verbs were being combined with words that evoke an attachment context.In the end, this led to a research list of 49 items (see Appendix).In order to elicit a distressing attachment context, we ask people to select a troublesome relationship (from the past or present) and to describe the nature of the relationship in a couple of sentences and to characterize the relationship by means of the list of attachment items.Considering the nature of the initial item pool, we expect to detect factors/scales that reflect the four attachment prototypes, but we are also attentive to possible facets within these overarching dimensions.
With regard to validation, we include a circumplex of interpersonal adjectives, which positions a spectrum of interpersonal behaviors on a circle around two bipolar axes, which are best known as power (submissive vs. assertive) and affiliation (cold vs. warm) (Horowitz, 2004;Leary, 1957;Wiggins, 1996).According to this circumplex, interpersonal behaviors are assumed to be composed of blends of agentic and communal orientations.For example, agentic separation (dominant, controlling) can be distinguished from agentic communion (open, enthusiastic); the same goes for submissive separation (distant, aloof) and submissive communion (accommodating, self-effacing).A complete circumplex comprising eight scales is presented in the method section.We predict that the secure attachment scale will be positioned in the communal or agentic-communal segment of the circumplex, the preoccupied attachment scales in the submissive-communal segment, the dismissive-avoidance scales in the (agentic) separate segments, and the fearful-avoidance scales in the submissive-separate segment.
For additional validation, we investigate the relationships of the newly formed attachment scales with the Self, Other, Positive, and Negative Affect Scales of the SCM (SOPN-scales, Hermans & Hermans-Jansen, 1995).As mentioned previously, valuation theory distinguishes six basic types of experiences, each representing a different theme.Hermans and Hermans-Jansen situated the SCM prototypes in a two-dimensional circular model, in which two (bipolar) dimensions can be recognized that underlie the hexagonal arrangement: (a) self vs. other, differentiating S types (+S and −S) from O types (+O and −O); and (b) positive vs. negative, distinguishing negative (−S, −LL, −O) from positive types (+S, +HH, +O).In the present study, we will use the thematic hexagonal configuration as a reference frame for mapping 'cases' and 'psychological scales' in a group space (cf.circumplex analysis).According to Van Geel (2000), Hermans' hexagonal arrangement of thematic experiences bears a lot of resemblance to the circumplex model of interpersonal behavior.Consequently, we use similar arguments regarding the positioning of the attachment scales within the boundaries of the hexagon of affective ex-periences.Secure attachment shows a theoretical resemblance to the theme of 'love and unity' (viz., they both concern a communal interpersonal orientation), whereas preoccupied attachment has a theoretical resemblance to the theme of unfulfilled longing (viz., they both concern a submissive-communal interpersonal orientation).Furthermore, dismissive-avoidance shows a resemblance to the theme of 'anger and opposition', and fearful-avoidance shows a theoretical resemblance to the theme of 'powerlessness and isolation' (viz., they both concern a submissive interpersonal orientation).Consequently, we predict attachment scales pertaining to security, preoccupation, dismissive-avoidance and fearful-avoidance to be situated in the vicinity of the +O, -O, -S and -LL segments, respectively.A concept mapping technique, so-called hexagon analysis, will be employed to investigate these kinds of relationships graphically.
In addition to these initial validation studies, we examine the texts concerning a troublesome relationship.The objective of this qualitative study is to examine the content expressed in texts that are classified as either secure, preoccupied, dismissive-avoidant, or fearful-avoidant.For that purpose, initially, cluster analysis is applied (using the newly formed attachment scales), with which texts are allocated to one of four attachment prototypes.Subsequently, within each of the attachment groups we explore the content of the texts, expecting to discover meaningful differences.By adopting such a hermeneutic approach, we not only expect to find characteristic themes of the four attachment orientations (thus replicating findings from previous research), but we also expect to discover new themes associated with insecure attachment orientations.

METHOD Participants
The participants of this study took part in an online survey about "The Experience of Difficult Relationships" of Goodman (2018), who collected data for her master's thesis in the summer of 2016 (under supervision of the first author of this article).Respondents were recruited from the student population of the Open University of the Netherlands (OU) via online discussion fora.Goodman (2018) and seven other psychology students (who started their bachelor research at the OU in 2016 under supervision of the first author of this article) contacted their social network (friends, acquaintances, colleagues, and students) by sending a personal email using the invitation letter, with the request to fill out the online survey anonymously.People were also asked to forward the invitation letter (including the link to the online survey) to their acquaintances or colleagues who might be eligible to participate in this study (cf.snowball method).Of the 248 respondents who responded to the call to fill in the questionnaire about "The Experience of Difficult Relationships", 106 respondents filled in the questionnaire (almost) completely.The majority of the study group was female (79.2%).The age of the men ranged from 41 to 74 (M = 58.1.,SD = 9.2) and the age for the women from 22 to 74 (M = 50.0,SD = 10.9).Hence, the mean age of the men was substantially higher than that of the women, η 2 = .09.

Procedure
The online survey consisted of two major parts.In the first part of the questionnaire, a participant was asked to think of a person with whom they have (had) a difficult relationship, and to describe the contact with this person in more detail on the basis of the following four open sentences: When I think of this person, then…; The contact is/was…; Our relationship can best be described as…; and With this person I feel/felt... (cf.sentence completion test).In the second part of the questionnaire, the respondent was invited to characterize the specific relationship using a list of attachment items, interpersonal adjectives and affect terms (see Measures).At the time, the study was considered as not being within the scope of "Medical Research Involving Human Participants Act" (WMO, Wet Medisch-wetenschappelijk Onderzoek met mensen), as the survey was presented only one time.It was assumed that filling out the questionnaire would not be too burdensome and emotionally within acceptable limits.In the invitation letter, participants were briefly informed about the nature of the study, the voluntary basis of the research and the anonymous processing of the data.

Measures
We used three types of measures to characterize the specific troublesome relationship: (1) a list of attachment items, (2) interpersonal adjectives, and (3) affect terms.

Attachment Relationships: Idiographically Assessed (Research List, ARIA-49)
The initial item pool was constructed by Goodman (2018) and consisted of 49 attachment items that, in theory, could be reduced to four subscales: (1) Secure, 16 items, e.g., "trusting each other" and "having an intimate connection with the other"; (2) Preoccupied, 10 items, e.g., "ruminating about the relationship" and "feeling rejected"; (3) Dismissive-avoidant, 13 items, e.g., ''to behave independently" and "presenting myself as invulnerable"; and (4) Fearfulavoidant, 10 items, e.g., "avoiding the other" and "to feel alienated" (see Appendix).The latter three subscales refer to "insecure" attachment, totaling 33 items.The instruction was: "Below is a list of words about the nature of a relationship.Which words would you use to describe your relationship with this specific person?The objective is to describe yourself in the relationship with the other, and ask yourself the question: What is/was characteristic of the relationship?"The 49 items were being scored on a 6-point scale (0 = not at all applicable, 1 = hardly applicable, 2 = slightly applicable, 3 = considerably applicable, 4 = strongly applicable, and 5 = completely applicable).

Interpersonal Adjectives Circumplex
The interpersonal circumplex was operationalized with a shortened Dutch version of Wiggins's (1979) Interpersonal Adjective Scales (IAS-32, Van Geel, 2013).It assesses agentic and communal experiences in interpersonal relationships, by means of eight scales that are being arranged in a two-dimensional circular structure.This version consists of 32 adjectives that describe interpersonal behavior (Van Geel, 2013).The four-item scales are alphabetically designated according to the anti-clockwise arrangement in the circumplex: ( 1 The objective is to describe yourself with the aid of this list, and ask yourself the question: What kind of person am/was I in the contact with the other?"The 32 adjectives were being scored on a 6-point scale (0 = not at all applicable, 1 = hardly applicable, 2 = slightly applicable, 3 = considerably applicable, 4 = strongly applicable, and 5 = completely applicable).

