STABILITY AND CHANGE OF ATTACHMENT STYLES FOLLOWING IDIOSYNCRATIC LIFE EVENTS: PRELIMINARY EVIDENCE BASED ON THE STATE ADULT ATTACHMENT MEASURE (SAAM)

Attachment security, anxiety, and avoidance styles are acquired in infancy to remain stable across the life-span. Nonetheless, major life events influence attachment styles. We examined the extent to which attachment ratings had stable trait and fluctuating state characteristics. To this purpose 46 participants took the State Adult Attachment Measure (SAAM) on two different occasions (T1 and T2) spaced one month apart. On T2 participants were asked if a major life change occurred between T1 and T2. 25 participants reported no change, 21 reported a life change, in most cases with a positive valence. Data were modeled in a PLS-SEM framework. For each of the three styles, an exogenous and an endogenous latent variable loaded on T1 and T2 items, respectively. A regression path was set to represent test-retest stability. According to PLS-SEM, the state account for a set of measures is supported if the square root of the Average Extracted Variance (sqrAVE) is larger than a test-retest coefficient. Multiple group analyses compared model’s parameters for participants who experienced a life change and no change. The sqrAVE was larger than the test-retest coefficient for security and anxiety scores, suggesting fluctuations in attachment states. Avoidance ratings exhibited more stable trait-like characteristics. Multi-group analyses revealed that T2 scores were less predictable for participants who experienced a life change, attaining statistical significance for attachment security, only. Participants reporting positive life changes tended to score higher on security on T2. Our data show that SAAM scores were affected by trait- and state-like components. Security exhibited state-like characteristics for participants who reported a major life change. SAAM is indeed sensitive to life events that influence attachment styles. Future studies, on larger samples, involving clinical populations, and multiple measurement occasions are needed to cross-validate study’s results.