Assessing the reliability of change: A comparison of two measures of adult attachment

Recently attachment researchers have become interested in how attachment changes within an individual due to social or cognitive context fluctuations. Such analyses of process are limited by unreliability of change scores. Traditional estimates of between-person reliability (Cronbach’s alpha) are not informative about how reliable a measure is at capturing within-person change. In two longitudinal studies, we examined the reliability of the State Adult Attachment Measure (SAAM; Gillath, Hart, Noftle, & Stockdale, 2009) and the Experiences in Close Relationships scale (ECR; Brennan, Clark, & Shaver, 1998), in capturing attachment change. We used generalizability theory analyses to estimate the between- and within-person reliabilities of both scales. Even with fewer items, the reliability of change for the SAAM was higher than that of the ECR.

► The SAAM was more reliable in measuring changes in attachment than the ECR.
► Researchers interested in assessing change in attachment should consider the SAAM.
► Studies of change should assess reliability using appropriate psychometric methods.
► These results may not generalize to studies with shorter or longer time frames.