Beyond Multiple Linear Regression
Kaup valmöguleikar
Beyond Multiple Linear Regression: Applied Generalized Linear Models and Multilevel Models in R is designed for undergraduate students who have successfully completed a multiple linear regression course, helping them develop an expanded modeling toolkit that includes non-normal responses and correlated structure. Even though there is no mathematical prerequisite, the authors still introduce fairly sophisticated topics such as likelihood theory, zero-inflated Poisson, and parametric bootstrapping in an intuitive and applied manner.
The case studies and exercises feature real data and real research questions; thus, most of the data in the textbook comes from collaborative research conducted by the authors and their students, or from student projects. Every chapter features a variety of conceptual exercises, guided exercises, and open-ended exercises using real data. After working through this material, students will develop an expanded toolkit and a greater appreciation for the wider world of data and statistical modeling.
Nánar um bókina
- Taylor & Francis
- 9780429527333
- 9780367680442
- ePub
- 1
- Paul Roback; Julie Legler
- English
- 2021-01-14
- 100
- 2
- 2
Kaflar
- Cover
- Half Title
- Series
- Title
- Copyright
- Dedication
- Contents
- Preface
- 1 Review of Multiple Linear Regression
- 1.1 Learning Objectives
- 1.2 Introduction to Beyond Multiple Linear Regression
- 1.3 Assumptions for Linear Least Squares Regression
- 1.3.1 Cases Without Assumption Violations
- 1.3.2 Cases With Assumption Violations
- 1.4 Review of Multiple Linear Regression
- 1.4.1 Case Study: Kentucky Derby
- 1.5 Initial Exploratory Analyses
- 1.5.1 Data Organization
- 1.5.2 Univariate Summaries
- 1.5.3 Bivariate Summaries
- 1.6 Multiple Linear Regression Modeling
- 1.6.1 Simple Linear Regression with a Continuous Predictor
- 1.6.2 Linear Regression with a Binary Predictor
- 1.6.3 Multiple Linear Regression with Two Predictors
- 1.6.4 Inference in Multiple Linear Regression: Normal Theory
- 1.6.5 Inference in Multiple Linear Regression: Bootstrapping
- 1.6.6 Multiple Linear Regression with an Interaction Term
- 1.6.7 Building a Multiple Linear Regression Model
- 1.7 Preview of Remaining Chapters
- 1.7.1 Soccer
- 1.7.2 Elephant Mating
- 1.7.3 Parenting and Gang Activity
- 1.7.4 Crime
- 1.8 Exercises
- 1.8.1 Conceptual Exercises
- 1.8.2 Guided Exercises
- 1.8.3 Open-Ended Exercises
- 2 Beyond Least Squares: Using Likelihoods
- 2.1 Learning Objectives
- 2.2 Case Study: Does Sex Run in Families?
- 2.2.1 Research Questions
- 2.3 Model 0: Sex Unconditional, Equal Probabilities
- 2.4 Model 1: Sex Unconditional, Unequal Probabilities
- 2.4.1 What Is a Likelihood?
- 2.4.2 Finding MLEs
- 2.4.2.1 Graphically approximating an MLE
- 2.4.2.2 Numerically approximating an MLE
- 2.4.2.3 MLEs using calculus (optional)
- 2.4.2.4 How does sample size affect the likelihood?
- 2.4.3 Summary
- 2.4.4 Is a Likelihood a Probability Function? (optional)
- 2.5 Model 2: Sex Conditional
- 2.5.1 Model Specification
- 2.5.2 Application to Hypothetical Data
- 2.6 Case Study: Analysis of the NLSY Data
- 2.6.1 Model Building Plan
- 2.6.2 Exploratory Data Analysis
- 2.6.3 Likelihood for the Sex Unconditional Model
- 2.6.4 Likelihood for the Sex Conditional Model
- 2.6.5 Model Comparisons
- 2.6.5.1 Nested models
- 2.7 Model 3: Stopping Rule Model (waiting for a boy)
- 2.7.1 Non-nested Models
- 2.8 Summary of Model Building
- 2.9 Likelihood-Based Methods
- 2.10 Likelihoods and This Course
- 2.11 Exercises
- 2.11.1 Conceptual Exercises
- 2.11.2 Guided Exercises
- 2.11.3 Open-Ended Exercises
- 3 Distribution Theory
- 3.1 Learning Objectives
- 3.2 Introduction
- 3.3 Discrete Random Variables
- 3.3.1 Binary Random Variable
- 3.3.2 Binomial Random Variable
- 3.3.3 Geometric Random Variable
- 3.3.4 Negative Binomial Random Variable
- 3.3.5 Hypergeometric Random Variable
- 3.3.6 Poisson Random Variable
- 3.4 Continuous Random Variables
- 3.4.1 Exponential Random Variable
- 3.4.2 Gamma Random Variable
- 3.4.3 Normal (Gaussian) Random Variable
- 3.4.4 Beta Random Variable
- 3.5 Distributions Used in Testing
- 3.5.1 χ2 Distribution
- 3.5.2 Student's t-Distribution
- 3.5.3 F-Distribution
- 3.6 Additional Resources
- 3.7 Exercises
- 3.7.1 Conceptual Exercises
- 3.7.2 Guided Exercises
