Quantitative Data Analysis: Doing Social Research to Test Ideas

ISBN-13: 9780470380031 / Angielski / Miękka / 2009 / 480 str.

Donald J Treiman

Opis

This book is an accessible introduction to quantitative data analysis, concentrating on the key issues facing those new to research, such as how to decide which statistical procedure is suitable, and how to interpret the subsequent results. Each chapter includes illustrative examples and a set of exercises that allows readers to test their understanding of the topic. The book, written for graduate students in the social sciences, public health, and education, offers a practical approach to making sociological sense out of a body of quantitative data. The book also will be useful to more experienced researchers who need a readily accessible handbook on quantitative methods. The author has posted stata files, updates and data sets at his website http: //tinyurl.com/Treiman-stata-files-data-sets.

Spis treści

Tables, Figures, Exhibits, and Boxes xi

Preface xxiii

The Author xxvii

Introduction xxix

1 CROSS–TABULATIONS 1

What This Chapter Is About 1

Introduction to the Book via a Concrete Example 2

Cross–Tabulations 8

What This Chapter Has Shown 19

2 MORE ON TABLES 21

What This Chapter Is About 21

The Logic of Elaboration 22

Suppressor Variables 25

Additive and Interaction Effects 26

Direct Standardization 28

A Final Note on Statistical Controls Versus Experiments 43

What This Chapter Has Shown 45

3 STILL MORE ON TABLES 47

What This Chapter Is About 47

Reorganizing Tables to Extract New Information 48

When to Percentage a Table "Backwards" 50

Cross–Tabulations in Which the Dependent Variable Is Represented by a Mean 52

Index of Dissimilarity 58

Writing About Cross–Tabulations 61

What This Chapter Has Shown 63

4 ON THE MANIPULATION OF DATA BY COMPUTER 65

What This Chapter Is About 65

Introduction 66

How Data Files Are Organized 67

Transforming Data 72

What This Chapter Has Shown 80

Appendix 4.A Doing Analysis Using Stata 80

Tips on Doing Analysis Using Stata 80

Some Particularly Useful Stata 10.0 Commands 84

5 INTRODUCTION TO CORRELATION AND REGRESSION (ORDINARY LEAST SQUARES) 87

What This Chapter Is About 87

Introduction 88

Quantifying the Size of a Relationship: Regression Analysis 89

Assessing the Strength of a Relationship: Correlation Analysis 91

The Relationship Between Correlation and Regression Coeffi cients 94

Factors Affecting the Size of Correlation (and Regression) Coeffi cients 94

Correlation Ratios 99

What This Chapter Has Shown 102

6 INTRODUCTION TO MULTIPLE CORRELATION AND REGRESSION (ORDINARY LEAST SQUARES) 103

What This Chapter Is About 103

Introduction 104

A Worked Example: The Determinants of Literacy in China 113

Dummy Variables 120

A Strategy for Comparisons Across Groups 124

A Bayesian Alternative for Comparing Models 133

Independent Validation 135

What This Chapter Has Shown 136

7 MULTIPLE REGRESSION TRICKS: TECHNIQUES FOR HANDLING SPECIAL ANALYTIC PROBLEMS 139

What This Chapter Is About 139

Nonlinear Transformations 140

Testing the Equality of Coeffi cients 147

Trend Analysis: Testing the Assumption of Linearity 149

Linear Splines 152

Expressing Coeffi cients as Deviations from the Grand Mean (Multiple Classifi cation Analysis) 164

Other Ways of Representing Dummy Variables 166

Decomposing the Difference Between Two Means 172

What This Chapter Has Shown 179

8 MULTIPLE IMPUTATION OF MISSING DATA 181

What This Chapter Is About 181

Introduction 182

A Worked Example: The Effect of Cultural Capital on Educational Attainment in Russia 187

What This Chapter Has Shown 194

9 SAMPLE DESIGN AND SURVEY ESTIMATION 195

What This Chapter Is About 195

Survey Samples 196

Conclusion 223

What This Chapter Has Shown 224

10 REGRESSION DIAGNOSTICS 225

What This Chapter Is About 225

Introduction 226

A Worked Example: Societal Differences in Status Attainment 229

Robust Regression 237

Bootstrapping and Standard Errors 238

What This Chapter Has Shown 240

11 SCALE CONSTRUCTION 241

What This Chapter Is About 241

Introduction 242

Validity 242

Reliability 243

Scale Construction 246

Errors–in–Variables Regression 258

What This Chapter Has Shown 261

12 LOG–LINEAR ANALYSIS 263

What This Chapter Is About 263

Introduction 264

Choosing a Preferred Model 265

Parsimonious Models 277

A Bibliographic Note 294

What This Chapter Has Shown 295

Appendix 12.A Derivation of the Effect Parameters 295

Appendix 12.B Introduction to Maximum Likelihood Estimation 297

Mean of a Normal Distribution 298

Log–Linear Parameters 299

13 BINOMIAL LOGISTIC REGRESSION 301

What This Chapter Is About 301

Introduction 302

Relation to Log–Linear Analysis 303

A Worked Logistic Regression Example:

Predicting Prevalence of Armed Threats 304

A Second Worked Example: Schooling Progression Ratios in Japan 314

A Third Worked Example (Discrete–Time Hazard–Rate Models): Age at First Marriage 318

A Fourth Worked Example (Case–Control Models):

Who Was Appointed to a Nomenklatura Position in Russia? 327

What This Chapter Has Shown 329

Appendix 13.A Some Algebra for Logs and Exponents 330

Appendix 13.B Introduction to Probit Analysis 330

14 MULTINOMIAL AND ORDINAL LOGISTIC REGRESSION AND TOBIT REGRESSION 335

What This Chapter Is About 335

Multinomial Logit Analysis 336

Ordinal Logistic Regression 342

Tobit Regression (and Allied Procedures) for Censored Dependent Variables 353

Other Models for the Analysis of Limited Dependent Variables 360

What This Chapter Has Shown 361

15 IMPROVING CAUSAL INFERENCE: FIXED EFFECTS AND RANDOM EFFECTS MODELING 363

What This Chapter Is About 363

Introduction 364

Fixed Effects Models for Continuous Variables 365

Random Effects Models for Continuous Variables 371

A Worked Example: The Determinants of Income in China 372

Fixed Effects Models for Binary Outcomes 375

A Bibliographic Note 380

What This Chapter Has Shown 380

16 FINAL THOUGHTS AND FUTURE DIRECTIONS: RESEARCH DESIGN AND INTERPRETATION ISSUES 381

What this Chapter is About 381

Research Design Issues 382

The Importance of Probability Sampling 397

A Final Note: Good Professional Practice 400

What This Chapter Has Shown 405

Appendix A: Data Descriptions and Download Locations for the Data Used in This Book 407

Appendix B: Survey Estimation with the General Social Survey 411

References 417

Index 431

O autorze

Donald J. Treiman, PhD, is Distinguished Professor, Department of Sociology, University of California, Los Angeles. He was also, until recently, director of the California Center for Population Research.

Blurb

Advance Praise for Quantitative Data Analysis

"This is a first–rate introduction to modern social science research methods by an acknowledged master and brilliant teacher of the art. Treiman simplifies complexity and offers a mountain of practical tips and best practices. Anyone who wants to learn to do modern quantitative social research (and anyone who plans to teach it) has much to gain from this book."
Ross M. Stolzenberg, professor of sociology, University of Chicago

"Packed with examples using interesting data sets, this book doesn′t just teach statistics– it teaches how to use statistics to answer questions about the social world. In simple language, from simple percentage tables to fixed–effects models, including how to do the runs in Stata, and packed with examples from real data on how to make and interpret tables, Treiman shows students how to do the craft they are too often never taught."
Paula England, professor of sociology, Stanford University

"After a distinguished career, Don Treiman shares his trade secrets in this remarkable book, which should be read by everyone wishing to master the art of theoretically informed quantitative research."
Yu Xie, Otis Dudley Duncan Distinguished University Professor of Sociology and Statistics, University of Michigan and coauthor, Statistical Methods for Categorical Data Analysis

"This book captures the wisdom arising from Professor Treiman′s many years of data analysis experience. It provides the applied perspective and practical advice that are missing in many books on statistical models."
Kenneth A. Bollen, H. R. Immerwahr Distinguished Professor of Sociology and director, Howard W. Odum Institute for Research in Social Science, University of North Carolina at Chapel Hill

"There are lots of books on the market that teach statistics. This book, written by a distinguished social scientist, achieves a more ambitious goal; it teaches students how to use statistics to do first–rate quantitative research."
Thomas A. DiPrete, professor of sociology, Columbia University

"Treiman′s wealth of teaching and research experience shows in this book. He delivers new insights into familiar methods and gives us tools we need to solve the puzzles that inevitably arise in real research."
Michael Hout, professor of sociology and demography, University of California, Berkeley, and member, National Academy of Sciences

"Having taken Don Treiman′s course at UCLA, I highly recommend the book to anyone who wants practical instruction in how survey research is done."
Christine R. Schwartz, assistant professor, Department of Sociology, University of Wisconsin, Madison