A Primer on Effect Sizes, Simple Research Designs, and Confidence Intervals was designed to help individuals learn to calculate effect sizes for their research designs. Effect sizes allow a clinician or researcher to determine the effect of a treatment. For example, an effect size of zero would indicate that the treatment had no effect, but generally effect sizes allow researchers to see the degree of effect of some treatment or intervention. Often, researchers and clinicians are not aware that effect sizes are connected to research designs. For years, statisticians have been aware of limits of null hypothesis significance testing (NHST). The Wilkinson Task Force (Wilkinson & Task Force on Statistical Inference, 1999) recommended that researchers report effect sizes and confidence intervals in addition to null hypothesis significance testing (NHST). The purpose of this book is to provide the connection among effect sizes, confidence intervals, and simple research designs. Also, some commonly used univariate and multivariate statistics are covered. Regression discontinuity designs, simple moderation and mediation designs, power analysis, and fit indices as effect sizes measure are presented. All calculations are demonstrated through a calculator and statistical packages such as Microsoft Excel, SPSS, SAS, Hayes' Process Analysis, and EQS. This book covers more than 25 effect sizes that are connected to simple research designs. It will be of interest to students taking a statistics class, research methods class, or research design class. Unlike many texts within this area, the current test will give students or researchers the understanding of how to calculate effect sizes with a simple calculator or with a few commands from statistical software programs. Hence, mathematical ability is not a prerequisite for this text. This text provides a nonmathematical treatment of effect sizes within the context of research designs. Finally, to aid understanding, critical material is repeated throughout this book. Preface 1. Introduction: r and d effect sizes History of Effect Sizes Reliability for Unidimensional Scales Reliability for Multidimensional Scales Validity Face Validity Content Validity Criterion Validity Predictive Validity Construct Validity Effect Sizes Definition of Multivariate Statistics Confidence Intervals Testing Calculated Validity Coefficients Against Hypothesized Values Standard Error of Estimate Confidence Intervals Around Validity A Practical Example of a One Sample Case Confidence Interval Discussion The Effect Size r Counternull Value of an Effect Meta-Analysis Confidence Intervals Around the Effect Size r Using SAS for Calculating d Effect Size Confidence Intervals SAS Control Lines to Compute an Exact 95% Confidence Interval for Effect Size d For Two Groups of Participants SAS Control Lines to Compute an Exact 95% Confidence Interval for Effect Size d For One Group of Participants Chapter Summary Practice Problems Answer to Practice Problems 2. Confidence Intervals for A Single Mean Problems Answers 3. Effect Size and Confidence Interval for Differences Between Two Means (Between Group Research Designs) Regression Discontinuity Designs 4. One-Group Pre-test Post-test Design Problems Answers 5. Effect Size for One-Way Analysis of Variance or Three or More Group means Test of Between-Subjects Effects Test of Homogeneity of Variances SAS Commands for 95% Confidence Interval for Eta Squared Welch and Brown-Forsythe Test for Unequal Variances Factorial Designs Fixed Effects, Random Effects, and Mixed Model Analysis of Variance (ANOVA) Disproportional Cell Size or Unbalanced Factorial Designs Three-Way Analysis of Variance (ANOVA) Multiple Comparisons Post Hoc Procedures Nested ANOVA One-Way Analysis of Covariance (ANCOVA) 6. Correlations as Effect Sizes 7. Effect Sizes for Two or More Predictors and One Dependent Variable Multiple Regression Schematic Design for Two-Predictor Case Analysis of Variance Table for Regression Multiple Regression Broken Down into Sums of Squares Assumptions of Multiple Regression Suppressor Variables in Multiple Regression Structure Coefficients within Multiple Regression Interaction Effects within Multiple Regression Cross-Validation Formulas with Multiple Regression Logistic Regression 8. Effect Sizes for Two or More Predictors and Two or More Dependent Variables Multivariate Regression 9. Effect Size for Two-Group Multivariate Analysis of Variance Discussion 10. Moderation and Meditation effects 11. Power Analysis A Priori and Post Hoc Estimations of Power 12. Path Analysis and Effect Sizes 13. Fit Indices as Effect Size Measures Book Summary
References Name Index Subject Index
|