Behavioral Data Analysis with R and Python

Spis treści ebooka

• Preface
  • Who This Book Is For
  • Who This Book Is Not For
  • R and Python Code
    • Code Environments
    • Code Conventions
    • Functional-Style Programming 101
    • Using Code Examples
  • Navigating This Book
  • Conventions Used in This Book
  • OReilly Online Learning
  • How to Contact Us
  • Acknowledgments
• I. Understanding Behaviors
• 1. The Causal-Behavioral Framework for Data Analysis
  • Why We Need Causal Analytics to Explain Human Behavior
    • The Different Types of Analytics
    • Human Beings Are Complicated
  • Confound It! The Hidden Dangers of Letting Regression Sort It Out
    • Data
    • Why Correlation Is Not Causation: A Confounder in Action
    • Too Many Variables Can Spoil the Broth
  • Conclusion
• 2. Understanding Behavioral Data
  • A Basic Model of Human Behavior
    • Personal Characteristics
      • Collecting data and ethical considerations
    • Cognition and Emotions
      • Collecting data and ethical considerations
    • Intentions
      • Collecting data and ethical considerations
    • Actions
      • Collecting data and ethical considerations
    • Business Behaviors
      • Collecting data
  • How to Connect Behaviors and Data
    • Develop a Behavioral Integrity Mindset
    • Distrust and Verify
    • Identify the Category
    • Refine Behavioral Variables
    • Understand the Context
  • Conclusion
• II. Causal Diagrams and Deconfounding
• 3. Introduction to Causal Diagrams
  • Causal Diagrams and the Causal-Behavioral Framework
    • Causal Diagrams Represent Behaviors
    • Causal Diagrams Represent Data
  • Fundamental Structures of Causal Diagrams
    • Chains
      • Collapsing chains
      • Expanding chains
    • Forks
    • Colliders
  • Common Transformations of Causal Diagrams
    • Slicing/Disaggregating Variables
    • Aggregating Variables
    • What About Cycles?
      • Understanding cycles: Substitution effects and feedback loops
      • Managing cycles
    • Paths
  • Conclusion
• 4. Building Causal Diagrams from Scratch
  • Business Problem and Data Setup
    • Data and Packages
    • Understanding the Relationship of Interest
  • Identify Candidate Variables to Include
    • Actions
    • Intentions
    • Cognition and Emotions
    • Personal Characteristics
      • Traits
      • Demographic variables
    • Business Behaviors
    • Time Trends
  • Validate Observable Variables to Include Based on Data
    • Relationships Between Numeric Variables
    • Relationships Between Categorical Variables
    • Relationships Between Numeric and Categorical Variables
  • Expand Causal Diagram Iteratively
    • Identify Proxies for Unobserved Variables
    • Identify Further Causes
    • Iterate
  • Simplify Causal Diagram
  • Conclusion
• 5. Using Causal Diagrams to Deconfound Data Analyses
  • Business Problem: Ice Cream and Bottled Water Sales
  • The Disjunctive Cause Criterion
    • Definition
    • First Block
    • Second Block
  • The Backdoor Criterion
    • Definitions
    • First Block
    • Second Block
  • Conclusion
• III. Robust Data Analysis
• 6. Handling Missing Data
  • Data and Packages
  • Visualizing Missing Data
    • Amount of Missing Data
    • Correlation of Missingness
  • Diagnosing Missing Data
    • Causes of Missingness: Rubins Classification
    • Diagnosing MCAR Variables
    • Diagnosing MAR Variables
    • Diagnosing MNAR Variables
    • Missingness as a Spectrum
  • Handling Missing Data
    • Introduction to Multiple Imputation (MI)
    • Default Imputation Method: Predictive Mean Matching
    • From PMM to Normal Imputation (R Only)
    • Adding Auxiliary Variables
    • Scaling Up the Number of Imputed Data Sets
  • Conclusion
• 7. Measuring Uncertainty with the Bootstrap
  • Intro to the Bootstrap: Polling Oneself Up
    • Packages
    • The Business Problem: Small Data with an Outlier
    • Bootstrap Confidence Interval for the Sample Mean
