Attachment in middle and late childhood : measurement validation and relation to mental health problems

Attachment theory is a widely appreciated perspective on the importance of close relationships for human development. Middle and late childhood (8 to 13 years) is one of the least studied age periods in attachment research. Challenges for attachment research in general, and this age period in particular, concern the adequacy of measurement procedures and the influence of attachment on mental health problems. This dissertation includes cross-sectional data from 150 children (M age 11.7) and 121 parents. Children were interviewed with the Child Attachment Interview (CAI), which is videotaped and coded according to a manual. Both parents and children completed questionnaires including measures of mental health problems, as well as risk and protective factors for mental health problems. Individual differences in attachment are traditionally conceived of in terms of categories (e.g. secure-insecure). Yet, there are both theoretical and empirical reasons to consider individual differences as being distributed along dimensions. Furthermore, attachment researchers have to a limited extent taken advantage of recent developments in psychometrics, in particular a latent variable approach like Confirmatory Factor Analysis (CFA), to study construct validity. The first paper in this dissertation applies a CFA to the CAI, and thereby examines the construct validity of a continuous conception of the attachment construct in middle and late childhood. In this study, we found evidence for a two-dimensional model comprising the factors Security-Dismissal and Preoccupation-Idealization. Furthermore, we found these factors to maintain the information inherent in the categories, but to add information about subtle differences between individuals. The continuous approach to the attachment construct was applied in the two other papers in this dissertation as well. A small, yet consistent, association is found between attachment and mental health problems across age groups; insecurity as a risk factor, and security as a protective factor. More complex models are used to understand the role of attachment and other risk and protective factors for mental health problems. The multi-variate risk factor model where attachment, family adversity, and ineffective parenting are considered, alone and in combinations, to influence mental health problems, has been influential. The diathesis-stress model is another influential way of understanding how attachment influences mental health problems. This model is based on a stress-regulatory approach to attachment, and hypothesizes that children with high levels of attachment security have more efficient means IV for coping with stress. In contrast, children with low levels of attachment security (i.e. attachment insecurity) have less efficient means. In paper two, we combine these perspectives by hypothesizing attachment as a part of a multi-variate model of family risks, however as being a moderator of the negative effect of the other risk factors on internalizing and externalizing problems. This hypothesis was supported for some, but not all, of the risk factors. Also, there was a dose-response relationship between a cumulative index of risk factors and mental health problems for children with low levels of attachment security, whereas children with high levels of security were unaffected by the accumulation of family risks. The association between low levels of attachment security (i.e. insecure attachment) and internalizing problems is unquestionable. However, there are children who have more or less internalizing problems than expected, given their level of attachment security. The purpose of paper three is to examine characteristics of these children. We found high levels of difficult temperament (negative emotionality and shyness), as well as family risks, to characterize children with more internalizing problems than expected, given their level of security. Low levels of difficult temperament and family risks, but not positive temperament (activity) and social support from peers, explained less internalizing problems than expected, given the child’s level of attachment security. This dissertation addressed several important topics in attachment research, particularly concerning middle and late childhood. The main findings in this dissertation are: - Attachment organization, measured with the Child Attachment Interview, may adequately be conceived as a two-dimensional construct, however with one main dimension ranging from security to dismissal - Attachment security buffers the negative effects of family risk of mental health problems in middle and late childhood. - Children with more internalizing problems than expected, given their level of attachment security, tend to have difficult temperament and experience family risks