Self, Other, Positive, and Negative Affect Scales
The SOPN-24 affect scales (Van Geel & De Mey, 2003) were developed within the context of Hermans' Self-Confrontation Method (SCM), a form of counselling in which agentic and communal themes of a person's self-narrative are identified and discussed in a profound dialogue (Hermans & Hermans-Jansen, 1995).The Self (S) scale consists of indicators expressing the (agentic) striving for self-enhancement (e.g., self-confidence, strength).The Other (O) scale includes feelings that reflect the (communal) striving for contact and union (e.g., love, tenderness).Feelings such as joy and happiness belong to the Positive (P) affect scale, whereas feelings such as worry and unhappiness cover the Negative (N) affective domain.In this study these scales were used to gain an impression about how one was feeling in a specific troublesome relationship.The instruction accompanying the SOPN affect list was: "Rate the extent to which the feelings presented below are/were evoked towards this person.What feelings do/did you experience in this relationship?"The 24 items were being scored on a 6point scale (0 = not at all/not applicable, 1 = slightly, 2 = to some extent, 3 = rather much, 4 = much, and 5 = very much).

Analysis
The analyses comprise two phases, a first quantitative phase and a second predominantly qualitative phase.In the first phase, the focus was on finding psychometrically sound and meaningful factors in the initial pool of attachment items (ARIA-49), by employing a mix of exploratory and confirmatory factor analytic techniques.For the purpose of validation, we examined the correlations of the derived (sub)factors with interpersonal adjectives scales (IAS-32 circumplex).For a graphical integration of the variables, a circumplex analysis was performed, a kind of factor analysis in which variables that can be arranged in a circular pattern (in two dimensions) serve as background for other variables.
Applying this method to our data, we first approximated the octagonal configuration of the eight interpersonal scales with the aid of orthogonal procrustes rotation (Verboon, 1994) and subsequently, projected the attachment scales within the boundaries of the octagon (Wiggins & Broughton, 1991) (see Figure 1).Next, the correlations of the derived attachment scales with the SOPN affect scales were examined.We also explored the relationships visually by employing a concept mapping technique (hexagon analysis) with which scales are being projected within the boundaries of a hexagonal frame representing six valuation types (see Figure 2).
In the second "qualitative" validation phase, we shifted the focus to the texts pertaining to troublesome relationships.We subdivided the dataset into four attachment clusters (prototypes) with the aid of cluster analysis, and selected ten persons from each of the clusters, attempting to find thematically meaningful texts for further study.As a heuristic guideline for selection, we were particularly keen on the length, coherence, richness (i.e., degree of self-reflection and self-disclosure), and presence of relevant attachment topics (e.g., childhood experiences, parental upbringing, intimacy in friendship and love relationships).While reading the texts, we were also attentive to interpersonal and emotional problems, and associated dejected states of mind.In so doing, we eventually selected 40 texts of 40 persons who were reflecting on their experiences in depth when thinking about an important troublesome relationship.In order to facilitate interpretation and improve comprehensibility to readers, the sentences or text fragments of the "sentence completion test" were assembled, summarized and syntactically more streamlined.Subsequently, these streamlined texts were independently reviewed (by the 1st and 3rd author of this article) and -after some discussion-provided with a concise description of the nature of the troublesome relationships (e.g., "unbalanced friendship, cf.therapist-client" or "stressful, transgressive and difficult relationship with dominant person") and additional main themes/topics (e.g., "self-sacrificing behavior" or "feeling disrespected, not being able to be myself").
Most statistical analyses were performed with the standard tools of SPSS 28 (exploratory factor analysis, calculations of Omega reliabilities, data-screening and clusteranalysis), but circumplex analysis and hexagon analysis were executed with specially written SPSS Syntax algorithms (Van Geel, 2006, 2011).Confirmatory factor analysis was performed with the R-package LAVAAN (Rosseel, 2012;2021), which has been made easily accessible via the graphical user interface Jamovi (version 2.2.5).

Exploratory factor analysis
In order to examine the factorial pattern of the 49 attachment items (see Appendix), we conducted principal components analysis (PCA) of the 16 secure and 33 insecure items separately (N =106).
In the PCA of the 16 secure items one conspicuous large factor emerged with an eigenvalue of 7.9, accounting for 49.6% of the variance.Psychometrically, this scale of 16 secure items seemed sound, as most corrected item-total correlations were larger than 0.50, except for one item (23, rit = 0.32).
In the PCA of the 33 insecure items three (unrotated) factors emerged with eigenvalues larger than two, accounting for 42.9% of the variance.This item pool of insecure items was further refined by screening for items with low loadings on all factors (< .30)and items with high secondary loadings (> .40) in the pattern matrix (Principal Axis Factoring, Oblimin rotation).As a result, eight items were progressively removed (i.e., 4, 10, 24, 25, 35, 39, 41, 47).The remaining 25 items clearly displayed a simple three-dimensional structure, which accounted for 47.7% of the variance.The first factor was a clear anxiety factor containing nine items related to the "preoccupation with others", including concerns about rejection, self-doubt and self-sacrificing tendencies.The second factor included seven items expressing independency, detachment from others and a denial of attachment needs (cf."dismissive-avoidance").The third factor was composed of nine items pertaining to "fearful" experiences in which one feels threatened, alienated and inclined to suppress feelings (cf.dissociation).The internal consistencies of all three candidate (higher-order) insecure attachment scales were marginally acceptable: within each subscale, all corrected item-total correlations were larger than 0.30.Hence, three higher-order factors corresponding to the classical insecure attachment prototypes seem to be detectable in the observed correlations among a selection of 25 items.Even though the internal consistencies of all (possible) subscales within all of the three insecure factors were The internal structures of the three insecure (higher order) factors were studied in more detail by extracting twofactors within each of these domains (PAF, Oblimin).The results suggested that within the anxiety factor a distinction could be made between two subfactors: "[fear of] rejection" (1, 6, 9, 33) and "worrying and self-doubt" (26,28,29,37,45); however, item 33 ("fear of getting hurt") displayed a moderate secondary loading on the other subfactor.Regarding the dismissive-avoidant factor, the two-factor solution disclosed a subfactor predominantly pertaining to the "denial of attachment needs" (7,22,32,34,43) and a very small subfactor expressing "independency" (17, 20).Finally, within the fearful factor it seemed that a separation could be made into a facet concerning "threat and abuse" (12,15,19,21) and a facet covering "avoidance and alienation" (2,13,16,31,36); however, item 21 ("to be cautious of the other person") displayed a moderately high secondary loading on the other subfactor.The internal consistencies of all (possible) subscales in all domains were acceptable: within each subscale, all corrected item-total correlations were larger than 0.30.

Confirmatory factor analysis and refinement of subscales
Before examining the complete CFA model, in which all of the (remaining) items of the secure, anxiety, dismissiveavoidant and fearful-avoidant factors were included, we first investigated the item pools of the four factors separately within a CFA framework.In contrast to EFA, the CFA framework offers the possibility of testing a theoretically based clustering of items (by fixing cross-loadings to zero), and estimating method effects (by permitting measurement errors to be correlated) (Brown, 2006).Moreover, in order to approximate an acceptable CFA model fit (CFI > .90,TLI > .90,RMSEA < .06,SRMR < .08,cf.Brown, 2006), items with large item-to-factor and large item-to-item error correlations may be progressively eliminated, as signaled by many large modifications indices (of an item).
Second, in accordance with the EFA results, we fitted two CFA models of the anxiety items.In the one-factor model all anxiety items were allowed to load on one general factor.In the two-factor model the items referring to "(fear of) rejection" were allowed to load on one factor and the items about "worrying and self-doubt" on a second factor.In order to improve model fit, one correlated error between similarly worded items was included in both models (item 1-item 6).The LAVAAN results indicated that the fit measures of the two-factor model (χ 2 /df = 46.15/25=1.85 , CFI = 0.92, TLI = 0.88, RMSEA = 0.09, SRMR = 0.07) were much better than those of the one-factor model (χ 2 /df = 73.37/26= 2.82, CFI = 0.82, TLI = 0.75, RMSEA = 0.13, SRMR = 0.10).Hence, these results imply that a two-factor model is most appropriate for adequately explaining the observed correlations (between the anxiety items).
Third, with regard to the dismissive-avoidant items, we fitted two CFA models.In the one-factor model all items were assumed to load on one general factor, whereas in the two-factor model the items referring to "denial of attachment needs" and "independency" were assumed to load on two different factors.The LAVAAN results indicated that the fit measures of the two-factor model (χ 2 /df = 19.27/13=1.48 , CFI = 0.96, TLI = 0.94, RMSEA = 0.07, SRMR = 0.05) were indeed much better than those of the one-factor model (χ 2 /df = 43.37/14= 3.10, CFI = 0.83, TLI = 0.75, RMSEA = 0.14, SRMR = 0.08).However, it must be mentioned that a one-factor model with an extra correlation between the measurement errors of item 17 and item 20, an equivalent model to the two-factor model is being obtained, i.e., with identical good fit measures (See Brown, 2006, p. 203-209).Hence, in subsequent analyses, we must keep in mind that a one-factor model (with correlated errors 17 and 20) is equally useful for adequately explaining the observed correlations (among the dismissive-avoidant items).
Fourth, we fitted two CFA models concerning the fearful-avoidant items.In the one-factor model all items were assumed to load on one general factor, whereas in the two- Note: The intercorrelations are based on "ipsatized" scale scores.The factor loadings were obtained with the aid of orthogonal Procrustes rotation; the ideal coordinates of the target solution (cf., 'ideal loadings' in brackets) are cosines and sines corresponding to the angular positioning in a regular octagon (cf., 'ideal angles' in brackets).The (near zero) correlations between perpendicular scales are in boldface.The (negative) correlations between opposite scales are in boldface italics.
factor model the items referring to "threat and abuse" and "avoidance and alienation" were assumed to load on two different factors (cf.EFA results).The one-factor model of all nine items did not directly result in an acceptable fit, due to some large residual correlations.Initially, the fit measures of the two-factor model were not impressive either, but after removing two items with high secondary loadings, i.e., "to feel restricted by the other person" (item 16) and "to be cautious of the other person" (item 21), almost perfect fit was reached (χ 2 /df = 5.50/13 = 0.42, CFI = 1.00,TLI = 1.06,RMSEA = 0.00, SRMR = 0.03).Hence, these results seem to imply that a trimmed two-factor model is most appropriate here.

