Exercises and solutions in biostatistical theory

Exercises and solutions in biostatistical theory, by Lawrence L. Kupper, Brian H. Neelon and Sean M. O'Brien, Boca Raton, Chapman & Hall/CRC Press, 2011, xvii + 402 pp., £24.99 or US$51.95 (paperback), ISBN 978-1-58488-722-5

This book contains the basic and main results of the theory of statistical methods in a classic overview and related exercises with their solutions. It is a compendium of solved problems for a university course of statistics and with an intentional orientation to biostatistics. Five chapters cover basic probability theory, univariate distribution theory, multivariate distribution theory, estimation theory, and hypothesis testing theory.

Why is theory its content? It must be indicated that in estimation and hypothesis testing, an available sample of data is not necessarily a simple random sample of the population to be observed. A simple random sample is a set of independent observations, identically distributed to the population. These conditions are usually uncheckable in practice since the experimenter has no control over the nature of the generation of available data both for simple or more complex random samples. Another issue is the confidence of the biostatistician or the statistician in his/her statistical model. I think that this is the main obstacle for the classic statistical methods of parametric inference to be credible as objective ones. These uncheckable hypotheses are the basis of the arguments of classic inference. That is to say, the theory is based on hypotheses, not on reality. As consequence, the theory is not the best for practical studies. An inference which does not have these obstacles is proposed in the book by Ruiz Espejo 1.

However the book is very well suited to its aims and is methodologically and mathematically correct. But this feature is necessary but not sufficient for a good practice in statistics and other methods. Reason without truth could be theory or ideology, but the truth is necessary in the statistical process to be scientifically acceptable in practice.

http://dx.doi.org/10.1080/02664763.2012.657407