Adult Attachment

Abstract

The Adult Attachment assessment—frequently operationalized through the Relationship Scales Questionnaire (RSQ) developed by Kim Bartholomew and Leonard M. Horowitz (1991)—is an established psychometric instrument designed to evaluate individual differences in adult interpersonal orientation and internal working models of self and others. Rooted in foundational attachment theory formulated by John Bowlby and extended to adulthood by Cindy Hazan and Phillip Shaver, this 19-item self-report questionnaire systematically maps adult relational patterns onto Bartholomew and Horowitz’s four-category typology: Secure, Preoccupied, Fearful, and Dismissing. Respondents rate each item on a 7-point Likert-type scale ranging from 1 (“Strongly Disagree”) to 7 (“Strongly Agree”), with an additional categorical option of 9 (“Don’t Know”).

Psychometrically, the instrument captures two fundamental underlying orthogonal continuous dimensions: Attachment Anxiety (Model of Self, reflecting anxiety concerning unworthiness, abandonment, and rejection) and Attachment Avoidance (Model of Other, reflecting defensive suppression of intimacy, interpersonal distrust, and compulsive self-reliance). Across extensive clinical, developmental, and social psychological research spanning three decades, the subscales have demonstrated adequate-to-robust internal consistency, with Cronbach’s alpha coefficients typically ranging from .68 to .85 across diverse demographic cohorts. Test-retest reliability across intervals ranging from two weeks to eight months exhibits moderate-to-high temporal stability (r = .60 to .78), establishing that while adult attachment is subject to life-transition revisions, core cognitive schemas maintain substantial dispositional consistency. Structural equation modeling and confirmatory factor analyses corroborate the dual-axis dimensional organization, affirming the construct validity, convergent validity with peer-report ratings and clinical attachment interviews, and discriminant validity against general neuroticism and extraversion. The Adult Attachment scale remains an essential, highly adaptable assessment tool for relationship research, clinical diagnostic formulation, psychotherapy process monitoring, and social neuroscience investigations.

Keywords

Adult Attachment, Relationship Scales Questionnaire, RSQ, Internal Working Models, Attachment Anxiety, Attachment Avoidance, Bartholomew and Horowitz, Attachment Styles, Psychometrics, Interpersonal Relationships

Authors

The conceptual framework, prototypical operationalization, and empirical foundation of the Adult Attachment four-category taxonomy and its affiliated questionnaire methodology were authored by:

• Kim Bartholomew, Ph.D.: Professor Emerita of Psychology, Department of Psychology, Simon Fraser University, Burnaby, British Columbia, Canada. Dr. Bartholomew’s seminal scholarship pioneered the systematic conceptualization of adult attachment via internal working models of self and other.
• Leonard M. Horowitz, Ph.D.: Professor Emeritus of Psychology, Department of Psychology, Stanford University, Stanford, California, United States. Dr. Horowitz is renowned for his fundamental contributions to interpersonal theory, interpersonal problems (Inventory of Interpersonal Problems), and psychopathology.

Purpose

The primary purpose of the Adult Attachment scale (Bartholomew & Horowitz version) is to operationalize, quantify, and categorize the complex psychological representations, emotional expectations, and behavioral strategies that adults employ within close interpersonal and romantic relationships. While early attachment researchers predominantly investigated infant-caregiver bonds or utilized Hazan and Shaver’s initial three-category forced-choice paragraphs, clinical and empirical scientists required a nuanced, multidimensional measurement tool capable of capturing avoidant defenses that manifest in divergent behavioral topologies.

In clinical contexts, the instrument serves several pivotal diagnostic and prognostic functions:

• Case Conceptualization: It enables psychotherapists to identify patients’ baseline internal working models, illuminating how relational trauma, emotional neglect, or inconsistent parental attunement manifest as defensive distancing or frantic bids for validation.
• Therapeutic Alliance Forecasting: Individuals exhibiting high fearful or dismissive attachment often struggle with therapeutic alliance formation, premature drop-out, or transference ruptures, whereas those with preoccupied attachment may present with acute boundary-testing or dependence.
• Treatment Customization: By distinguishing between fearful avoidance (motivated by fear of rejection) and dismissive avoidance (motivated by defensive self-reliance and emotional deactivation), clinicians can tailor interventions—such as Emotion-Focused Therapy (EFT), Mentalization-Based Treatment (MBT), or schema-focused cognitive therapies—to the patient’s specific emotional regulation strategy.

