Distinguishing Attachment and Affiliation in Early Adolescents' Narrative Descriptions of Their Best Friendship

Abstract

This study was designed to test for specificity in the relationship between individual friendship provisions and adjustment across early adolescence. Using a narrative procedure, attachment (i.e., accessing care) and affiliation (i.e., forming cooperative partnerships) were found to be distinct functional themes organizing 293 adolescents' (M_age = 13) internal representations of their best friendship across three annual measurement occasions. Longitudinal, cross-lag analyses revealed a unique transactional relationship between friendship affiliation and greater social competence over time, controlling for friendship stability, maternal relationship quality, socioeconomic status, and gender. By contrast, friendship attachment predicted fewer subsequent internalizing symptoms from ages 14 to 15. Together, findings point to the importance of understanding individual differences in the content of adolescents' internal representations of friendship.

Figure 1

Path model displaying all possible cross-lagged paths between friendship attachment, friendship affiliation, and parent and teacher reports of internalizing symptoms over three years. The covariates Maternal Relationship Quality, Adolescent Gender, and SES were specified as exogenous predictors of all downstream variables in this model but, for ease of illustration, are not displayed here. All path coefficients shown are standardized values. All path coefficients shown are standardized values. Bolded paths reflect regression coefficients that are significant at p < .05. Light grey paths were included in the model, but were not significant. ** p < .01, * p < .05.

Figure 2

Path model displaying all possible cross-lagged paths between friendship attachment, friendship affiliation, and parent and teacher reports of social competence over three years. The covariates Maternal Relationship Quality, Adolescent Gender, and SES were specified as exogenous predictors of all downstream variables in this model but, for ease of illustration, are not displayed here. All path coefficients shown are standardized values. Bolded paths reflect regression coefficients that are significant at p < .05. Light grey paths were included in the model, but were not significant. ** p < .01, * p < .05.