Correlations between idiographic attachment scales and data screening
We examined the correlational structure of our idiographic attachment inventory, consisting of scales created by unit weighting of items, according to the six-factor model (see Table 1).We started by (1) inspecting the fit between the frequency distributions and the assumptions of multivariate analysis (i.e., extreme deviations from 'normality'), (2) identifying univariate and multivariate outliers, (3) checking pairwise linearity and (4) evaluating multicollinearity (cf.Tabachnick & Fidell, 2001;Field, 2018).Overall, no serious deviations from normality were detected, although the distribution of the security scale was somewhat positively skewed and slightly peaked (skewness = 0.98; kurtosis = 1.47), but only one univariate outlier was found for this scale (z = 3.50).The (absolute) skewness and kurtosis statistics of the other scales were consistently low (< 1.00) and no univariate outliers were detected (z > 3.25).In addition, no multivariate outliers were detectedi.e., with a large Mahalonobis distance, Χ 2 (4) > 22.458, p < .001.
We also checked pairwise scattergrams, which showed that the person with the highest Mahalonobis distance score (20.595) could possibly be considered as an influential case (i.e., located somewhat outside the ellipse of datapoints in some of the scattergrams), but in most of the graphs there was nevertheless a distinct "lineair" pattern visible.Finally, no exceptional problems were encountered with respect to multicollinearity: all VIF's were < 2.5.
Table 2 reports some descriptive statistics, reliabilities and intercorrelations of the attachment scales.It shows that most attachment scales are (positively or negatively) correlated with each other, except for the Preoccupation facet scale of Anxiety.This latter anxiety facet scale exhibits near zero correlations with Security and with Defensive Separation, low positive correlations with Avoidance and Threat and only a moderate positive correlation with the Rejection facet scale of Anxiety (0.39).Hence, the correlation pattern as a whole does not show a clear higher-order grouping into anxiety and avoidance aspects.A second-order principal factor analysis (PAF, Promax) of the six scales suggested that there were two (correlated) factors present in this dataset (explaining 50.1% of the variance): one bipolar factor consisted of a contrast of Security (-0.74)versus Avoidance (0.84), Defensive Separation (0.64), Threat (0.47) and Rejection (0.46) and the other unipolar factor consisted of the two anxiety scales, Preoccupation (0.85) and Rejection (0.34); all other (absolute) loadings in the pattern matrix of the rotated solution were lower than 0.20.Although the rejection subscale cannot be unambiguously conceptualized as a facet of an overarching higher-order factor, it seems that there are nevertheless two higher-order dimensions observable in the attachment scales pertaining to troublesome relationships.One dimension contrasts secure attachment with dismissive-avoidance (denial of attachment needs) as well as fearful-avoidant aspects (alienation and threat).The other dimension covers aspects that are predominantly related to preoccupied attachment (worrying and fear of rejection).

The interpersonal features of the attachment scales
In order to examine the agentic and communal interpersonal features of the attachment scales, we explored the relation- ships with the eight IAS scales.First, we examined the distributions of the IAS scales and looked for outliers.In doing so, we discovered no serious deviations from normality, although the DE scale was somewhat positively skewed and slightly peaked (skewness = 1.22; kurtosis = 1.54), and two univariate (but no extreme) outliers were found for this scale (z > 3.25).The absolute skewness and kurtosis statistics of the other scales were consistently low (< 1.00) and no univariate outliers were detected (z > 3.25).Next, we inspected the reliabilities (McDonald's ω), some descriptive statistics (M en SD), and the intercorrelations of the "ipsatized" IAS scales (see Table 3).The internal consistencies of all scales were adequate (ranging from 0.71 to 0.90), albeit somewhat lower for the submissive-communal scale (JK).Overall, respondents tended to describe their experiences (with the other) as more communal (JK and LM means were above the midpoint) and as less separate (BC and DE were below the midpoint).
A requirement for circumplex analysis is that the eight IAS-32 scales can be arranged in an octogonal pattern, in which the scales are evenly spread around a circle.This model is known as the "circulant correlation model", referring to the descending and ascending pattern of the "banddiagonal elements" of the correlation matrix (Gurtman & Pincus, 2000); To clarify, two of these characteristic band diagonals are accentuated in boldface and boldface italics in Table 3.One way to test whether the 28 correlations in Table 3 match with the "ordinal properties" of a circumplex model is to compare the magnitude of correlations between adjacent (at 45 o ), orthogonal (at 90 o ), octants at 135 o and opposite octants (at 180 o ), for example: r (PA,BC) > r (PA, DE), > r (PA, FG) > r (PA,HI).A circular model comprising eight evenly distributed scales results in 288 predictions about the relative magnitudes of correlations among scales (Locke, 2019).We used the R-Package RANDALL that incorporates a randomization procedure to test for these hypothesized ordinal relations (Tracey, 2000;2016).RAN-DALL computes a Correspondence Index (CI) equal to the proportion of predictions met minus the proportion of predictions violated.The CI can range from −1.0 (all predictions violated) to 1.0 (perfect fit).In the current data, 284 out of 288 predictions were met, CI = 0.96, p < 0.0004, indicating good conformity to a circular model.Regarding the band-diagonal structure, there is a descending and ascending pattern discernible (see Note to Table 3).However, there are also some deviations from a perfect circumplex model.For example, close inspection of Table 3 shows that the correlation between the "opposite" octants LM and DE (at 180 o ) is identical to that between LM and FG (at 145 o ).Notwithstanding some deviations from the ideal circumplex, we Figure 1.Circumplex of interpersonal adjectives (IAS-32) containing attachment scales believe that this circumplex offers a useful reference frame for the interpretation of other variables.The communal and agentic loadings and corresponding angular locations (see last three columns in Table 3), also demonstrate that the octagonal configuration as envisioned can be sufficiently approximated with the aid of orthogonal Procrustes rotation.
Table 4 reports the correlations between the attachment and IAS scales.The correlational pattern of Avoidance and Defensive Separation with the circumplex scales were somewhat similar, disclosing weak to strong positive correlations with DE and FG, and moderate to strong negative correlations with the LM and NO scales.For Security the opposite pattern of correlations was found, showing moderate to strong negative correlation with DE and FG, and strong positive correlations with LM and NO.The correlational patterns of Preoccupation and Threat were somewhat similar, disclosing weak to moderate positive correlations with HI and JK and weak to strong negative correlations with PA and NO.The correlational pattern of Rejection was somewhat comparable with that of the Threat scale, showing moderate positive correlations with FG and HI, and a weak negative correlations with NO.
A similar grouping of scales can be observed in Figure 1, which presents an integrated perspective of the attachment scales against the background the interpersonal circumplex.For correct understanding: the projection shows how all of the variables are interrelated in terms of interpersonal behavior.Variables that are close together in the same segment of the circumplex will more often than not intercorrelate positively, but the projection may deform those relations.The position of Security in the upper right quadrant of the circumplex signifies that this attachment scale is associated with an agentic-communal behavioral attitude.
The positioning of the anxiety subscales Preoccupation and Rejection in the lower part of the circumplex implies that they are associated with a submissive orientation.Submission also seems characteristic of troublesome relation-ships in which people experience a distinct threat.The projection of Defensive Separation and Avoidance in the lower left quadrant indicates that these scales are related to a (purely) separate and submissive-separate interpersonal orientation, respectively.

