With humor, extraordinary clarity, and carefully paced explanations and examples, Bruce Thompson shows readers how to use the latest techniques for interpreting research outcomes as well as how to make statistical decisions that result in better research. Utilizing the general linear model to demonstrate how different statistical methods are related to each other, Thompson integrates a broad array of methods involving only a single dependent variable, ranging from classical and robust location descriptive statistics, through effect sizes, and on through ANOVA, multiple regression, loglinear analysis and logistic regression. Special features include SPSS and Excel demonstrations that offer opportunities, in the book’s datasets and on Thompson’s website, for further exploration of statistical dynamics.
Table of Contents:
Preface
1. Introductory Terms and Concepts
Definitions of Some Basic Terms
Levels of Scale
Some Experimental Design Considerations
Some Key Concepts
Reflection Problems
2. Location
Reasonable Expectations for Statistics
Location Concepts
Three Classical Location Descriptive Statistics
Four Criteria for Evaluating Statistics
Two Robust Location Statistics
Some Key Concepts
Reflection Problems
3. Dispersion
Quality of Location Descriptive Statistics
Important in Its Own Right
Measures of Score Spread
Variance
Situation-Specific Maximum Dispersion
Robust Dispersion Descriptive Statistics
Standardized Score World
Some Key Concepts
Reflection Problems
4. Shape
Two Shape Descriptive Statistics
Normal Distributions
Two Additional Univariate Graphics
Some Key Concepts
Reflection Problems
5. Bivariate Relationships
Pearson's r
Three Features of r
Three Interpretation Contextual Factors
Psychometrics of the Pearson r
Spearman's rho
Two Other r -Equivalent Correlation Coefficients
Bivariate Normality
Some Key Concepts
Reflection Problems
6. Statistical Significance
Sampling Distributions
Hypothesis Testing
Properties of Sampling Distributions
Standard Error/Sampling Error
Test Statistics
Statistical Precision and Power
pCALCULATED
Some Key Concepts
Reflection Problems
7. Practical Significance
Effect Sizes
Confidence Intervals
Confidence Intervals for Effect Sizes
Some Key Concepts
Reflection Problems
8. Multiple Regression Analysis: Basic GLM Concepts
Purposes of Regression
Simple Linear Prediction
Case #1: Perfectly Uncorrelated Predictors
Case #2: Correlated Predictors, No Suppressor
Effects
Case #3: Correlated Predictors, Suppressor
Effects Present
b Weights versus Structure Coefficients
A Final Comment on Collinearity
Some Key Concepts
Reflection Problems
9. A GLM Interpretation Rubric
Do I Have Anything?
Where Does My Something Originate?
Stepwise Methods
Invoking Some Alternative Models
About the Author :
Bruce Thompson is Distinguished Professor of Educational Psychology and Distinguished Professor of Library Sciences, Texas A&M University, and Adjunct Professor of Family and Community Medicine, Baylor College of Medicine, Houston. He is the coeditor of the teaching, learning, and human development section of the American Educational Research Journal and past editor of Educational and Psychological Measurement, the series Advances in Social Science Methodology, and two other journals. He is the author or editor of 10 books, has written over 175 research articles, and has made contributions that have been influential in promoting greater emphasis on effect size reporting and interpretation, and improved understanding of score reliability.
Review :
'An important contribution to improving data analysis and interpretation methods in the social sciences. This book treats many important topics that are ignored in most statistics books... The Statistical Significance chapter is excellent and one of the crown jewels of this book. This is the kind of understanding of significance testing that we should seek to impart to students and researchers. No other stat book, to my knowledge, has a chapter like this one. This is the direction in which things are moving today.' "- Frank L. Schmidt, College of Business, University of Iowa"" " 'This book would be suited for professional and/or class use. The benefit for professional use is the vast number of resources and references cited throughout. This wealth of seminal works from some of the best minds in the field of statistics would provide professionals with supplemental knowledge and information on many crucial statistical issues that have recently been proposed, debated, challenged, and mandated (APA).'" - Victoria Rodlin, Statistical Consultant (former faculty, Department of Psychology, California State University, Fullerton)"
