Content area
Full Text
Hierarchical Linear Models: Applications and Data Analysis Methods, 2nd ed., by Stephen W. Raudenbush and Anthony S. Bryk (2002). Thousand Oaks, CA: Sage, 504 pages. ISBN 0-7619-1904-X.
The organizational sciences, along with other disciplines, have seen a growing interest in multilevel modeling in recent years. This should not come as a surprise, given the hierarchical nature of organizational data, with observations nested in individuals, individuals nested in groups/teams/departments, groups/teams/departments nested in organizations, and so on. We have had a long-standing theoretical interest in variables at each of these levels as well as relationships between variables that are both contained within levels and span across levels. But statistical challenges have prevented us from "cleanly" testing and estimating these models.
Hierarchical linear modeling (HLM), which is one name for a class of modeling approaches that also goes by, among others, multilevel modeling and random coefficients modeling, is one method for analyzing such data that (a) avoids many problems of earlier approaches (e.g., running a single-level analysis with either the higher-level variables assigned down to lower-level units or lower-level variables aggregated to the higher-level unit) and (b) has caught on with organizational researchers. Undoubtedly, this was helped along by the first edition of this book (published in 1992 and sharing the same title but with the order of authorship reading Bryk and Raudenbush) as well as the related software package, currently in its fifth version (Raudenbush, Bryk, Cheong, & Congdon, 2000). The decade since the publication of the first edition has seen several advances in HLM. Hence, the authors' and editor's perceived need for the second edition. I could not agree more. But more on that later.
There are two categories of organizational scholars who should consider adding this book to their collection. The first is the researcher who actually uses the HLM software to fit multilevel models. To such a person, I would suggest that this book is indispensable (but then, you are probably already aware of this!). The reasoning is simple. The manual that ships with the software is good, but it is not comprehensive. In fact, indepth discussion of some topics, topics you will want to be informed about when running these models, is not offered in the manual. Rather, the reader is referred to...