The affective features of the attachment scales
In order to examine the agentic and communal affective features of the attachment scales, we explored the relationships of the attachment scales with the Self, Other, Positive, and Negative affect scales (SOPN-scales).First, we examined the distributions of the SOPN scales and looked for outliers.In doing so, no multivariate outliers were detected-i.e., with a large Mahalonobis distance, Χ 2 (4) > 18.467, p < .001.Furthermore, no severe deviations from normality were visible, although the N scale was positively skewed and somewhat peaked (skewness = 1.22,kurtosis = 1.98).
Next, we inspected the reliabilities (McDonald's ω), some descriptive statistics (M en SD), and the intercorrelations of the SOPN-scales (see Table 5).The internal consistencies of all scales were adequate (ranging from 0.77 to 0.90).As expected, respondents mainly outlined their troublesome interpersonal relationships in negative terms, i.e., with relatively low scores on S, O, and P, and high scores on N. The S and O scales were strongly correlated, and both scales displayed even stronger correlations with Positive affect (P).In addition, moderate negative correlations were found between Negative affect and the other three scales.
The correlations of the attachment scales with the SOPN-scales are presented in Table 6.Overall, the pattern of correlations were similar for Rejection, Defensive Separation, Avoidance and Threat, showing moderate to strong negative correlations with S, O, and P and a moderate to strong positive correlation with N; Note that the correlations for Avoidance are much stronger than for the other three (insecure) attachment scales.For Security the opposite pattern of correlations was found, showing very strong positive correlations with S, O, and P and a moderate negative correlation with N. Most notably for the Preoccupation scale were the strong positive correlation with N and moderate negative correlation with the S scale.
For a graphical integration of the attachment variables, a so-called 'hexagon analysis' was used (Van Geel, 2011).This kind of concept mapping provides a depiction of the attachment scales against a hexagonal 'two-dimensional' background of SCM prototypes (Hermans & Hermans-Jansen, 1995).Technically, scales are being projected into the hexagon by calculating the correlations with the two underlying main axes, i.e., 'Self vs. Other' (Dimension 1), and 'Negative vs. Positive' (Dimension 2).The technique is fairly straightforward as it uses the factor score coefficients derived from a principal component analysis of SOPNscales of the (six) extreme prototypes (Van Geel & De Mey, 2004).Usually, the following formulas are used in the calculations: DIM1 = (O -S); DIM2 = 0.38269*(O + S) + 0.92388*(P -N) -1.91343.Figure 2 presents the result of this fixed hexagon analysis, showing how the eight attachment scales are interrelated affectively and thematically, i.e., in terms of their shared similarity with the SCM proto- types.Variables that are close together in the hexagon will more often than not inter-correlate positively, but the projection may deform those relations.The location of Security at the top of the hexagon indicates that this scale is associated with a mix of experiences related to 'strength and unity' (+HH) and 'unity and love' (+O).The positioning of Preoccupation in the bottom right of the hexagon signifies that this attachment aspect is associated with 'unfulfilled longing and loss' (-O).The compact cluster of four insecure attachment in the bottom left of the hexagon signifies that these aspects (in troublesome relationships) are predominantly associated with 'powerlessness and isolation' (-LL).

Four attachment clusters
We subdivided the dataset into four attachment clusters (prototypes) with the aid of cluster analysis, aiming to select thematically meaningful texts from these clusters for further study (see next paragraph).Several hierarchical cluster algorithms were used (on the standardized scores of six attachment scales), but the most promising and comprehensible results were obtained with Ward's method (i.e., with squared Euclidean distances).To ascertain the stability of the solution an additional 'nonhierarchical' clustering was performed with the centroids from the hierarchical procedure as seeds.The agreement measure between these two four-cluster solutions indicated substantial convergence (Kappa = .84).The results from the nonhierarchical four-cluster solution with a given start configuration were used in this study.Table 7 summarizes the means of the four clusters on the six attachment scales.To some extent the profiles of mean scores of these clusters are in agreement with the four attachment prototypes.Persons allocated to the first cluster represent a secure group: they display a high average score on the secure scale and below average scores on all insecure scales.The second cluster can be described as a preoccupied cluster, with a moderate positive average score on security, a high average score on preoccupation and below average score on defensive separation.The third cluster represents a dismissive-avoidant cluster, with low average scores on security and preoccupation and a high average score on defensive separation.The fourth cluster is clearly a fearfulavoidant cluster as it displays a low average score on the secure scale and high average scores on all insecure scales.The effect sizes (η 2 ) indicate that all six sales contribute substantially to the differentiation between clusters, but the largest differences are related to differences in Security (η 2 = .51)and Avoidance (η 2 = .59).We conclude that four meaningful clusters have been extracted from the data.

Qualitative exploration: (in)secure attachment themes in texts
Previously, we examined the agentic and communal features of the attachment scales by focusing on the numerical similarities with interpersonal scales (IAS) and affect scales (SOPN).Although such a quantitative approach has its merits, it is also limiting as texts remain hidden behind a label or typification.We now turn to the texts that have been allocated to one of four attachment prototypes.By adopting such a hermeneutic approach, we were expecting to find characteristics and, possibly, new themes for the (in)secure attachment orientations.Four tables each contain the texts of ten persons that have been allocated to either the secure (Table 8), the preoccupied (Table 9), the dismissiveavoidant (Table 10), or the fearful-avoidant attachment cluster (Table 11).As explained previously, these texts were selected because of their richness of content (i.e., presence of attachment, interpersonal and emotional issues).
The texts of the secure cluster (see Table 8) reveal a diversity of relationships with a family member, a friend or (ex-)partner, of which some seem more problematic than others.Most pertain to still continuing (long-term) relationships, and only one text refers to a broken up relationship (S10).Some of the texts include a temporal aspect indicating that the relationship has improved over the years (S1, S4, S9) or deteriorated as compared to the past (S5, S7, S10).Most texts are associated with relatively high security scores, but the elevated scores for some of the insecure scales disclose an ambivalent quality for many of the relationships.Some texts describe a predominantly good relationship characterized by (a) mutual appreciation and lovein a respectful relationship (S1, +HH), or (b) comfort and acceptancein a former conflictual sister-relationship (S4, +HH), whereas others refer to an ambivalent intimate relationship in which 'love/intimacy' is being alternated by (c) experiences of confusionin a budding love affair (S2, +HH), (d) pain and uncertaintyin an adventurous relationship (S3, -O), (e) annoyancein a fading love affair (S7, +HH), or (f) despondencyin a complicated intense love affair (S8, +O).Within the secure cluster, the lowest security scores were registered for texts about (a) a deteriorated long-term friendshipstill evoking resistance (S5, +S), (b) a one-sided friendship, cf.therapist-client (S6, +O), (c) a former suffocating friendship which has improved somewhat lately (S9, -O), and (d) a broken up relationship with a constricting dominant partner (S10, -O).Additionally, in a few texts somewhat high scores on preoccupation (doubt, ruminating and worrying) were observed, revealing a serious downside of a capricious love relationship (S8) or excessive consideration for a friend (S9).Furthermore, some noteworthy high scores on defensive separation (i.e., denial of attachment needs) were observed (S4, S5, S7, S9), which mostly outweighed the security score, underlining the ambiguity of these "secure" relationships.
All texts from the preoccupied cluster (see Table 9) concern a problematic unbalanced/ ambivalent relationship with a friend, (ex-)partner or parent (P10), with whom in most cases there is no longer any contact (P1, P2, P4, P6, P8, P9).Most texts show relatively high preoccupation scores (cf. the defining feature of this cluster), but most rejection scores are also elevated (above midpoint: > 2.5).Some of the texts convey a (strong) mutual consideration for each other, expressed in (a) (childlike) limitlessness/openness for each otherin a past symbiotic relationship (P1, -O), (b) sharing of psychic heartachein a past psychologically burdening relationship (P2, -LL), and (c) taking each other's vulnerabilities into account [not hurting each other] in a tense relationship (P3, -O).Noticeably, the previous three texts (P1, P2, P3) not only display high anxiety scores, but also the highest security scores within this cluster, disclosing some sense of security and mutual connection in these past and present relationships.Other texts reveal a one-sided submissive interpersonal orientation, i.e., (a) selfeffacing by taking on the role of counselorin a past unequal partner relationship (P4, -LL), (b) pleasing the otherin a past depressing relationship with a dominant partner (P6, -LL), (c) taking the other into account a lotin a deteriorating old friendship (P7, -O), (d) and being dominatedin a past unequal relationship (P9, -LL).Considering the nature of this cluster, it is no coincidence that in some of the texts an explicit reference is being made to the submissive theme of 'rejection' (P5, P8, P10).However, a novel communal-submissive theme may be discerned in some of the texts -labelled 'forgivingness'-which is expressed in (a)   (f, 41) In the past, our relationship was best described as a suffocating friendship.The other wanted an exclusive friendship that completely consumed me, for fear that I would leave her.I had to focus on her and not pay attention to others.For example, I always had to answer app messages immediately.It made me feel trapped, confined and swallowed up.I felt like I couldn't escape and was being watched.Personally, I thought it was enough if we saw each other once every few weeks.Now the relationship is better.In the past, the relationship was often good when we saw each other, but because of her grip on me, I kept thinking about her when we were apart and I filled in the feelings she might have.For example, if I didn't respond to texts, I thought "she won't like that, because she'll feel abandoned".She did everything for me, was always there for me, but I often thought "leave me alone".I always felt like I was falling short as a friend.

