The Latent Structure of the Adult Attachment Interview: Large Sample Evidence from Consortium Data

• In Event: 1-028 - Attachment in the Age of Big Science: Theoretical Progress Through Consortium-Building

Integrative Statement

The Adult Attachment Interview (AAI) is a well-validated and widely used measure that assesses adults’ states of mind regarding their early attachment-related experiences with primary caregivers. The standard system for coding the AAI (Main, Goldwyn, & Hesse, 2003–2008) recommends classifying adults categorically as having an autonomous/secure, dismissing, preoccupied, or unresolved attachment state of mind. Studies have empirically explored the latent structure of the AAI using factor analytic techniques (which identify the number of distinct constructs underlying a set of observations) and taxometric procedures (which evaluate whether the variation in a construct is categorical or dimensional). The evidence from these prior studies suggests that: (a) adults’ attachment states of mind are captured by two factors reflecting adults’ dismissing and preoccupied states of mind and (b) individual differences on these factors are more consistent with a continuous than a categorical model (e.g., Haltigan, Roisman, & Haydon, 2014; Fraley & Roisman, 2014).
Attachment scholars have debated the theoretical implications of these findings, especially the controversial observation that ratings of adults’ preoccupied states of mind and ratings of adults’ unresolved states of mind about loss and trauma have loaded on a common factor (e.g., Roisman, Fraley, & Booth-LaForce, 2014; Van IJzendoorn & Bakermans-Kranenburg, 2014). The purpose of the current study was to evaluate the competing hypotheses about the latent structure of AAI states of mind by leveraging individual participant data from 40 studies (N = 3,218). The unprecedented sample size allows for statistically powerful tests of the potential distinction between preoccupied and unresolved states of mind as well as the unsettled question of whether variation in attachment states of mind is dimensionally or categorically distributed.
Confirmatory factor analyses indicated that: (a) a 2-factor model that included one factor for dismissing states of mind and a second factor for preoccupied and unresolved states of mind and (b) a 3-factor model that distinguished unresolved from preoccupied states of mind were both compatible with the data (Figure 1). Comparisons of the relative fit of these two models were ambiguous, as the χ2 test favored the 3-factor model but the BIC values favored 2-factor model. In the 3-factor model, the correlation between the factors for preoccupied and unresolved states of mind was large (latent r = .87).
Two taxometric tests (MAXCOV MAXEIG and L-Mode) indicated that variation in both dismissing and preoccupied states of mind was more consistent with a dimensional than a categorical model. The results of the third taxometric test (MAMBAC) did not clearly support either a categorical or a dimensional model (Table 1).These findings generally suggest that a dimensional model provides a more plausible explanation for the variation in adults’ states of mind than a categorical one.
Altogether, the results of this large-sample study suggest there is a need to alter the traditional assumptions about the latent structure of the AAI, especially the assumption of taxonicity. Future studies that test whether these dimensional indices yield new insights into the processes underlying the intergenerational transmission of attachment would be valuable.

Group Authors

The Collaboration on Attachment Transmission Synthesis

Authors

• Lee Raby, University of Utah
  Presenting Author

• Marije L. Verhage, Vrije Universiteit Amsterdam
  Non-Presenting Author

• Pasco Fearon, Research Department of Clinical, Educational and Health Psychology, University College London
  Non-Presenting Author

• Chris Fraley, Department of Psychology, University of Illinois at Urbana-Champaign
  Non-Presenting Author

• Glenn Roisman, University of Minnesota
  Non-Presenting Author

• Marinus H van IJzendoorn, Erasmus University Rotterdam
  Non-Presenting Author

• Carlo Schuengel, Vrije Universiteit Amsterdam
  Non-Presenting Author

• Marian J Bakermans-Kranenburg, Vrije Universiteit Amsterdam
  Non-Presenting Author

• Annie Bernier, University of Montreal
  Non-Presenting Author

• Karin Ensink, University of Laval
  Non-Presenting Author

• Airi Hautamäki, Social Psychology and Psychology, Swedish School of Social Science, University of Helsinki
  Non-Presenting Author

• Sheri Madigan, University of Calgary and the Alberta Children’s Hospital Research Institute, Calgary
  Non-Presenting Author

• Sarah Mangelsdorf, Department of Psychology, University of Wisconsin-Madison
  Non-Presenting Author

• Mirjam Oosterman, Vrije Universiteit Amsterdam
  Non-Presenting Author

• Lynn Priddis, School of Arts and Humanities, Edith Cowan University
  Non-Presenting Author

• Maria S Wong, Stevenson University
  Non-Presenting Author