Abstract
The goal of the present study is to demonstrate the ways in which multilevel models can be applied to family research. We emphasize the conceptual issues in family research that this data analytic technique helps us to address. The family represents a nested, hierarchical structure in which multiple children from the same family are not independent from one another. Multilevel models can be used to accommodate the complex structure of families. We draw on two data structures to illustrate the utility of the analytic technique. The first data structure involves children nested within families. With this data structure, it is possible to: 1) differentiate between family-wide and child-specific processes, 2) examine the way in which adverse family environments may exacerbate differences across siblings and 3) examine the way in which individual child characteristics may modify the impact of the family environment. In addition to children nested within families, data structure # 2 involves a cross-classification, as each parent reports on the emotional problems of multiple children. This hierarchical, cross-classified model allows us to examine predictors of children's emotional problems, predictors of informant agreement on children's emotional problems and the extent of children's similarity with their siblings on emotional problems.