Being claimed by the other
Taking the other person into account a lot Feeling guilty ("falling short" in the past) 2.20 3.00 0.00 2.75 3.00 0.33 S10 (f, 62) I had a difficult relationship with this person.I tried to build something together, but I noticed that he decided what was going to happen.After the first crush, I noticed that our life rhythms were different: I am a morning person while he was moody in the morning.When he blossomed in the evening, I was exhausted.It always took a lot of time to come to an agreement about what we wanted to do together.I felt cheated and restrained by him.He seemed very reliable but crossed boundaries, which was very shocking to me.

Constricting relationship with a dominant other
Boundaries being ignored by the other 2.10 2.00 2.50 1.25 1.25 0.33 Note: In the first column are the subject numbers S1 to S10, each provided with gender (m for male and f for female) and age; Columns 1 to 6 represent the attachment subscales: 1= Security, 2 = Preoccupation, 3 = Rejection, 4 = Defensive Separation, 5 = Avoidance, 6 = Threat.P2 (f, 49) When I think of [my ex-partner] I feel love and connection and at the same time an underlying anger because this relationship has not been able to continue.We had a relationship in which a lot was discussed, in which a lot of pain was suffered and in which there was also a lot of appreciation.In addition, [I did not] understand the [psychological problems] of the other person and therefore [I] did not have a clear view of my own "blind spots".I found our contact mainly confusing, it was attraction and repulsion.Not wanting to lose each other but also getting stuck all the time.I felt shortchanged by him.

Complex ambivalent relationship
Loss of an intimate love affair Resentment about the loss of the relationship

Sharing psychic heartache with each other
Being burdened by the psychological misery of the other (attracting and repelling) 2.20 4.00 4.50 1.75 1.50 2.00 P3 (f, 45) With this person I have a careful and tense relationship.We try not to hurt each other, like "two hedgehogs revolving around each other".Unfortunately, as a result we have not been able to build up a good relationship.With this person I feel warmth, tension, sometimes a lot of sadness, sometimes a lot of love and sometimes I feel disarmed.

Concern for each other
Taking each other's vulnerabilities into account 2.70 4.67 4.00 3.25 3.50 1.00 P4 (f, 43) The relationship with this person was unequal, it was a one-way street with difficult communication.I felt insecure, unloved and more a counselor than a partner.I still feel angry and cheated.
Anger 1.50 4.33 4.50 2.00 1.50 0.67 P5 (f, 48) With this person I feel good, seen, appreciated on the one hand and on the other hand insecure, rejected and not good enough.We have an unequal, unbalanced and non-transparent relationship with unclear, unspoken, evasive and capricious contact.I don't know what I'm getting out of this person and that makes me fidgety.

Feeling rejected
Being non-transparent/avoiding each other 1.90 3.67 3.50 2.50 2.25 3.00 P6 (m, 74) I am happy and feel satisfied with the decision to end the relationship with this person.I feel liberated from the daily pressure to please each other.At first I felt very happy with her, but I allowed myselfas always in my contact with womento "be dominated" [which] made me very insecure and depressed.We had a fairly superficial contact, were too different from each other and we did not dare to express deep feelings to each other.It seemed as if there was a blockade that stopped an intimate, growing connection.
(-LL: 1.33 1.83 2.00 3.17) Unbalanced relationship (daily pressure to please each other) Depressing relationship (with a dominant other) Lack of intimacy 1.50 3.67 2.50 2.75 3.75 2.67 Table 9 continues on next page Table 9 continued P7 (f, 35) I love this person but can also really be wary of the interaction.I understand her and at the same time I have enormous difficulties with the choices she makes.I like to be with her but actually often prefer not to be with her.This feels conflicted.I always felt at home with her, welcomed and enjoyable.Now I don't have those feelings anymore and I miss them.We have a complex relationship where there is a lot of adjustment and alignment in the interaction, mainly on my part.She hardly notices what her behavior and presence does in contact with others.In recent years she has become increasingly pessimistic and negative, while I am always doing quite well and I am and remain optimistic.
( P9 (m, 68) When I think of this person, I think of her with compassion [even though] I didn't feel safe with her.It was not possible to maintain a relationship with her.In the past, the interaction was reasonable, but she always wanted to dominate me.

Unequal relationship
Being dominated by the other

Feeling unsafe
Forgivingness towards the other person 1.80 3.00 2.50 3.25 2.25 1.00 P10 (f, *) With this person [mother] I have a complex relationship where I feel both intense sadness and intense love.For the first 16 years of my life, I felt unaccepted by her, not worth anything.In retrospect, it is clear to me that she loved me but did not communicate that to me.I now see that she did her best but that her own past determined her reactions.The relationship was always unpredictable: from very good periods to periods in which she pretended that I did not exist because she was angry since I did not meet her unspoken expectations.Over time, I have gained more and more insight about her, which made the relationship more loving.

Unpredictable contact
Forgivingness: a growing sense of understanding for the shortcomings of the other 2.10 2.00 3.75 3.25 3.00 1.00 Note: In the first column are the subject numbers P1 to P10, each provided with gender (m for male and f for female) and age; Columns 1 to 6 represent the attachment subscales: 1= Security, 2 = Preoccupation, 3 = Rejection, 4 = Defensive Separation, 5 = Avoidance, 6 = Threat; * = age unknown  (f,51) With this person I have had a short-lived relationship in the past, in which I initially felt catered to my every whim.But the relationship did not go well and was difficult and dependent on his will to contact me.I now feel ignored and used by him.He could not cope well with the breakup and did not contact me again.Then there was another "revival" and then the same thing happened again.Now I have finally put an end to it.

Short termed ambivalent love affair
One-sided depended relationship Feeling ignored and used 1.00 1.33 4.25 4.50 2.50 1.67 D5 (f, 56) The relationship with my sister is best described as tense.I don't feel "sisterly love", at most compassion for the burden she lives with, which she doesn't want to face.I feel powerless to change anything substantially and that makes me restless.The relationship used to be bad, I had the feeling that she was waltzing over me.At the moment, our contact is reasonable, mainly because I am more assertive now and have more control, as a result of which she has less control over my feelings.Still, I sometimes feel like I'm walking on eggshells because my sister's mood can change quickly.

Feeling helpless, powerless
Sympathy for the problems of the other Contact used to be bad: the other dominated me Being wary ("the feeling of walking on egg shells") 0.70 1.67 2.50 4.00 2.75 1.33 D6 (f, 63) With my sister I have a difficult but loving and good relationship.I feel wary of her and don't say everything out loud because I don't want her "well-intentioned advice."I feel a lot of irritation that I do not want and dare to express, which costs me a lot of energy.The irritations that my sister arouses in me are so great that I hold off to contact her, which often leads to postponement.I feel guilty that I don't contact her more often.

