Introduction to biostatistics (2nd edition) , by R. R. Sokal and F. J. Rohlf. Pp 363. £37·95.1987. ISBN 0-716-71805-7 (Freeman)
Introduction to biostatistics (2nd edition) , by R. R. Sokal and F. J. Rohlf. Pp 363. £37·95.1987. ISBN 0-716-71805-7 (Freeman)
复制标题
DOI：
10.2307/3618952
复制
发表时间：
1988-06
期刊：
影响因子：
--
通讯作者：
A. J. Willis
中科院分区：
文献类型：
--
作者：
A. J. Willis
讨论了与假设检验相关的错误类型，并且没有警告如果E的值太小，则使用f的危险。随着书的进展，可读性的风格变差了离散和连续变量的讨论特别令人困惑。我也不满意定义方差从（另一个例子的公式出现的魔术）;后来，方差被定义为变量（J 0 = E（X - nf，但我们没有被告知这两个定义是如何相关或协调，和使用（n - 1）在r公式意味着讨论无偏估计不带出点时，将使教学这一主题。我们还被告知，\ar（X）公式适用于离散和连续分布，但期望仅定义为离散情况（因为积分是一个禁区）。所涉及的实际计算。许多用户可能会发现独特的“盒子”（其中有28个，设置在灰色背景上）很有帮助，因为它们显示了各种生物统计问题的计算程序;这些涵盖的方面包括，例如，从置信限的计算到Mann-Whitney U检验和Kolmogorov-Smirnov双样本检验。前几章涉及基本统计的标准特征，包括抽样，人口，变量，位置和分散的统计，二项和泊松分布以及正态概率分布。文中用工作实例和有关的实验数据很好地说明了这一点。这个版本的新功能包括茎和叶图和悬挂直方图。与其他章节一样，这些章节在这里以一系列练习结束（给出了数字答案）。一个重要的章节题为“估计和假设检验”涵盖了置信限和零假设和备择假设，有效地把一些重点放在检验的力量。方差分析占四章，包括单分类和双因素分析以及方差分析的假设。一个相当充分的帐户，作者正确地强调在序言中的需要，今天的生物学家有一个彻底的基础，在这一领域。这里一个有用的新特性是用于进行多重比较的Bonferroni方法。的/测试被视为方差分析的一个特殊情况下，有人认为，如果方差分析是早期理解，需要使用t分布减少。
discussion of the types of errors associated with hypothesis testing, and no warning of the dangers of using f if the values of E, are too small. The readable style deteriorates as the book progresses—the discussion of discrete and continuous variables is particularly confusing. I am also unhappy with defining variance as from the (another example of a formula appearing by magic); later on, variance is defined as Var(J0 = E(X — nf but we are not told how these two definitions are related or reconciled, and the use of (n — 1) in the r formula means that the discussion of unbiased estimators doesn't bring out the points one would make when teaching this topic. We are also told that the \ar(X) formula applies to both discrete and continuous distributions, but expectation has only been defined for the discrete case (because integration is a no-go area). the actual computations involved. Many users are likely to find the distinctive "boxes" (of which there are 28, set on a grey background) helpful, as they show the computational procedures for a wide variety of biostatistical problems; these cover aspects ranging, for example, from the calculation of confidence limits to the Mann-Whitney U-test and the Kolmogorov-Smirnov two-sample test. The early chapters deal with standard features of elementary statistics, including sampling, populations, variables, statistics of location and dispersion, the binomial and Poisson distributions and the normal probability distribution. The text is well illustrated with worked examples and by relevant experimental data. New features of this edition include stem-and-leaf diagrams and hanging histograms. The chapters conclude here, as elsewhere, with a series of exercises (numerical answers are given). An important chapter entitled "Estimation and Hypothesis Testing" covers confidence limits and null and alternative hypotheses, usefully putting some emphasis on the power of a test. Analysis of variance occupies four chapters; both single-classification and two-way analysis and also the assumptions of anova are covered. A quite full account is given, the authors rightly stressing in the Preface the need for today's biologists to have a thorough foundation in this field. A useful new feature here is the Bonferroni method for making multiple comparisons. The / test is treated as a special case of analysis of variance and it is argued that, if anova is understood early, the need to use the t distribution is reduced.