Department of Statistics, UCLA

There are now quite a number of books on multilevel analysis targeted at applied researchers in the behavioral and social sciences. They are not intended primarily for methodologists, psychometricians, or statisticians. The idea is to provide information about this relatively new class of techniques to researchers who want to use them in their research.

The latest entry in the field is the book by Joop Hox, reviewed here. Earlier competitors are the classic work by Bryk and Raudenbush [1992], now Raudenbush and Bryk [2002], and the textbooks by Snijders and Bosker [1999] and Kreft and de Leeuw [1998]. The books by Longford [1993] and by Goldstein [1995], no matter how excellent, are written for an essentially different target group, with more statistical sophistication. And there are the edited books by Reise and Duan [2003] and by Leyland and Goldstein [2001]. Although these books cover the same material, they cannot really be compared, because they do not speak with a single voice, and their contributions vary wildly in terms of scope and level of difficulty.