Not expressing emotions
Feelings of guilt (about the avoidance of contact) 3.00 1.00 0.50 4.50 4.00 0.33  (m, 67) The relationship I used to have with my wife is best described as a disguised father-daughter relationship, in which I was the all-forgiving father and she was the dependent daughter.I feel like she was never able to deal with the loss of her beloved father at a young age.In our relationship she was needy, dependent but also loving, while also exhibiting alcoholic behavior.I often exhibited bossy behavior, sometimes using seduction and manipulation to get things done.Our contact was strongly influenced by my own stress and business success.After the separation, the contact was limited to practical matters.Nowadays, there is also room, based on compassion, for more openness about emotional matters from the past.
(   Inferiority ("to feel small like a child") 1.70 4.00 4.50 3.25 3.75 3.00 Note: In the first column are the subject numbers F1 to F10, each provided with gender (m for male and f for female) and age; Columns 1 to 6 represent the attachment subscales: 1 = Security, 2 = Preoccupation, 3 = Rejection, 4 = Defensive Separation, 5 = Avoidance, 6 = Threat not being angry when thinking about being rejected by an ex-partner (P8, -O), (b) thinking with compassion about a past relationship with an unreliable dominant partner (P9, -LL), and (c) a growing sense of understanding for the shortcomings of the otherin a complex mother-daughter relationship (P10, -O).Many texts of the dismissive-avoidant cluster (see Table 10) reveal a problematic tense relationship with a relative (parent or sibling) or (ex-)partner, with whom in some cases there is no longer any contact (D2, D4, D8, D9) or contact possible (D3).Most texts show quite high scores on defensive separation/denial of attachment needs (cf. the defining feature of this cluster), but the scores on avoidance/alineation are also elevated for most of the texts (above midpoint: > 2.5) and a few texts display elevated scores on rejection or exceptional high scores on threat/abuse.In correspondence with the nature of this cluster, several texts reveal aspects in which a person is defensively (behaviorally) keeping a distance in an actual relationship, i.e., (a) being wary in the obligatory contact with a family member (D1, -S), (b) keeping control and walking on egg shells in a tense relationship with a sister (D5, -LL), (c) avoiding contact with a sister that evokes irritation (with her advice)" (D6, +O), (d) being on guard and in survival mode in the contact with a transgressive and dominant person (D7, -LL); Notice that these relationships are still continuing.A related aspect is becoming (emotionally) withdrawn as expressed in (a) absence of connectedness ("closing my heart") in a past unsafe relationship with a father (D2, -LL), (b) not sharing ones experiences ("stopped telling things") to an unresponsive manipulative father (D3, -LL), and (c) absence of an intimate connection (i.e., "pain of not being seen" and not being allowed to show true emotions) in a reversed mother-daughter relationship (D10, -O).It is striking that in the dismissive-avoidant texts with high scores on 'threat' an explicit reference is being made to 'unsafety', showing how distressing the contact is/has been with these parents/adults: (a) "I didn't feel safe [with him], which still hurts.He was my father" (D2), (b) "I have an unsafe and difficult relationship with [her], (…) if she's not feeling well, I have to endure it" (D7), (c) "With [her] I felt loved and unsafe.(…) I blame her for protecting herself as an adult at my expense" (D9).
The ten texts that we selected from the fearful-avoidant cluster (see Table 11) predominantly expose unsafe past relationships with a mother (F1, F3, F8) or an unspecified relative/acquaintance (F5, F6, F9, F10); Only three texts concern an actual relationship with a relative (F2) or unspecified acquaintance (F4, F7).Most texts display extremely high scores for the five insecure attachment scales (cf. the features of this cluster), obviously emphasizing the arduous nature of these relationships; Affectively, all texts refer to powerlessness and isolation (-LL).In our view, most texts within this cluster expose psychologically (severely) damaging relationships, articulating experiences of humiliation, inferiority or profound rejection (F1, F3, F4, F5, F6, F8, F9, F10).This is clearly visible in the texts which refer to: (a) feeling depressed in a long-term derogatory contact with a manipulative mother ("I was never good enough") (F1), (b) feeling miserable in a "deadly" contact with a humiliating and rejecting mother ("… empty and thrown away, ... unwanted, with no right to exist") (F3), feeling inferior in an alienating relationship ("never felt good enough") (F4), feeling sad and powerless anger in an edgy contact with a dominant constricting other ("felt unseen as to who I am") (F5), feeling inferior in a depressing tense relationship (F6), feeling rejected and alienated in a cold detached relationship with a mother ("She never said anything positive about me") (F8), feeling humiliated in a degrading contact with a disrespectful other ("she tried to convince me that I was no good") (F9), and feeling inferior in an alienating relationship ("feeling small as a child") (F10).In two other texts (F2, F7) the contact is not that explicitly described as being harmful or transgressive, but also these texts clearly reflect a disturbed contact, i.e., (a) unsafety and restrained anger in a constricting relationship ("I remain silent") (F2) and (b) distrust in an uncomfortable edgy relationship ("there is a lot of dishonesty") (F7).Furthermore, in some of the texts there are signs of physical reactions, viz., (a) "tension in my chest and clench my jaws" (F2), (b) "pain in my belly … my stomach is being squeezed" (F6), or "feeling sweaty" (F10)-which may be an indication of how taxing these relationships can be.

