原版图书,内容细致:
This book has grown out of research undertaken at the Department of Biostatistics
of the Harvard School of Public Health and the Dana-Farber Cancer
Institute in Boston on the one hand and at the Limburgs Universitair Centrum
(transnational University Limburg) in Belgium on the other hand, in
close collaboration with a number of colleagues located at various institutions.
Research interests in the modeling of clustered and repeated categorical
data have been brought together with research in the modeling of data
from toxicological experiments in general and developmental toxicity studies
in particular.
the field of developmental toxicity is exciting and raises a large
range of substantive and methodological questions. Nevertheless, the methodology
presented here has much wider ramifications than just this field of application.
Therefore, a modular concept seemed most appropriate. In particular,
the motivating examples have been collected in a separate chapter, followed by
chapters on model building and on particular estimation procedures (generalized
estimating equations and pseudo-likelihood). The “modeling chapters”
have been written with a general clustered or even correlated data setting in
mind and are therefore of use far beyond the developmental toxicology context.
In later chapters, specific issues have been tackled. Some of these are
rather particular to toxicology and dose-response modeling (e.g., the chapters
on quantitative risk assessment and exact dose-response inference) while
others are more general in scope (e.g., the chapters on goodness-of-fit, model
misspecification, individual level covariates, and combined continuous and discrete
outcomes). In this way, both the very focused as well as the more broadly
interested reader will have no difficulty selecting the material of interest to
her. To underscore the large potential of methods for clustered data, a chapter
has been included on the analysis of clustering effects in complex survey data.