ABSTRACT
A common form of data found in the behavioral and social sciences has subjects nested within one or more hierarchical levels of categories. In an educational setting, subjects may be nested within classes, each with a different teacher, and the classes in turn may be nested within schools, and these in turn within school districts. In an organizational behavior study, subjects may be employees studied within stores, stores in turn being classified within cities, cities in turn within regions. A hospital researcher may study patients within wards, the wards in turn are in hospitals, and hospitals may be within hospital systems, and these in turn in regions of the country. If at each level of categorization there is variation in the units within that category, this may represent potential sources of causal variation that has an influence on the observed variables of the study. For example, variations in store managers may bring different leadership styles to bear on the employees working under them and influence their performance on various variables. In turn, different cities within which the stores are located may introduce variations in the culture, transportation systems, health systems, and housing availability that may influence differently the people working in each city and influence rates of tardiness, turn-over, absenteeism, wage, and health insurance demands. The cities in turn may be placed under the supervision of different higher level managers, who establish different policies in the regions within which the cities are located. These variations in policy may also influence the variation in behavior of the employees within the regions.
