[Page 866]Family health research data are particularly complex as a consequence of involving repeated measurements over multiple family members and possibly also over multiple time points. For the family to be the unit of analysis, outcome (dependent, response) variables need to be measured for multiple family members. Different family members can have different perspectives on the family or on certain family members that are important to account for in analyses, for example, mother's and father's perspectives on family functioning or on a child's behavioral problems. Different families in a study can have different types, for example, single-parent versus two-parent families or different numbers within the same types, such as one or two participating parents for two-parent families. Family data are often dyadic, involving two family ...