Analysis of Health Surveys

How to apply statistical methods to survey data--a guide toeffective analysis of health surveys. With large health surveys becoming increasingly available forpublic use, researchers with little experience in survey methodsare often faced with analyzing data from surveys to addressscientific and programmatic questions. This practical book providesstatistical techniques for use in survey analysis, making healthsurveys accessible to statisticians, biostatisticians,epidemiologists, and health researchers. The authors clearlyexplain the theory and methods of survey analysis along withreal-world applications. They draw on their work at the NationalInstitutes of Health as well as up-to-date information from acrossthe literature to present: * The sampling background necessary to understand health surveys. * The application of such techniques as t-tests, linear regression,logistic regression, and survival analysis to survey data. * The use of sample weights in survey data analysis. * Dealing with complications in variance estimation in large healthsurveys. * Applications involving cross-sectional, longitudinal, andmultiple cross-sectional surveys, and the use of surveys to performpopulation- based case-control analyses. * Guidance on the correct use of statistical methods found insoftware packages. * Extensive bibliography.

The book fulfills the authors' purpose of providing neededexpertise for statisticians who are limited in their knowledge ofsurvey research. There are excellent questions at the end of eachchapter so it would make a good textbook for a course in surveyanalysis. (Journal of Official Statistics, Vol. 16, 2000) A strength of this book is the exercises at the ends of eachchapter [...]. This makes the book ideal either for independentstudy or for a survey methods course at final year undergraduate orpostgraduate level. Since the book is based on the two authors'many years of experience working in this field, it shows highlevels of sound judgement and good advice. (Biometrics, 2000) The analytic components of books about survey sampling oftenrestrict their discussion to the analysis of relatively simpleparameters of interest, such as population means and linearregression slopes. On the other hand, texts not devoted to surveysampling rarely discuss in detail the effect of complex sampledesigns on data analysis. This book provides a great service to thehealth research community by tying together the tools of modernstatistical analysis and survey research in one package. Korn andGraubard write clearly and concisely, using well-chosen examples toilluminate potentially confusing concepts. (JASA, March 2001) If you are involved in analyzing data from very large surveys, thenyou will want to buy this book. Statisticians who enjoy expandingtheir horizons in statistics by reading practical and usefulstatistics books should also get a copy of the book. More and moreof this health-survey data becomes available all the time, soperhaps data analysis no longer needed in industry can find a newhome somewhere analyzing health-survey data. (Technometrics, May2001, Vol. 42, No. 4) "...provides a great service to the health research community bytying together...tools of modern statistical analysis and surveyresearch...." (Journal of the American Statistical Association)

EDWARD L. KORN, PhD, is head of the Clinical Trials Section,Biometric Research Branch, at the National Cancer Institute. BARRY I. GRAUBARD, PhD, is a senior investigator in theBiostatistics Branch at the National Cancer Institute.