Skip to main content
. 2014 Jun 12;180(2):197–207. doi: 10.1093/aje/kwu110

Figure 2.

Figure 2.

Diagram for the 4-level classification models of individuals, sibling groups, areas, and schools in the Aberdeen Children of the 1950s Study. Areas are enumeration districts from the 1961 census, and schools are primary schools in Aberdeen, Scotland, in 1962. Individuals are nested within sibling groups, and sibling groups are nested within areas. One cross-classification arises from individuals from the same sibling group not attending the same school. This cross-classification does not exist for sibling groups containing 1 child nor for sibling groups containing 4 children, because all 4 children in these sibling groups attended the same school. The cross-classification occurs for sibling groups of 2 or 3 children; 130 (16%) sibling groups of 2 children attended different schools, and 13 (19%) sibling groups of 3 children attended different schools. There is a further cross-classification of areas and schools where individuals from the same area attend different schools. None of the 104 areas was nested within a school; all areas had some children who attended different schools. The absence of arrows displays these cross-classifications; where an arrow exists, it displays the typical nested multilevel model structure (25). IQR, interquartile range.