Common errors in statistics (and how to avoid them) / Phillip I. Good, James W. Hardin.

Format:

Book

Author/Creator:

Good, Phillip I.

Contributor:

Hardin, James W. (James William)

Class of 1953 Fund.

Language:

English

Subjects (All):

Statistics.

Physical Description:

xi, 221 pages : illustrations ; 24 cm

Place of Publication:

Hoboken, NJ : Wiley-Interscience, [2003]

Summary:

A guide to choosing and using the right techniques

High-speed computers and prepackaged statistical routines would seem to take much of the guesswork out of statistical analysis and lend its applications readily accessible to all. Yet, as Phillip Good and James Hardin persuasively argue, statistical software no more makes one a statistician than a scalpel makes one a surgeon. Choosing the proper technique and understanding the analytical context is of paramount importance to the proper application of statistics. The highly readable Common Errors in Statistics (and How to Avoid Them) provides both newly minted academics and professionals who use statistics in their work with a handy field guide to statistical problems and solutions.

Good and Hardin begin their handbook by establishing a mathematically rigorous but readily accessible foundation for statistical procedures. They focus on debunking popular myths, analyzing common mistakes, and instructing readers on how to choose the appropriate statistical technique to address their specific task. A handy checklist is provided to summarize the necessary steps.

Topics covered include: Creating a research plan Formulating a hypothesis Specifying sample size Checking assumptions Interpreting p-values and confidence intervals Building a model Data mining Bayes'Theorem, the bootstrap, and many others

Common Errors in Statistics (and How to Avoid Them) also contains reprints of classic articles from statistical literature to re-examine such bedrock subjects as linear regression, the analysis of variance, maximum likelihood, meta-analysis, and the bootstrap. With a final emphasis on finding solutions and on the great value ofstatistics when applied in the proper context, this book will prove eminently useful to students and professionals in the fields of research, industry, medicine, and government.

Contents:

1. Sources of Error 3

Ad Hoc, Post Hoc Hypotheses 7

2. Hypotheses: The Why of Your Research 11

Null Hypothesis 14

Neyman-Pearson Theory 15

Deduction and Induction 19

Losses 20

Decisions 21

3. Collecting Data 25

Preparation 25

Measuring Devices 26

Determining Sample Size 28

Fundamental Assumptions 32

Experimental Design 33

Four Guidelines 34

Part II Hypothesis Testing and Estimation 39

Prevention 41

Desirable and Not-So-Desirable Estimators 41

Interval Estimates 45

Improved Results 49

5. Testing Hypotheses: Choosing a Test Statistic 51

Comparing Means of Two Populations 53

Comparing Variances 60

Comparing the Means of K Samples 62

Higher-Order Experimental Designs 65

Contingency Tables 70

Inferior Tests 71

Multiple Tests 72

Before You Draw Conclusions 72

6. Strengths and Limitations of Some Miscellaneous Statistical Procedures 77

Bootstrap 78

Bayesian Methodology 79

Meta-Analysis 87

Permutation Tests 89

7. Reporting Your Results 91

Tables 94

Standard Error 95

p Values 100

Confidence Intervals 101

Recognizing and Reporting Biases 102

Reporting Power 104

Drawing Conclusions 104

8. Graphics 107

The Soccer Data 107

Five Rules for Avoiding Bad Graphics 108

One Rule for Correct Usage of Three-Dimensional Graphics 115

One Rule for the Misunderstood Pie Chart 117

Three Rules for Effective Display of Subgroup Information 118

Two Rules for Text Elements in Graphics 121

Multidimensional Displays 123

Choosing Effective Display Elements 123

Choosing Graphical Displays 124

Part III Building A Model 127

9. Univariate Regression 129

Model Selection 129

Estimating Coefficients 137

Further Considerations 138

10. Multivariable Regression 145

Generalized Linear Models 146

Reporting Your Results 149

Building a Successful Model 152

11. Validation 155

Methods of Validation 156

Measures of Predictive Success 159

Long-Term Stability 161.

Notes:

Includes bibliographical references (pages 191-209) and index.

Local Notes:

Acquired for the Penn Libraries with assistance from the Class of 1953 Fund.

ISBN:

0471460680

OCLC:

51769193

Online:

Publisher description