DISCUSSION
The primary aim of this study was to develop attachment scales that can be applied idiographically in a clinical counseling context.The initial results of the exploratory factor analyses, revealed clues of a four-dimensional model reflecting the classical attachment prototypes, but by employing a mix of exploratory and confirmatory techniques, we also discovered facets/subfactors within some of the higherorder factors.Ultimately, we constructed six scales, in which the traces of the four prototypes were reasonably well visible.
In order to explore the interpersonal and affective meaning of the ARIA-scales, we performed two supplementary validation studies using instruments that accentuate agentic and communal interpersonal orientations.The circumplex shows that, consistent with our expectations, Security is interpersonally characterized by 'agentic-communion', Defensive Separation by 'separation' and Avoidance/Alienation by 'submissive-separation'. Contrary to our expectations, the two anxiety scales (Preoccupation and Rejection), were both more strongly associated with 'submission' than with 'submissive-communion'.Threat was also projected in the lower 'submissive' part of the circumplex, disclosing a stronger interpersonal similarity with the two anxiety subscales than anticipated.When looking at the configuration of scales as a whole, it resembles an upside down 'kite' of which the vertices correspond to the four attachment prototypes.In summary, this configuration supports the construct validity to some degree, visualizing that most ARIA-scales pertaining to the secure, preoccupied, dismissive-avoidant and fearful-avoidant prototypes are interpersonally characterized by, respectively, agentic-connection, submission, cold-disconnection, and submissive-disconnection.The hexagon reveals a quite different picture, whereby a contrast between Security (upper right) and a compact cluster of four insecure scales (bottom left) catches the eye.As anticipated, Security is characterized by a mixture of 'unity and strength' (+HH) and 'unity and love' (+O), and Preoccupation by 'un-fulfilled longing and loss' (-O).However, the compact grouping of Defensive Separation (cf.dismissive-avoidant prototype), Avoidance/Alienation and Threat (cf.fearful avoidant prototype) and Rejection (theoretically, cf.preoccupied and fearful-avoidant prototype) in the bottom left of the hexagon signifies that these scales are affectively characterized by 'powerlessness and isolation' (-LL).Hence, it seems that the differentiation in four prototypes is more transparently reflected by the interpersonal circumplex than by the hexagon.
With the aid of cluster analysis we constructed (heterogeneous) groups representing four attachment prototypes (of texts).The profiles of mean scores on the attachment scales indicate that these groups can be efficiently discriminated from each other.First of all, all six attachment scales are effective for distinguishing the secure cases from those in the other three clusters.Furthermore, both the preoccupied and dismissive-avoidant cases can sharply be delineated from the fearful-avoidant cases by taking the scores on Avoidance/Alienation, Threat, and (also) Rejection into account.Finally, the Preoccupation and Defensive Separation scales seem particularly useful for the discrimination of cases in the preoccupied cluster from those in the dismissive-avoidant cluster.Noticeably, the findings of the cluster analysis show that Rejection cannot be interpreted in terms of a (simple) attachment anxiety facet (characteristic of the preoccupied and fearful-avoidant prototype), but instead seems to be a distinctive feature of texts assigned to the fearful-avoidant prototype.
By studying a selection of texts of the four attachment prototypes, we gained an impression of the kind of problems that people were experiencing in (problematic) actual relationships.The themes that emerged were not only characteristic for the four prototypes-which supports the content validity of the ARIA-scales-but we also found novel interpersonal themes for some of the prototypes.The texts from the secure cluster illustrate that even in troublesome ambiguous relationships some (mutual) sense of connection can exist.Several persons mentioned that they sometimes experience/experienced the relationship as loving (S1, S2, S3, S7, S8), referring to mutual respect (S1), comfort and acceptance (S4), a joyful valuable relationship (S8), empathy for a vulnerable other (S3, S6) and consideration for the other (S9).In some texts the longevity of the relationship is mentioned (S4, S5, S7), which may be a way of expressing how valuable the contact has been.In the texts of the preoccupied cluster we discovered characteristic themes referring to a strong/excessive mutual consideration for each other (P1, P2, P3), a (friendly) submissive interpersonal orientation (P4, P6, P7, P9), and rejection (P5, P8, P10).Friendlysubmission was also apparent in texts expressing an exceptional mildness towards a person that has caused suffering, cf.forgivingness (P8, P9, P10).In some of the texts from the dismissive-avoidant cluster we detected clear signs of a disconnected interpersonal orientation, i.e., defensively keeping a distance/control (D1, D5, D6, D7) and an emotionally withdrawn attitude (D2, D3, D10).In the texts of the fearful-avoidant cluster there were also signs of disconnection (from the other), yet the relationships described were much more harmful and transgressive.Most texts within this cluster express profound disruptive experiences in which one feels inferior or feels alienated from oneself (cf.dissociation).Hence, it seems that the experience of being fundamentally disrespected (by the other) and feeling inferior and insignificant are distinctive features of a fearfulavoidant relationship.
In our view, the ARIA-scales can be useful in a narrative psychotherapeutic context in which a client is reflecting upon interpersonal and intimate relationships in depth.By connecting the attachment items to accurate descriptions of experiences with a number of significant others -preferably obtained in a collaborative dialogical relationship (cf.coconstruction in the SCM)-an idiographic assessment of attachment relationships can be obtained.Such a systematic investigation of relationships may affect and enrich the communication between counselor and client in several ways.Foremost, it places the discussion of the client's personal relationships explicitly within an attachment theoretical perspective.Hence, some explanation of the basic concepts and assumptions of attachment theory to a client is required, preferably in the phase where the numerical results of the assessment are being discussed.To ensure that a client can properly understand the results, the therapist may highlight the following issues: (a) Attachment (in adolescence or adulthood) can be described as how one experiences relationships with significant others when being distressed or vulnerable.Secure attachment is related to trust in others, experiencing intimacy, being accepted/respected and receiving support from others.We feel insecurely attached when we are anxious about being rejected/abandoned (not being accepted) or when we avoid intimacy/closeness and deny the need for support; (b) Early childhood experiences may resound throughout life and unknowingly influence our affective relationships in adulthood; (c) There are times and situations where we may need support and comfort from others; dependency need not be pathologized.Dependency and autonomy are two sides of the same attachment coin (Dallos & Vetere, 2014); (d) When we are seriously distressed and finding support from others (cf.primary strategy of proximity seeking) doesn't work well, we may either intensify the seeking for support in a clingy manner (cf.hyperactivating strategy) or downplay the need for support and act independently (cf.deactivating strategy); (e) When we have had traumatic experiences with caregivers (i.e., who were alternatingly caring, neglecting, frightening), it may seem that we have no sense of recollection and have difficulty to put these painful experiences into words (cf.dissociation); (f) Four theoretical attachment prototypes can be distinguished, which can be recognized by characteristic interpersonal themes in the descriptions of our (intimate, friendship and work) relationships; (g) Attachment-related issues (e.g., fear of rejection, discomfort with intimacy, dissociation) are likely to emerge in the here-and-now interaction with the therapist (cf.transference and countertransference) (Wallin, 2007).
An idiographic assessment of attachment relationships also has implications for the nature of the psychological intervention, in that the overview of attachment themes/issues (as described in real-life relationships) may reveal a dominant attachment orientation and dysfunctional aspects of the 'narrative attachment system'.It is obvious that some sort of numerical classification scheme should be developed with which profiles can efficiently be allocated to one of four attachment prototypes (e.g., by using centroids in a non-hierarchical cluster analysis, an equation from a discriminant function analysis, or some other method).The idea is that the frequencies of prototypes reveal a dominant attachment orientation, which may offer clues for fine-tuning the goals in psychotherapy (together with the client).For example, a person with a preponderance of fearful-avoidant relationships (in past and present) asks for a much more delicate approach than someone with several secure and only a few fearful-avoidant relationships.Dependent on the dominant insecure attachment orientation, a variety of issues may emerge, such as: (a) preoccupied themes (e.g., excessive concern for others, excessive fear of rejection/abandonment, difficulty with being alone); (b) dismissing-avoidant themes (e.g., denial of attachment needs, distancing from others, avoidance of closeness, need for control/independence); or (c) fearful-avoidant themes (e.g., distrust in others, fear and avoidance of intimacy, suppression of emotions, dissociation) (cf.Van Geel et al., 2019;Van Geel & Houtmans, 2022).Detailed guidelines about how these problems can be addressed in psychotherapy, is beyond the scope of this article.The clinical techniques and interpersonal skills that are needed for being a sensitive and responsive therapist, with the allied challenges related to transference and negative countertransference, are outlined in detail in more clinically oriented texts (Beier & Young, 1984;Dallos, 2006;Mikulincer & Shaver, 2007, Chapter 14;Wallin, 2007).Another distinct advantage of a narrative idiographic approach is that it may direct the communication to specific issues of a client's interpersonal life.For example, a client may discover that the attachment profile of the relationship with an intimidating superior is very similar to that of a (deceased) neglecting alcoholic mother, which may be a reason to find out more specifically why the relationship with the supervisor is experienced in this way.The client may eventually come up with a text that explains the similarity ("I am disgusted by dominant people, like my mother") to which the therapist-who is familiar with the dysfunctional aspects of self-narratives, knowing that such a rigid perspective isn't going to change the relationship with his boss-may respond with an invitation to think in more constructive ways for improving this contact (e.g., mediation).However, it might be necessary to first resolve the traumatic experiences of his youth before he can think of improving the contact with his superior.This fictional example illustrates how the interpersonal similarity between two different relationships can trigger further discussion and shape therapeutic intervention.To conclude, we think that attachment theory in an idiographic narrative format may provide clients with a fresh perspective on their relationships.The thematic overview according to the attachment typology may increase their awareness of attachment issues and associated (dysfunctional) emotional reactions, which can be helpful when trying to construct a coherent narrative of (pleasant and painful) experiences with others (Goodman, 2018;Van der Zande, 2022).
When considering the validity and utility of an attachment measure, it seems relevant to question and check whether all theoretical aspects of the attachment prototypes are sufficiently represented.Regarding the ARIA-instrument, secure attachment appears sufficiently represented, even though it may be possible to distinguish multiple facets within this domain (i.e., trust and intimacy; mutual respect and inspiring contact; being able to be myself, cf.Van Geel et al. 2011).Furthermore, it seems that more items/aspects are needed to cover the domain of preoccupied attachment more fully-Items that might be eligible are: longing for a good relationship, being dominated by the other, not being accepted, being very dependent on the other, showing much consideration for the other, self-effacing, and displaying a forgiving attitude (See Table 9).The dismissive-avoidant aspect seems sufficiently represented by the items referring to the denial of attachment needs (the core aspect of this prototype).However, the measurement of dismissive-avoidance might be broadened by including items that refer directly to defensive separation (shutting out the other, keeping the other at a distance), imply low empathy (not empathize with the other, show little interest in the other) or denote low self-disclosure (avoid intimate subjects, not disclosing much about myself) (For empirical support of this idea, see Grasman, 2023;Steunenberg, 2023).The content validity of the threatening aspect of fearful-avoidance (Factor IVb, See Table 1) might be improved by including some extra items (e.g., feeling belittled, being humiliated, feeling inferior, See Table 11).All in all, it seems a worthwhile idea to distinguish multiple facets within each of the four prototypes and to develop a (two-dimensional) circumplex model of attachment scales in which (eight) scales are evenly distributed on a circle.
Limitations of this study refer to self-report biases and response sets and a possible self-selection bias due to the voluntary nature of respondent participation.The response group consisted predominantly of a (nonclinical) group of middle-aged women, the few male participants being noticeably older.The factor structure which resulted from the data, therefore, may not generalize to the Dutch population.Some caution is also warranted, as considerable steps were needed (in which 21 items were eliminated) to reach an acceptable model fit.Hence, the factor structure requires replication in other samples.Another methodological limitation is that participants were asked to describe only "one" (troublesome) relationship, without being able to provide a more differentiated and nuanced picture of their interpersonal life.Moreover, participants were describing their relationships in an online survey without the possibility to fine-tune their descriptions in a dialogue with someone who listens attentively.Consequently, future research may benefit from gathering data in adult groups of younger age who are invited to discuss and describe several relationships in a dialogue with an experienced psychologist (cf.SCM procedure).We believe that the ARIA-28 scales provide a useful starting point for the further development of idiographic attachment scales that can be applied in a therapeutic context.

