Missing data are a real bane to researchers across all social science disciplines. For most of
our scientifi c history, we have approached missing data much like a doctor from the ancient
world might use bloodletting to cure disease or amputation to stem infection (e.g, removing
the infected parts of one’s data by using list-wise or pair-wise deletion). My metaphor should
make you feel a bit squeamish, just as you should feel if you deal with missing data using
the antediluvian and ill-advised approaches of old. Fortunately, Craig Enders is a gifted quantitative
specialist who can clearly explain missing data procedures to diverse readers from
beginners to seasoned veterans. He brings us into the age of modern missing data treatments
by demystifying the arcane discussions of missing data mechanisms and their labels (e.g.,
MNAR) and the esoteric acronyms of the various techniques used to address them (e.g., FIML,
MCMC, and the like).
Enders’s approachable treatise provides a comprehensive treatment of the causes of missing
data and how best to address them. He clarifi es the principles by which various mechanisms
of missing data can be recovered, and he provides expert guidance on which method
to implement and how to execute it, and what to report about the modern approach you
have chosen. Enders’s deft balancing of practical guidance with expert insight is refreshing
and enlightening. It is rare to fi nd a book on quantitative methods that you can read for its
stated purpose (educating the reader about modern missing data procedures) and fi nd that
it treats you to a level of insight on a topic that whole books dedicated to the topic cannot
match. For example, Enders’s discussions of maximum likelihood and Bayesian estimation
procedures are the clearest, most understandable, and instructive discussions I have read—
your inner geek will be delighted, really.