In academic and experimental research, the Adult Attachment scale serves as an indispensable vehicle for examining the interplay between relational orientations and a vast array of psychological outcomes, including dyadic conflict resolution, sexual satisfaction, stress reactivity, physiological cortisol regulation, social anxiety, and resilience against major depressive disorder. Its theoretical rationale rests on Bowlby’s proposition that childhood relational experiences crystallize into generalized relational expectations, which dynamically influence social perception, attributional biases, and affective regulation throughout adulthood.

Psychological Construct

The psychological construct evaluated by this instrument is adult attachment style, operationalized through Bartholomew and Horowitz’s two-dimensional, four-category paradigm. The model defines attachment styles along two orthogonal axes: the Model of Self (which correlates with attachment anxiety) and the Model of Other (which correlates with attachment avoidance).

1. Model of Self (Attachment Anxiety / Dependence)

The internal working model of the self reflects the degree to which an individual internalizes a sense of intrinsic self-worth, lovability, and emotional security versus pervasive self-doubt, hypervigilance toward abandonment, and fear of unworthiness. A negative model of self generates high attachment anxiety, characterized by cognitive preoccupation with relationship maintenance, compulsive reassurance-seeking, and heightened emotional reactivity during interpersonal distress.

2. Model of Other (Attachment Avoidance / Closeness-Seeking)

The internal working model of others assesses the degree to which an individual perceives relational partners, peers, and attachment figures as trustworthy, dependable, emotionally accessible, and benevolent versus rejecting, untrustworthy, and intrusive. A negative model of other fosters attachment avoidance, prompting strategies of emotional detachment, defensive suppression of vulnerability, minimization of intimacy needs, and rigid self-sufficiency.

The Resulting Four-Category Attachment Typology

The orthogonal intersection of these two dimensions delineates four distinct prototypical attachment styles:

• Secure Attachment (Positive Self, Positive Other): Characterized by high self-esteem, self-efficacy, and a natural comfort with interpersonal intimacy and mutual dependence. Secure adults easily navigate closeness, experience low anxiety regarding abandonment, and express emotions constructively. In the 19-item version, this is captured by items such as “I am comfortable depending on others” and “I know that others will be there when I need them.”
• Preoccupied Attachment (Negative Self, Positive Other): Defined by intense yearning for emotional merger and total intimacy, paired with chronic anxiety regarding partner availability and fear of abandonment. These individuals define their worth primarily through external validation. Prototypical items include: “I find others are reluctant to get as close as I would like” and “My desire to merge sometimes scares people away.”
• Fearful-Avoidant Attachment (Negative Self, Negative Other): Marked by a simultaneous yearning for emotional intimacy and profound terror of interpersonal vulnerability. Individuals with this style view themselves as unlovable and others as untrustworthy, resulting in agonizing conflict between closeness and self-protection. Representative items include: “I worry that I will be hurt if I allow myself to become too close to others” and “I want emotionally close relationships‚ but I find it difficult to trust others completely.”
• Dismissing-Avoidant Attachment (Positive Self, Negative Other): Characterized by defensive self-reliance, deactivation of attachment needs, and explicit minimization of the importance of close emotional bonds. While the self is ostensibly viewed as strong and autonomous, others are dismissed as untrustworthy or unnecessary. This is embodied by items such as: “I find it difficult to allow myself to depend on others” and “People are never there when you need them.”

Theoretical Framework

The theoretical architecture of the Adult Attachment scale is grounded in John Bowlby‘s (1969, 1973, 1980) evolutionary-ethological attachment theory and Mary Ainsworth‘s (1978) empirical infant-mother attachment paradigms (the “Strange Situation”). Bowlby posited that human infants possess an innate motivational system evolved to ensure proximity to protective caregivers (attachment figures). Repeated micro-interactions with these caregivers become mentally represented as generalized “internal working models”—cognitive-affective schema complexes comprising expectations, perceptual filters, and declarative/procedural scripts regarding self-worth and caregiver responsiveness.

