Ten Questions That Measure How Adults Bond: A New Chinese Attachment Test Passes Its Toughest Trial

Attachment theory has long held that the way adults bond with romantic partners traces back to a behavioral system evolved to keep us safe under threat. For decades, psychologists have measured that system with the Experiences in Close Relationships-Revised, a 36-item questionnaire widely regarded as the gold standard for assessing adult attachment. Now, a team of Taiwanese researchers has put the instrument’s abbreviated Chinese versions through one of the most rigorous validations to date, and the results carry implications far beyond test construction. Published in Current Psychology, the study by Yu-Lien Huang of Chung Shan Medical University and Kuang-Hui Yeh of Academia Sinica systematically compared five shortened Traditional Chinese forms of the ECR-R across three independent samples, ultimately crowning a ten-item version as the most reliable and versatile tool for capturing attachment processes in Chinese-speaking populations.

The theoretical foundation of the work rests on a two-dimensional model of adult insecurity. Attachment anxiety reflects a heightened sensitivity to rejection, a chronic worry that partners will not be available or responsive in moments of need. Attachment avoidance, by contrast, describes a self-reliant distancing from intimacy and a skepticism toward others’ goodwill. According to the framework developed by Mikulincer and Shaver, anxious individuals deploy hyperactivating strategies, intensifying distress signals and proximity-seeking when threatened, while avoidant individuals deploy deactivating strategies that suppress attachment-related distress and maintain emotional distance. These regulatory patterns, the theory holds, shape how people interpret interpersonal cues, respond to stress, and sustain close relationships across the entire lifespan.

Why bother shortening a validated questionnaire at all? Respondent burden is a genuine obstacle in both research and clinical practice, and abbreviated versions of the ECR-R have proliferated internationally, including 18-, 16-, 12-, 10-, and 8-item forms validated in German, Czech, Thai, and other contexts. Yet most prior studies examined only a single short form in a single, usually nonclinical, sample. What remained unknown was which abbreviated version offers the optimal balance between brevity and psychometric robustness, particularly for people experiencing relational adversity, and whether any of them perform adequately in Chinese and Taiwanese cultural settings where direct comparisons had never been made.

The research team tackled this gap with an unusually comprehensive design. Study 1 recruited 380 nonclinical college students, aged 18 to 54, who completed the full 36-item Traditional Chinese ECR-R along with measures of mentalization, mindfulness, filial piety beliefs, depression, anxiety, and perceived stress. The researchers first confirmed that the full-length scale reproduced its original two-factor structure, with factor loadings between 0.26 and 0.84 for anxiety and 0.38 and 0.85 for avoidance, and exceptionally high internal consistency. They then used confirmatory factor analysis to test each of the five short forms against the same data, evaluating fit with standard indices including the Comparative Fit Index, the Tucker-Lewis Index, the root mean square error of approximation, and the standardized root mean square residual.

The verdict was clear. Initial models for every abbreviated version showed poor fit, but after permitting theoretically defensible covariances among specific error terms, the 18-item and 10-item versions emerged as clear winners, achieving fit statistics that met the strictest criteria. Standardized factor loadings in the final models ranged from 0.53 to 0.74, and the correlation between the anxiety and avoidance factors was modest, at 0.19 and 0.15 for the two versions respectively, confirming that the dimensions are related but distinct. Internal consistency was strong for both, and a subsample of 128 participants who repeated the questionnaire four weeks later produced intraclass correlation coefficients indicating moderate to good temporal stability. All subsequent validity analyses focused exclusively on these two contenders.

Study 2 extended the testing to 402 community adults, a more heterogeneous group recruited through social media. Both short forms again displayed a stable two-factor structure, and multi-group confirmatory factor analysis established that the measures achieved metric and scalar invariance across the student and community samples, meaning the scales measure the same constructs in the same way across developmental stages. The validity correlations were theoretically coherent: higher attachment anxiety and avoidance predicted greater difficulties in emotion regulation and more problematic social media use, as measured by the Bergen Social Media Addiction Scale. Notably, when the researchers split participants using the scale’s clinical cut score, individuals with problematic social media use showed significantly elevated attachment anxiety but not elevated avoidance, hinting that anxious hyperactivation, rather than avoidant withdrawal, drives compulsive digital connection-seeking.

The most demanding test came in Study 3, which enrolled 317 trauma-exposed adults aged 18 to 77, all of whom met DSM-5 Criterion A for posttraumatic stress disorder by experiencing at least one traumatic event. Here the ten-item version maintained its excellent two-factor fit, and its associations with clinical outcomes were striking. Attachment anxiety and avoidance both correlated positively with PTSD symptom severity, problematic alcohol use, impaired mentalized affectivity, and emotion regulation difficulties, with effects ranging from small to medium. When participants were divided into those whose worst trauma was impersonal, such as accidents or disasters, and those whose worst trauma was interpersonal, such as assault or betrayal, the interpersonal trauma group showed significantly higher attachment anxiety. The ten-item questionnaire, in other words, could detect a theoretically predicted vulnerability signature in one of the most psychologically burdened populations imaginable.

One intriguing null finding deserves attention. Attachment avoidance showed no association with problematic alcohol use, a pattern the authors explain through the deactivating strategy framework. Highly avoidant individuals defensively turn their attention away from both their emotional distress and their attachment figures, a self-regulatory style that minimizes felt distress and thereby reduces the need to rely on alcohol for affect regulation. The finding illustrates why distinguishing anxiety from avoidance matters clinically: the two forms of insecurity follow different paths to maladaptation, and a measure that collapses them would obscure these divergent mechanisms.

The study also embedded attachment within a distinctly Taiwanese cultural framework by including filial piety beliefs as external criteria. The dual filial piety model distinguishes reciprocal filial piety, grounded in emotional intimacy and gratitude toward parents, from authoritarian filial piety, centered on hierarchy and obligation. Both attachment dimensions correlated negatively with mindfulness, and avoidance showed a robust negative association with reciprocal filial piety specifically. The authors interpret this through the culture-fit hypothesis: in collectivistic contexts that prize interdependence, avoidant individuals’ rigid emotional boundaries obstruct the genuine, bidirectional exchange of love and gratitude that reciprocal filial piety demands, making avoidance especially damaging to parent-child relationships in Chinese societies.

The findings are not without caveats. Multi-group analysis across the trauma sample supported metric but not scalar invariance, meaning that while relationships among variables can be compared across groups, direct comparisons of average attachment levels between nonclinical and trauma-exposed respondents should be made cautiously, possibly because trauma-exposed individuals apply heightened subjective thresholds when reporting interpersonal distress. The samples were also predominantly female, lacked clinically diagnosed patients, and the cross-sectional design precludes causal claims. Still, the practical payoff is substantial: a ten-item questionnaire that takes moments to complete yet reliably tracks attachment anxiety and avoidance, their links to emotion dysregulation and impaired mentalization, and their sensitivity to interpersonal trauma. For researchers studying attachment across cultures and for clinicians screening for relational vulnerability, the CECR-R-10 offers a rare combination of efficiency, rigor, and cultural grounding, demonstrating that sometimes the smallest instruments reveal the largest patterns of the human heart.

Subject of Research: Validation of Traditional Chinese short forms of the ECR-R adult attachment questionnaire in Taiwanese samples

Article Title: Assessing adult attachment in taiwan: validity and reliability of traditional chinese short forms of the experiences in close relationships-revised and their implications for attachment mechanisms

Article References: Huang, Y.-L., & Yeh, K.-H. (2026). Assessing adult attachment in taiwan: validity and reliability of traditional chinese short forms of the experiences in close relationships-revised and their implications for attachment mechanisms. Current Psychology, 45(17), Article 1443. https://doi.org/10.1007/s12144-026-09996-5

Image Credits: AI Generated

DOI: 10.1007/s12144-026-09996-5

Keywords: adult attachment, ECR-R, psychometrics, attachment anxiety, attachment avoidance, Taiwan, trauma, PTSD, emotion regulation, mentalization, filial piety, questionnaire validation