P. Sprent & N.C. Smeeton (2007). Applied Nonparametric Statistical Methods (4th ed.).
P. Sprent & N.C. Smeeton (2007). Applied Nonparametric Statistical Methods (4th ed.).
复制标题
P·斯普伦特
DOI：
10.1007/s11336-010-9166-4
复制
发表时间：
2010
期刊：
影响因子：
3
通讯作者：
L. M. Schultz
中科院分区：
文献类型：
--
作者：
L. M. Schultz
应用非参数统计方法第四版比前一版略有改进。各章已重新编排，并增加了关于“现代”非参数计量学的新章节。值得注意的是，第一章已被修改，以提供基本统计概念的回顾，而第二章在大多数入门统计教科书中包括的水平上给出了非参数方法的广泛概述。其余章节对第2章中首次介绍的技术进行了扩展。在重组后的第9章中增加了生存分析的覆盖面，并且现在有整整一章(第8章)致力于析因设计。新的第15章重点介绍了密度估计、低值曲线平滑、Logistic回归和其他针对大数据集的方法。应该指出的是，自举的内容仍然有限(第14章有16页)，而只有4页(在第2章)专门介绍排列和重采样方法。这篇文章没有电子资源可用。这本书最大的优点是，它的写作水平是读者完全可以理解的，只需一两门入门级的统计课程。因此，它既适合作为本科生统计专业非参数方法第一门课程的教科书使用，也适合作为其他领域从业人员的参考。它也非常适合实验心理学或其他学科的研究生使用非参数方法分析数据的补充统计教科书。关键概念的讲授使用了来自不同领域的精心设计的例子。大多数章节以有用的总结、“真实世界”应用程序的例子和大量练习结束；选定练习的解决方案包含在附录中。正如标题所暗示的那样，这是一本应用课本；你会发现很少有数学推导，也很少有关于基本理论的讨论。虽然统计学研究生和其他对学习基础数学感兴趣的读者会被建议参考更高级的文本(例如，Hollander和Wolfe，1999；Wasserman，2006)，但这本书写得恰到好处，适合那些希望学习更多非参数方法的初等统计基础知识的人。正如作者在序言中所说，他们对非参数测量的治疗是“宽泛的，而不是深入的”。事实上，与Conover(1999)或Corder and Foreman(2009)等备选方案相比，这一文本涵盖了更广泛的主题。另一个优点是，应用非参数统计方法比它的许多竞争对手便宜得多(例如，2004年希金斯的价格为83.95vs.134.99美元)。自从这篇文章首次出现在《》中以来，计算能力有了巨大的发展。令人失望的是，作者没有做出更多的努力来更新这个版本，以反映R的广泛使用。这个免费的统计软件环境(R Development Core Team，2009)具有许多用于非参数检验的内置函数(例如Wilcoxon)。泰斯特和克鲁斯卡尔。测试)。诸如Coin(Hothorn，van de Wiel和Zeileis 2008)、Boot(Canty和Ripley，2009)和sm(Bowman和Azzalini，2010)等专门的R包也可免费用于排列测试、引导和平滑，分别用于非参数回归和密度估计。相反，作者几乎完全参考了StatXact(Mehta和Patel，2007)，这是一种相当昂贵的替代方案(995美元)，不太可能被本科生或其他非专业人士购买。此外，作者既没有包括说明
The fourth edition of Applied Nonparametric Statistical Methods represents a modest improvement over the previous edition. The chapters have been reorganized, and a new chapter on “modern” nonparametrics has been added. Notably, Chapter 1 has been modified to provide a review of basic statistical concepts, while Chapter 2 gives a broad overview of nonparametric methods at the level included in most introductory statistics textbooks. The remaining chapters expand upon the techniques first introduced in Chapter 2. Increased coverage is given to survival analysis in a restructured Chapter 9, and an entire chapter (Chapter 8) is now devoted to factorial designs. The new Chapter 15 focuses on density estimation, lowess curve smoothing, logistic regression, and other methods for large data sets. It should be noted that coverage of bootstrapping remains limited (16 pages in Chapter 14), while only four pages (in Chapter 2) are devoted to permutation and resampling methods. No electronic resources are available for this text. The greatest strength of this book is that it is written at a level that is perfectly understandable by readers with only a course or two of introductory-level statistics. As such, it is appropriate for use as either a textbook for a first course in nonparametric methods for undergraduate statistics majors or as a reference for practitioners in other fields. It is also quite suitable as a supplementary statistics textbook for graduate students in experimental psychology or other disciplines who may need to use nonparametric methods to analyze their data. Key concepts are taught using worked-out examples from a variety of fields. Most chapters conclude with a helpful summary, examples of “real-world” applications, and numerous exercises; solutions for selected exercises are included in an appendix. As the title implies, this is an applied text; you will find few mathematical derivations and very little discussion of underlying theory. While statistics graduate students and other readers interested in learning the underlying mathematics would be advised to consult a more advanced text instead (eg, Hollander and Wolfe, 1999; Wasserman, 2006), this book is written at just the right level for somebody with basic knowledge of elementary statistics who wishes to learn a bit more about nonparametric methods. As the authors state in the preface, their treatment of nonparametrics is “broad rather than deep.” Indeed, compared to alternatives such as Conover (1999) or Corder and Foreman (2009), this text covers a wider spectrum of topics. Another plus is that Applied Nonparametric Statistical Methods is considerably less expensive than many of its competitors (eg, US $83.95 vs. US $134.99 for Higgins, 2004).Since this text first appeared in 1989, there has been a massive evolution in computing capabilities. It is disappointing that the authors have not made more effort to update this edition to reflect the widespread usage of R. This free statistical software environment (R Development Core Team, 2009) has numerous built-in functions for nonparametric tests (eg, wilcoxon. test and kruskal. test). Specialized R packages such as coin (Hothorn, van de Wiel, and Zeileis 2008), boot (Canty and Ripley, 2009), and sm (Bowman and Azzalini, 2010) are also freely available for permutation tests, bootstrapping, and smoothing for nonparametric regression and density estimation, respectively. The authors instead refer almost exclusively to StatXact (Mehta and Patel, 2007), a rather expensive (US $995 academic) alternative that is unlikely to be purchased by undergraduates or other nonspecialists. Furthermore, the authors neither include instructions