In 1987, Cindy Hazan and Phillip Shaver revolutionized social psychology by demonstrating that these infantile attachment orientations extend into adult romantic love, defining three styles: secure, anxious/ambivalent, and avoidant. However, Bartholomew (1990) and Bartholomew and Horowitz (1991) identified a critical theoretical conflation within Hazan and Shaver’s conceptualization of avoidance: it collapsed individuals who actively avoid intimacy due to fear of rejection with those who avoid intimacy because they genuinely disavow its value.

To resolve this ambiguity, Bartholomew drew directly upon Bowlby’s original assertion that internal working models consist of two distinct, interacting components: a Model of the Self and a Model of the Attachment Figure (Other). By crossing the valence of these two continuous dimensions (+/- Self and +/- Other), Bartholomew formalized a rigorous 2 × 2 matrix generating four distinct adult attachment profiles:

This formulation seamlessly accounts for the dual-level nature of human defenses. The secure individual operates via “primary attachment strategies,” freely seeking support and expressing vulnerability. The preoccupied individual operates via “hyperactivating strategies,” amplifying emotional distress to compel proximity. The dismissing individual uses “deactivating strategies,” suppressing distress and denying attachment needs. Finally, the fearful individual experiences chronic distress resulting from conflict between hyperactivation and deactivation, experiencing desire for closeness alongside deep dread of vulnerability.

Validity

The validity of the Bartholomew and Horowitz Adult Attachment framework and its corresponding item pool has been exhaustively demonstrated through construct, convergent, discriminant, and predictive validation studies across four decades.

Construct and Structural Validity

Construct validity was initially verified by Bartholomew and Horowitz (1991) across two distinct multi-method studies employing semi-structured attachment interviews, peer reports, friend ratings, self-report questionnaires, and interpersonal problem inventories. The internal working models extracted from self-reports converged substantially with blind clinical interview ratings (correlations ranging from r = .40 to .56, p < .001). Individuals categorized as Secure demonstrated superior self-esteem, healthy coherence, and constructive social engagement. Preoccupied participants exhibited high emotional expressiveness, intense crying behavior, and hyper-dependence. Dismissing individuals exhibited high self-confidence paired with low interpersonal warmth and low emotional disclosure. Fearful individuals demonstrated marked social insecurity, low self-esteem, and pervasive interpersonal distrust.

Convergent Validity

The scale correlates robustly with alternative gold-standard attachment measures, such as the Experiences in Close Relationships (ECR and ECR-R; Brennan, Clark, & Shaver, 1998) inventory. The Anxiety dimension derived from the RSQ correlates between .65 and .78 with the ECR Anxiety scale, while the Avoidance dimension correlates between .60 and .74 with the ECR Avoidance scale. Furthermore, the measure demonstrates substantial convergence with measures of romantic jealousy, rejection sensitivity, and intimacy motivation.

Discriminant Validity

Discriminant validity has been demonstrated by showing that the two underlying attachment dimensions are empirically distinct from general Big Five personality traits. While attachment anxiety correlates moderately with Neuroticism (r ≈ .35 to .45), it accounts for substantial unique variance in relational outcomes after controlling for general neuroticism. Similarly, attachment avoidance correlates negatively with Extraversion and Agreeableness (r ≈ -.25 to -.38), but uniquely predicts defensive deactivation, physiological suppression under attachment threats, and lower dyadic responsiveness.

Predictive and Criterion Validity

Predictive validity has been substantiated across a broad spectrum of real-world outcomes. Longitudinal investigations establish that baseline fearful and preoccupied scores predict heightened marital dissatisfaction, domestic conflict, and post-breakup depression. In clinical cohorts, high baseline scores on fearful avoidance strongly predict treatment non-compliance and poor initial therapeutic alliance, whereas secure scores reliably forecast favorable responses across diverse modalities of psychotherapy.

Reliability

The psychometric reliability of the Adult Attachment scale has been rigorously evaluated across non-clinical college populations, community adult samples, and clinical cohorts. Findings indicate acceptable-to-strong reliability coefficients across both categorical subscales and continuous dimensional scoring protocols.