(
In the first column are the subject numbers D1 to D10, each provided with gender (m for male and f for female) and age; Columns 1 to 6 represent the attachment subscales: 1 = Security, 2 = Preoccupation, 3 = Rejection, 4 = Defensive Separation, 5 = Avoidance, 6 = Threat

Table 1 .
Standardized loadings of the Six-Factor confirmatory solution (ARIA-28, N = 106) Note:The numbers correspond with those in the complete ARIA-49 research list, see Appendix acceptable, a more sophisticated confirmatory technique is needed for a closer examination of the number and composition of (sub)factors in this domain (see next paragraph).

Table 8 .
Texts of ten persons allocated to the "secure" cluster We have a friendly relationship with both the "wish for more", but we both do not dare to go for it yet.I want to "focus my senses on him" [closeness and intimacy], and at the same time I keep my distance.We have daily in-depth contact, which is totally new to me.I feel safe with this person until we discuss our "difficult relationship", then I get totally confused and I don't know what to do with him and myself anymore.The (40-year-old) friendship with my girlfriend is difficult, in the beginning it was fun, then bad and now neutral.When I think of her, I feel resistance because she [admittedly] seems very socially moved, but [in essence] she isat all costs and despite everyonemainly concerned with her own well-being, How I think about her is a mirror for myself.With [my partner] I "emotionally get into a split": I [love] him, but I am also (too) annoyed by him.In the past, our contact was intense and passionate, now it is superficial.I felt very loved.Our relationship is a very loyal love affair with an open character.I see the beauty of the relationship but still have mixed feelings.It is difficult to go through life together.The contact consists of ups, in which the relationship seems very solid, and downs where we navigate dark valleys together.I experience many different emotions in this relationship.Sometimes I am despondent and want to end the relationship, at other times I find joy in this intensely valuable relationship.
S3 (f, 44)Thinking about this person, I feel my belly tighten slightly and at the same time I also feel myself becoming softer.We have a love affair in which we both don't know how to deal with each other's "wounded layers"[vulnerabilities].Our contact is a great adventurepainful and (fortunately also) softwhere I experience many different feelings: uncertain and threatened versus safe, relaxed, [accepted], and flourishing.Miraculously, there is also an opportunity for love.

Table 9 .
Texts of ten persons allocated to the "preoccupied" cluster With this person, I had an immature open "total" relationship, in which we were dependent [on each other] and were focused on each other (not autonomous) "holding nothing back"-looking for symbiosis and sexual confirmation.We were honest, confrontational, loving and devoted to each other.I have felt seen, needy, loved, accepted, and used.
The relationship with my [ex-partner] was capricious.It didn't start with intense [loving] passion but we did have impetuous moments and I had strong feelings for that person.I sometimes felt very cozy, sometimes just comfortable, sometimes lustful, sometimes far away or rejected by her.She has done terrible things (I and our son weren't a priority for her) and completely talking things out didn't work.Still, I'm not angry with the other person anymore, [because we also had good times together].

Table 10 .
Texts of ten persons allocated to the "dismissive-avoidant" cluster When I think of this person, "my neck hairs rise up".I feel wary of this person, [because] at any moment a new incident can occur.The connection with this person was disrupted early on, [because it turned out that] we are two completely different people.We are biologically [as a family] obliged to contact each other and make appointments to see each other.When I think of [my father], I feel defensive and "close my heart."Our relationship doesn't really exist because it was never there.There was little contact and if there was contact, it was reactive.I didn't feel safe [with him], which still hurts."He was my father".The relationship with my father was unequal, unpleasant and sometimes downright frustrating.I felt tense, frustrated and unheard.My father demanded attention and contact from me and used emotional blackmail to achieve that.Subsequently he was not interested in me at all, but only in himself.He never listened to what I had to say.Eventually, I stopped telling him things.Now that he's dead, I sometimes feel guilty that I don't miss him.

Table 10
continues next pageTable 10 continued D7 (f, 59) I have an unsafe and difficult relationship with this person.If she's not feeling well, I have to endure it, which causes anxiety, anger, and a lot of stress [for me].She is very dominant, cannot look at herself (projects problems onto others), and tries to influence my life.If she doesn't get her way, she gets angry and tries to blackmail me.I used to be [indulgent], [but] now I'm not indulgent anymore and then a fight ensues.[Only] when she is in a good mood, I can relax [somewhat], but [also] always remain on my guard.Contact with her "feels like survival to me".I feel disrespected in who I really am and retreat to my own territory.
With this person, I felt loved and unsafe.The relationship was complicated and conflicted, because[it]was both loving and difficult.I understand [now] how difficult her life has been, but I blame her for protecting herself as an adult at my expense.I find this hard to forgive.We have a mother-daughter relationship in which I fulfill (and have always fulfilled) the mother role.There is no personal intimate connection between us and no open communication.I feel the pain of not being seen and not getting what I wanted from her.I now know that she can't give it to me because she never had it herself.That makes me angry.My own pain and repressed anger at times come to the surface, even though I understand what's going on.I have the idea that my mother will die if I don't take care of her, even though I rationally know that's not true.With my mother I had the feeling that I was not allowed to show my true feelings, because she cannot react to them with warmth or reactions and that hurts.I tend to be strong and tough, disregarding myself and support her at the expense of myself.There is no place for my feelings.

Table 11 .
Texts of ten persons allocated to the "fearful-avoidant" cluster This person has failed me from birth until now.I tried to adapt, but it was never good enough.I had to walk on my toes because she was manipulating and everything was really about her.It was better not to show emotions, because that was always made personal and then she was hurt.My confidence has suffered and I have long thought that I was a bad and nasty person.I felt alone, lonely, not heard and not important.I am disappointed in her and I have hated her.I'm still disgusted with her and can't stand her around me. [When I see her] I get restless and irritated.We are not in contact anymore.I did contact her out of guilt, but I regret doing that now.With this person I feel unsafe, anxious and angry: I feel tension in my chest and clench my jaws and feel anger rise.We have a difficult relationship, in which we are obliged to see each other because this is important to others."I never want to see you again, but I can't do that."[During the contact] I remain silent, the tension rises and [afterwards] I spend days thinking about everything that has been said that I do not agree with.With this person I had a "deadly" relationship that you can't really call a relationship.It all originated from one side, I gave life, the other chose death.This person humiliated me to feel better herself, but continued feeling bad.The contact was complex and ambiguous, with recriminations and even death threats against me.I often sought contact which ended in disappointment for me.I felt miserable, empty and thrown away with this person "like a doll that was once interesting, but is now gathering dust in the closet"unwanted, with no right to exist.For this person I best never should have existed orrather -I better should "never have grown up".Being a little baby things were adequate, since a baby can't say anything back.Our relationship is no longer there, and actually it was never there.When I think of this person, I get very sad and angry; I [still] get pain in my belly, it feels like my stomach is being squeezed.With this person I have had a very tense and difficult relationship (now broken).We had no trust in each other and [I was] always vigilant to what was happening and being said.I felt inferior to this person.When I think of my mother, I keep wondering why she has always approached me in a negative way.She never said anything positive about me.Is it from jealousy, or is there more to it?We were strangers to each other.Although our relationship was very cold, detached and frigid and I felt very bad with her, I nevertheless tried to improve the contact.I feel rushed, sweaty and small as a child with this person.In this apparently good relationship, there is a lot going on under the surface.While I'm trying to be polite, she can be annoying or annoyed.I feel alienated from myself [therefore] when I am with her.Even though I agree with myself [in advance] that I will approach her as I really am, after a few hours [contact with her] I find myself again starting to alienate.And then I can't turn this around anymore.She always looks at me as being a child who still has a lot to learn and we always have to dig deeper, she thinks.But I just want to be able to be myself.