- Shopping Bag ( 0 items )
From The CriticsReviewer: Howard M. Kravitz, DO, MPH(Rush University Medical Center)
Description: This short monograph is one in the Arnold Applications of Statistics Series. The principal focus is on measurement error. The author uses a problem-based approach with examples from the field of mental health to illustrate his points.
Purpose: According to the author, the aim is "to create an intermediate text suitable for practising research workers in psychiatry and related disciplines." He intends this book to be "something different," a book about "statistics in psychiatric epidemiology" that reflects his interests and experience in this area. Topics were chosen to reveal the distinctive "flavor" of the subject. Preference was given to observational studies and the issues of reliability and validity, and clinical trials methods were purposefully omitted. The author's worthy objectives are not fully met, however — this is a taste, not a meal.
Audience: The audience is identified as psychiatric researchers, clinical, behavioral, and social scientists, and applied statisticians with limited experience in the mental health field. Readers are assumed to have a basic grasp of general introductory statistics. The author is the editor of the journal Statistical Methods in Medical Research and is a credible authority.
Features: Seven chapters are covered in 117 pages of text, plus a reference section and index. Chapter 1 is the introduction. Instrument reliability and validity are covered in Chapters 2 and 3. Hidden variables and multiple indicators are covered in Chapter 4 and prevalenceestimation in Chapter 5. The focus in Chapter 6 is on continuity and change, and in Chapter 7 on missing data. These last four chapters each include an appendix with a software program code. Tables and figures help expand the textual material. In a couple of instances, text and table do not correlate.
Assessment: This book contains occasional pearls and flashes of brilliance, but overall it is disappointingly short on depth and detail and is likely to be of little practical usefulness for the targeted readers. The author's terse style and ex cathedra statements beg for further elaboration. Whereas the discussion of missing data is brief but nicely done, coverage of factor analysis and covariance structure models was disproportionately overemphasized and baffling at times. The price is reasonable; however, psychiatric or social science researchers wanting a good, reasonably priced, intermediate level text on mainstream epidemiologic methods that are applicable to psychiatric epidemiology should try Szklo's Epidemiology: Beyond the Basics (Aspen Publishers, 2000).