※ 引述《pingo87131 (pingo87131)》之銘言： : 想請問有沒有專門在講觀念 方法論的書呢 : 或是講一些統計分析容易解釋錯誤的例子 : 我現在初統學完 各種檢定方法大致上有了個認識 : 但是看PAPER的時候 對那些跑出來的數怎麼做解釋都不是很能理解 : 自己是文組科系 很怕自己以後做出來因為觀念薄弱而做了錯誤的分析 : 不知道有沒有這類型的書呢 非常感謝~ Phillip I. Good & James W. Hardin (2006) COMMON ERRORS IN STATISTICS(AND HOW TO AVOID THEM) John Wiley & Sons, Inc. ISBN-13: 978-0-471-79431-8 這個標題好像就是你要的？再節錄一點前言 The primary objective of the opening chapter is to describe the main sources of error and provide a preliminary prescription for avoiding them. The hypothesis formulation—data gathering—hypothesis testing and estimate cycle is introduced, and the rationale for gathering additional data before attempting to test after-the-fact hypotheses is detailed. Chapter 2 places our work in the context of decision theory. We emphasize the importance of providing an interpretation of each and every potential outcome in advance of consideration of actual data. Chapter 3 focuses on study design and data collection, for failure at the planning stage can render all further efforts valueless. The work of Berger and his colleagues on selection bias is given particular emphasis. Desirable features of point and interval estimates are detailed in Chapter 4 along with procedures for deriving estimates in a variety of practical situations.This chapter also serves to debunk several myths surrounding estimation procedures. Chapter 5 reexamines the assumptions underlying testing hypotheses. We review the impacts of violations of assumptions and detail the procedures to follow when making 2- and k-sample comparisons. In addition, we cover the procedures for analyzing contingency tables and 2-way experimental designs if standard assumptions are violated. Chapter 6 is devoted to the value and limitations of Bayes’ theorem, meta-analysis, and resampling methods. Chapter 7 lists the essentials of any report that will utilize statistics, debunks the myth of the “standard” error, and describes the value and limitations of p-values and confidence intervals for reporting results. Practical significance is distinguished from statistical significance and inductionis distinguished from deduction. Chapter 8 covers much the same material,but the viewpoint is that of the report reader rather than the report writer. Of particular importance is a section on interpreting computer output. Twelve rules for more effective graphic presentations are given in Chapter 9 along with numerous examples of the right and wrong ways to maintain reader interest while communicating essential statistical information. Chapters 10 through 13 are devoted to model building and to the assumptions and limitations of a multitude of regression methods and data mining techniques. A distinction is drawn between goodness of fit and prediction, and the importance of model validation is emphasized. Seminal articles by David Freedman and Gail Gong are reprinted. 台大圖書館有此書電子版可供下載 --

※ 發信站: 批踢踢實業坊(ptt.cc) ◆ From: 140.112.5.16 ※ 編輯: primavere 來自: 140.112.5.16 (08/22 23:52)