Differentiates among indices of agreement, reliability, and non-independence in organizational multilevel research and provides an overview of what each measure reveals about the group-level properties of one's data. Particular attention is devoted to the concept of non-independence and to the relationship between one form of the intraclass correlation coefficient and eta-squared. The 2nd purpose of this chapter is to extend the discussion of agreement, reliability, and non-independence from an exclusive focus on validating one's measurement model to a broader focus that includes testing substantive models and detecting emergent phenomena. In the discussion of bottom-up processes, particular attention is paid to a category that the author defines as "fuzzy" composition processes. (PsycINFO Database Record (c) 2012 APA, all rights reserved)