ABSTRACT
In the social sciences, nested data structures are very common. As Burstein (1980) noted, “much of what goes on in education occurs within some group context” (p. 158). Nested data (which can yield correlated observations) occurs whenever participants are clustered together in groups, as is frequently found in social science research. For example, students in the same school will typically be more alike than students from different schools. Responses of clients to counseling for those clients clustered together in therapy groups will depend to some extent on the group dynamics, resulting in a within-therapy group dependency (Kreft & de Leeuw, 1998). Yet, a key inferential assumption made in virtually any statistical technique (including regression, ANOVA, etc.) used in the social sciences (and covered in this text) is that the observations are independent.