    • Bootstrap Confidence Intervals for Ad Hoc Statistics
  • The Bootstrap for Regression Analysis
  • When to Use the Bootstrap
    • Conditions for the Traditional Central Estimate to Be Sufficient
    • Conditions for the Traditional CI to Be Sufficient
    • Determining the Number of Bootstrap Samples
  • Optimizing the Bootstrap in R and Python
    • R: The boot Package
    • Python Optimization
  • Conclusion
• IV. Designing and Analyzing Experiments
• 8. Experimental Design: The Basics
  • Planning the Experiment: Theory of Change
    • Business Goal and Target Metric
      • Business goal
      • Target metric
      • Pitfalls of poor target metrics
    • Intervention
    • Behavioral Logic
  • Data and Packages
  • Determining Random Assignment and Sample Size/Power
    • Random Assignment
      • Code implementation
      • Pitfalls of random assignment
        • Random assignment timing
        • Random assignment level
    • Sample Size and Power Analysis
      • A little bit of statistics theory without math
      • Traditional power analysis
      • Power analysis without statistics: Bootstrap simulations
        • Connecting simulations and statistical theory
        • Writing our analysis code
        • Power simulation
  • Analyzing and Interpreting Experimental Results
  • Conclusion
• 9. Stratified Randomization
  • Planning the Experiment
    • Business Goal and Target Metric
    • Definition of the Intervention
    • Behavioral Logic
    • Data and Packages
  • Determining Random Assignment and Sample Size/Power
    • Random Assignment
      • Random assignment level
      • Standard randomization
      • Stratified randomization
    • Power Analysis with Bootstrap Simulations
      • Single simulation
      • Simulations at scale
      • Understanding the power/significance trade-off
  • Analyzing and Interpreting Experimental Results
    • Intention-to-Treat Estimate for Encouragement Intervention
    • Complier Average Causal Estimate for Mandatory Intervention
  • Conclusion
• 10. Cluster Randomization and Hierarchical Modeling
  • Planning the Experiment
    • Business Goal and Target Metric
    • Definition of the Intervention
    • Behavioral Logic
  • Data and Packages
  • Introduction to Hierarchical Modeling
    • R Code
    • Python Code
  • Determining Random Assignment and Sample Size/Power
    • Random Assignment
    • Power Analysis
      • Using permutations when randomness is limited
      • Code for permutations
      • Power curve
  • Analyzing the Experiment
  • Conclusion
• V. Advanced Tools in Behavioral Data Analysis
• 11. Introduction to Moderation
  • Data and Packages
  • Behavioral Varieties of Moderation
    • Segmentation
      • Segmenting observational data
      • Segmenting experimental data
    • Interactions
    • Nonlinearities
  • How to Apply Moderation
    • When to Look for Moderation?
      • Including moderation in the experimental design stage
      • Including moderation in the data analysis stage
      • Nonlinearities
    • Multiple Moderators
      • Parallel moderators
      • Interacting moderators
    • Validating Moderation with Bootstrap
    • Interpreting Individual Coefficients
      • Setting meaningful reference points
      • Calculating effects at the level of business decisions
  • Conclusion
• 12. Mediation and Instrumental Variables
  • Mediation
    • Understanding Causal Mechanisms
    • Causal Biases
    • Identifying Mediation
    • Measuring Mediation
      • Total effect
      • Mediated effect
      • Direct effect
      • When the mediator is a binary variable
  • Instrumental Variables
    • Data
    • Packages
    • Understanding and Applying IVs
      • Step 1: Leftmost relationship
      • Step 2: Total effect
      • Step 3: Relationship of interest
    • Measurement
      • Python code
      • R code
    • Applying IVs: Frequently Asked Questions
  • Conclusion
• Bibliography
• Index

pokaż cały spis treści

ukryj recenzje