- 4 Poisson Regression
- 4.1 Learning Objectives
- 4.2 Introduction to Poisson Regression
- 4.2.1 Poisson Regression Assumptions
- 4.2.2 A Graphical Look at Poisson Regression
- 4.3 Case Studies Overview
- 4.4 Case Study: Household Size in the Philippines
- 4.4.1 Data Organization
- 4.4.2 Exploratory Data Analyses
- 4.4.3 Estimation and Inference
- 4.4.4 Using Deviances to Compare Models
- 4.4.5 Using Likelihoods to Fit Models (optional)
- 4.4.6 Second Order Model
- 4.4.7 Adding a Covariate
- 4.4.8 Residuals for Poisson Models (optional)
- 4.4.9 Goodness-of-Fit
- 4.5 Linear Least Squares vs. Poisson Regression
- 4.6 Case Study: Campus Crime
- 4.6.1 Data Organization
- 4.6.2 Exploratory Data Analysis
- 4.6.3 Accounting for Enrollment
- 4.7 Modeling Assumptions
- 4.8 Initial Models
- 4.8.1 Tukey's Honestly Significant Differences
- 4.9 Overdispersion
- 4.9.1 Dispersion Parameter Adjustment
- 4.9.2 No Dispersion vs. Overdispersion
- 4.9.3 Negative Binomial Modeling
- 4.10 Case Study: Weekend Drinking
- 4.10.1 Research Question
- 4.10.2 Data Organization
- 4.10.3 Exploratory Data Analysis
- 4.10.4 Modeling
- 4.10.5 Fitting a ZIP Model
- 4.10.6 The Vuong Test (optional)
- 4.10.7 Residual Plot
- 4.10.8 Limitations
- 4.11 Exercises
- 4.11.1 Conceptual Exercises
- 4.11.2 Guided Exercises
- 4.11.3 Open-Ended Exercises
- 5 Generalized Linear Models: A Unifying Theory
- 5.1 Learning Objectives
- 5.2 One-Parameter Exponential Families
- 5.2.1 One-Parameter Exponential Family: Poisson
- 5.2.2 One-Parameter Exponential Family: Normal
- 5.3 Generalized Linear Modeling
- 5.4 Exercises
- 6 Logistic Regression
- 6.1 Learning Objectives
- 6.2 Introduction to Logistic Regression
- 6.2.1 Logistic Regression Assumptions
- 6.2.2 A Graphical Look at Logistic Regression
- 6.3 Case Studies Overview
- 6.4 Case Study: Soccer Goalkeepers
- 6.4.1 Modeling Odds
- 6.4.2 Logistic Regression Models for Binomial Responses
- 6.4.3 Theoretical Rationale (optional)
- 6.5 Case Study: Reconstructing Alabama
- 6.5.1 Data Organization
- 6.5.2 Exploratory Analyses
- 6.5.3 Initial Models
- 6.5.4 Tests for Significance of Model Coefficients
- 6.5.5 Confidence Intervals for Model Coefficients
- 6.5.6 Testing for Goodness-of-Fit
- 6.5.7 Residuals for Binomial Regression
- 6.5.8 Overdispersion
- 6.5.9 Summary
- 6.6 Linear Least Squares vs. Binomial Regression
- 6.7 Case Study: Trying to Lose Weight
- 6.7.1 Data Organization
- 6.7.2 Exploratory Data Analysis
- 6.7.3 Initial Models
- 6.7.4 Drop-in-Deviance Tests
- 6.7.5 Model Discussion and Summary
- 6.8 Exercises
- 6.8.1 Conceptual Exercises
- 6.8.2 Guided Exercises
- 6.8.3 Open-Ended Exercises
- 7 Correlated Data
- 7.1 Learning Objectives
- 7.2 Introduction
- 7.3 Recognizing Correlation
- 7.4 Case Study: Dams and Pups
- 7.5 Sources of Variability
- 7.6 Scenario 1: No Covariates
- 7.7 Scenario 2: Dose Effect
- 7.8 Case Study: Tree Growth
- 7.8.1 Format of the Data Set
- 7.8.2 Sources of Variability
- 7.8.3 Analysis Preview: Accounting for Correlation
- 7.9 Summary
- 7.10 Exercises
- 7.10.1 Conceptual Exercises
- 7.10.2 Guided Exercises
- 7.10.3 Note on Correlated Binary Outcomes
- 8 Introduction to Multilevel Models
- 8.1 Learning Objectives
- 8.2 Case Study: Music Performance Anxiety
- 8.3 Initial Exploratory Analyses
- 8.3.1 Data Organization
- 8.3.2 Exploratory Analyses: Univariate Summaries
- 8.3.3 Exploratory Analyses: Bivariate Summaries
- 8.4 Two-Level Modeling: Preliminary Considerations
- 8.4.1 Ignoring the Two-Level Structure (not recommended)
- 8.4.2 A Two-Stage Modeling Approach (better but imperfect)
- 8.5 Two-Level Modeling: A Unified Approach
- 8.5.1 Our Framework
- 8.5.2 Random vs. Fixed Effects
- 8.5.3 Distribution of Errors: Multivariate Normal
- 8.5.4 Technical Issues when Testing Parameters (optional)
- 8.5.5 An Initial Model with Parameter Interpretations
- 8.6 Building a Multilevel Model
- 8.6.1 Model Building Strategy
- 8.6.2 An Initial Model: Random Intercepts
- 8.7 Binary Covariates at Level One and Level Two
- 8.7.1 Random Slopes and Intercepts Model
- 8.7.2 Pseudo R-squared Values
- 8.7.3 Adding a Covariate at Level Two
- 8.8 Adding Further Covariates
- 8.8.1 Interpretation of Parameter Estimates
- 8.8.2 Model Comparisons
- 8.9 Centering Covariates
- 8.10 A Final Model for Music Performance Anxiety
- 8.11 Modeling Multilevel Structure: Is It Necessary?
