Causal Inference Methods

Causal Agent supports a comprehensive range of causal inference methods, from experimental designs to observational studies. This section provides detailed documentation for each method, including when to use them, their assumptions, and implementation details.

Method Documentation

• Overview of Causal Inference Methods
  • What is Causal Inference?
  • The Fundamental Problem of Causal Inference
  • Method Categories in Causal Agent
  • How Causal Agent Selects Methods
  • The Decision Tree Process
  • Method Assumptions and Validity
  • Best Practices
  • Getting Started
• Method Selection Decision Tree
  • Complete Decision Tree Algorithm
  • Dataset Property Influence Visualization
  • Decision Criteria Explained
  • Step-by-Step Decision Walkthroughs
  • Method Selection Examples
  • Decision Node Documentation
  • Understanding Method Recommendations
  • Customizing Method Selection
  • Validating Method Choice
  • Interactive Tools and Utilities
  • Decision Tree Algorithm Implementation
  • Next Steps
• Experimental Methods
  • Randomized Controlled Trials (RCT)
  • Overview
  • Method Details
  • Implementation in CAIS
  • Best Practices
  • Common Challenges
• Quasi-Experimental Methods
  • Difference-in-Differences (DiD)
  • Instrumental Variables (IV)
  • Regression Discontinuity Design (RDD)
  • Overview
  • Method Details
  • Implementation in Causal Agent
  • Assumption Validation
  • Best Practices
  • Common Pitfalls
• Observational Methods
  • Propensity Score Matching
  • Overview
  • Method Details
  • Implementation in Causal Agent
  • Assumption Validation
  • Balance Assessment
  • Best Practices
  • Sensitivity Analysis
  • Common Challenges
  • Advanced Topics

Method Selection Guide

Not sure which method to use? Causal Agent can automatically select the most appropriate method based on your data and research design. However, understanding the different approaches will help you make informed decisions and interpret results correctly.

Start Here

• Overview of Causal Inference Methods - Introduction to causal inference and Causal Agent methods

• Method Selection Decision Tree - Interactive guide to method selection

Method Categories

Experimental Methods (Experimental Methods)

Gold standard for causal inference when randomization is possible.

• Randomized Controlled Trials (RCT)

• A/B Testing

• Field Experiments

Quasi-Experimental Methods (Quasi-Experimental Methods)

Leverage natural experiments and policy changes for causal identification.

• Difference-in-Differences (DiD)

• Instrumental Variables (IV)

• Regression Discontinuity (RDD)

Observational Methods (Observational Methods)

Extract causal insights from observational data with careful identification strategies.

• Propensity Score Matching

• Propensity Score Weighting

• Backdoor Adjustment

• Linear Regression (with controls)

Method Comparison

Method Comparison Overview

Method	Data Requirements	Key Assumptions	Strength of Evidence	Common Use Cases
RCT	Randomized treatment	No spillovers	Highest	Medical trials, A/B tests
DiD	Panel data, treatment timing	Parallel trends	High	Policy evaluation
IV	Valid instrument	Exclusion restriction	High	Natural experiments
RDD	Continuous assignment variable	Continuity at cutoff	High	Threshold-based policies
Propensity Score	Rich covariates	Unconfoundedness	Medium	Observational studies

Choosing the Right Method

The choice of causal inference method depends on several factors:

1. Research Design: Experimental vs. observational data

2. