The classifications derived from ratings on 17 subscales. The first three classifications are convergent with Ainsworth’s original three models: Secure (called Secure/Free/Autonomous), Avoidant (Dismissing of Attachment) and Anxious/Ambivalent (Preoccupied/Entangled). Main and colleagues also added two new types: Unresolved and Cannot Classify. Unresolved individuals seem distracted and disorganized. This may be caused by unresolved issues such as abuse or neglect in childhood. Cannot Classify individuals are just that, unclassifiable. They do not fit into any category and usually seem wholly incoherent.
This type of measure agrees with Bowlby’s original hypothesis that the infant-caregiver bond is an internal working model for later life relationships. The AAI focuses on an adult’s experience in childhood and the way that relationship affects subsequent relationships in his or her life. It puts little to no emphasis on other relationships that the individual has been involved in or how their bonding patterns may have changed. It is a categorical rather than a dimensional scale and does not allow for the possibility of an individual belonging to more than one attachment style.
Another downfall of the AAI is the length of time it takes to administer, transcribe and rate. It does not have great research utility. Raters must be trained in a two training program and pass a reliability trial to administer the interview. Although this extreme amount of time and energy does ensure that the interview is given properly, and, as a result, the AAI rates very high in psychometric values such as reliability and validity.
Others have attempted to reevaluate the AAI and revise it to make it more complete. George and West (2001) adapted the AAI and created the Adult Attachment Projective. It is a projective test based on the comments made by participants to seven