- 8.12 Notes on Using R (optional)
- 8.13 Exercises
- 8.13.1 Conceptual Exercises
- 8.13.2 Guided Exercises
- 8.13.3 Open-Ended Exercises
- 9 Two-Level Longitudinal Data
- 9.1 Learning Objectives
- 9.2 Case Study: Charter Schools
- 9.3 Initial Exploratory Analyses
- 9.3.1 Data Organization
- 9.3.2 Missing Data
- 9.3.3 Exploratory Analyses for General Multilevel Models
- 9.3.4 Exploratory Analyses for Longitudinal Data
- 9.4 Preliminary Two-Stage Modeling
- 9.4.1 Linear Trends Within Schools
- 9.4.2 Effects of Level Two Covariates on Linear Time Trends
- 9.4.3 Error Structure Within Schools
- 9.5 Initial Models
- 9.5.1 Unconditional Means Model
- 9.5.2 Unconditional Growth Model
- 9.5.3 Modeling Other Trends over Time
- 9.6 Building to a Final Model
- 9.6.1 Uncontrolled Effects of School Type
- 9.6.2 Add Percent Free and Reduced Lunch as a Covariate
- 9.6.3 A Final Model with Three Level Two Covariates
- 9.6.4 Parametric Bootstrap Testing
- 9.7 Covariance Structure among Observations
- 9.7.1 Standard Covariance Structure
- 9.7.2 Alternative Covariance Structures
- 9.7.3 Non-longitudinal Multilevel Models
- 9.7.4 Final Thoughts Regarding Covariance Structures
- 9.7.5 Details of Covariance Structures (optional)
- 9.8 Notes on Using R (optional)
- 9.9 Exercises
- 9.9.1 Conceptual Exercises
- 9.9.2 Guided Exercises
- 9.9.3 Open-Ended Exercises
- 10 Multilevel Data With More Than Two Levels
- 10.1 Learning Objectives
- 10.2 Case Studies: Seed Germination
- 10.3 Initial Exploratory Analyses
- 10.3.1 Data Organization
- 10.3.2 Exploratory Analyses
- 10.4 Initial Models
- 10.4.1 Unconditional Means
- 10.4.2 Unconditional Growth
- 10.5 Encountering Boundary Constraints
- 10.6 Parametric Bootstrap Testing
- 10.7 Exploding Variance Components
- 10.8 Building to a Final Model
- 10.9 Covariance Structure (optional)
- 10.9.1 Details of Covariance Structures
- 10.10 Notes on Using R (optional)
- 10.11 Exercises
- 10.11.1 Conceptual Exercises
- 10.11.2 Guided Exercises
- 10.11.3 Open-Ended Exercises
- 11 Multilevel Generalized Linear Models
- 11.1 Learning Objectives
- 11.2 Case Study: College Basketball Referees
- 11.3 Initial Exploratory Analyses
- 11.3.1 Data Organization
- 11.3.2 Exploratory Analyses
- 11.4 Two-Level Modeling with a Generalized Response
- 11.4.1 A GLM Approach
- 11.4.2 A Two-Stage Modeling Approach
- 11.4.3 A Unified Multilevel Approach
- 11.5 Crossed Random Effects
- 11.6 Parametric Bootstrap for Model Comparisons
- 11.7 A Final Model for Examining Referee Bias
- 11.8 Estimated Random Effects
- 11.9 Notes on Using R (optional)
- 11.10 Exercises
- 11.10.1 Conceptual Exercises
- 11.10.2 Open-Ended Exercises
- Bibliography
- Index