The multilevel dynamics of sibling relationships: Influences over time

A multilevel, longitudinal design was used to examine predictors of change in sibling affection and sibling hostility over time. Up to two children in a family reported on their relationships with their siblings resulting in a maximum of three sibling dyads per family. Mothers reported on predictor variables. One-hundred and nineteen families with 218 children as targets took part. A cross-classified, multilevel model allowed for the partitioning of variance into family, informant, target, mutuality and dyad. Sibling negativity was found to cluster significantly within families, whereas sibling affection showed no familial clustering. Variance partitioning showed that the bulk of the variance for sibling affection was attributable to mutuality and informant effects. Family, dyadic and individual level predictors of change in sibling affection and hostility were identified.