Internal Consistency

The internal consistency of the subscales, assessed via Cronbach’s alpha (α), varies depending on whether researchers utilize categorical scale averages or continuous orthogonal dimensions:

• Secure Subscale: Cronbach’s α typically ranges from .60 to .75 across diverse adult cohorts. The moderate alpha reflects the theoretical heterogeneity of security, which encompasses both emotional confidence and interpersonal autonomy.
• Preoccupied Subscale: Cronbach’s α ranges from .69 to .82, reflecting high conceptual cohesion around emotional merger, fear of rejection, and hyperactivating strategies.
• Fearful Subscale: Cronbach’s α demonstrates strong internal consistency, consistently reporting α values between .76 and .86 across both community and clinical samples.
• Continuous Dimensions (Anxiety and Avoidance): When items are converted into overarching continuous axes (Model of Self / Anxiety and Model of Other / Avoidance), reliability improves significantly, yielding Cronbach’s α coefficients between .81 and .88, and McDonald’s omega (ω) values exceeding .84.

Test-Retest Temporal Stability

Test-retest reliability assessments demonstrate robust temporal consistency over both brief and extended intervals. Over a 2- to 4-week window, test-retest correlation coefficients range from r = .75 to .84. Over longer developmental horizons—spanning 8 to 24 months—stability coefficients remain substantial (r = .56 to .68), confirming that attachment schemas function as stable dispositional working models while remaining open to structural revision following major life transitions or corrective relationship experiences.

Factor Analysis

Factor analytic investigations of the Bartholomew and Horowitz 19-item Adult Attachment inventory have been central to settling the historical debate between categorical taxonomies and dimensional continuum models in attachment science.

Exploratory Factor Analysis (EFA)

Initial Exploratory Factor Analyses using principal axis factoring and oblique (Promax / Direct Oblimin) rotations consistently reveal a clear two-factor solution accounting for the preponderance of total variance (typically 42% to 56%):

• Factor 1: Attachment Anxiety (Negative Model of Self): Items indexing intense anxiety regarding partner availability, dread of rejection, and excessive desire for emotional merger load strongly onto this factor (loadings typically ranging from .52 to .79). Key loading items include Item 1 (“I find others are reluctant to get as close as I would like”), Item 6 (“…I sometimes worry that others do not value me…”), Item 10 (“I often want to get closer…”), and Item 18 (“My desire to merge sometimes scares people away”).
• Factor 2: Attachment Avoidance (Negative Model of Other): Items reflecting aversion to intimacy, emotional vulnerability, and distrust load heavily onto this factor (loadings ranging from .48 to .81). Key markers include Item 4 (“I find it difficult to allow myself to depend on others”), Item 5 (“I am nervous when anyone gets too close”), Item 8 (“…I find it difficult to trust others completely”), and Item 13 (“I find it difficult to trust others completely”). Reverse-keyed items assessing comfort with intimacy (e.g., Item 9, Item 12, Item 17) load negatively onto this factor.

Confirmatory Factor Analysis (CFA)

Confirmatory Factor Analyses evaluating competing structural models have demonstrated that a two-dimensional orthogonal or lightly correlated model provides significantly superior fit indices compared to a single-factor general distress model or four uncorrelated categorical factors. Typical goodness-of-fit indices reported across large adult samples include:

• Comparative Fit Index (CFI) ≥ .92 to .96
• Tucker-Lewis Index (TLI) ≥ .91 to .95
• Root Mean Square Error of Approximation (RMSEA) ≤ .045 to .062 (90% CI [.038, .058])
• Standardized Root Mean Square Residual (SRMR) ≤ .051

These structural findings confirm that adult attachment is most accurately represented psychometrically as a continuous two-dimensional plane (Anxiety × Avoidance), from which the four prototype categories can be derived through quadrant cross-classification or calculated as distance scores from idealized prototypes.

Instrument / Measurement Tool

The technical specifications, test parameters, and scoring protocols for the Bartholomew and Horowitz 19-item Adult Attachment scale are structured as follows:

• Instrument Name: Adult Attachment / Relationship Scales Questionnaire (RSQ) – Bartholomew & Horowitz Version
• Construct Assessed: Individual differences in adult attachment representations, internal working models of self and others, and interpersonal intimacy strategies.
• Test Type: Standardized self-report inventory / psychometric rating scale.
• Item Count: 19 declarative statements assessing attitudes toward intimacy, emotional vulnerability, and interpersonal trust.
• Response Format: A 7-point Likert-type response scale anchored by:
  • 1 = Strongly Disagree
  • 2 = Disagree
  • 3 = Slightly Disagree
  • 4 = Neutral / Mixed
  • 5 = Slightly Agree
  • 6 = Agree
  • 7 = Strongly Agree
  • 9 = Don’t Know (Treated as a non-response / missing data point in quantitative psychometric scoring)
• Administration Time: Approximately 5 to 10 minutes.
• Target Population: Adults and adolescents aged 16 and older capable of reflecting on personal and relational patterns.
• Subscale Derivation and Scoring Protocols:
  • Secure Attachment Subscale: Composed of 7 items reflecting comfort with emotional closeness and mutual dependency. Scores are derived by computing the arithmetic mean of items representing secure representations (e.g., Items 9, 12, 14, 15, 17, with appropriate reverse coding where indicated).
  • Preoccupied Attachment Subscale: Composed of 6 items capturing hyperactivating strategies, craving for total intimacy, and anxiety over partner withdrawal (e.g., Items 1, 3, 6, 10, 16, 18).
  • Fearful Attachment Subscale: Composed of 6 items measuring desire for closeness accompanied by acute fear of rejection and interpersonal mistrust (e.g., Items 2, 4, 5, 7, 8, 13, 19).
  • Dimensional Scoring:
    • Attachment Anxiety (Model of Self): Computed by summing or averaging items reflecting anxiety, fear of abandonment, and low relational self-worth.
    • Attachment Avoidance (Model of Other): Computed by summing or averaging items reflecting discomfort with closeness, distrust, and compulsive independence (reverse-coding items reflecting interpersonal comfort).

Permissions & Fee and Test Year

The foundational theoretical framework and empirical items underpinning this adult attachment instrument were published by Kim Bartholomew and Leonard M. Horowitz in 1991 in the Journal of Personality and Social Psychology. In alignment with academic psychometric traditions, the Adult Attachment questions and the Relationship Scales Questionnaire are considered open-access instruments for non-commercial academic research, pedagogical purposes, and individual clinical evaluation, provided appropriate intellectual attribution and citation are accorded to the authors and original publisher.

No commercial royalty fees or formal licensing contracts are required for standard academic research investigations. However, organizations seeking to incorporate the instrument into proprietary diagnostic software platforms, commercial corporate assessment suites, or fee-for-service digital applications should seek formal copyright clarification from the copyright holders and the American Psychological Association (APA).

References

• Ainsworth, M. D. S., Blehar, M. C., Waters, E., & Wall, S. (1978). Patterns of attachment: A psychological study of the strange situation. Lawrence Erlbaum Associates.
• Bartholomew, K. (1990). Avoidance of intimacy: An attachment perspective. Journal of Social and Personal Relationships, 7(2), 147–178. https://doi.org/10.1177/0265407590072001
• Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among young adults: A test of a four-category model. Journal of Personality and Social Psychology, 61(2), 226–244. https://doi.org/10.1037/0022-3514.61.2.226
• Bowlby, J. (1969). Attachment and loss: Vol. 1. Attachment. Basic Books.
• Bowlby, J. (1973). Attachment and loss: Vol. 2. Separation: Anxiety and anger. Basic Books.
• Bowlby, J. (1980). Attachment and loss: Vol. 3. Loss: Sadness and depression. Basic Books.
• Brennan, K. A., Clark, C. L., & Shaver, P. R. (1998). Self-report measurement of adult attachment: An integrative overview. In J. A. Simpson & W. S. Rholes (Eds.), Attachment theory and close relationships (pp. 46–76). Guilford Press.
• Griffin, D., & Bartholomew, K. (1994). Models of the self and other: Fundamental dimensions underlying measures of adult attachment. Journal of Personality and Social Psychology, 67(3), 430–445. https://doi.org/10.1037/0022-3514.67.3.430
• Hazan, C., & Shaver, P. (1987). Romantic love conceptualized as an attachment process. Journal of Personality and Social Psychology, 52(3), 511–524. https://doi.org/10.1037/0022-3514.52.3.511
• Mikulincer, M., & Shaver, P. R. (2016). Attachment in adulthood: Structure, dynamics, and change (2nd ed.). Guilford Press.
• Ravitz, P., Maunder, R., Hunter, J., Sthankiya, B., & Lancee, W. (2010). Adult attachment measures: A 25-year review. Journal of Psychosomatic Research, 69(4), 419–432. https://doi.org/10.1016/j.jpsychores.2009.08.006

Items of the Scale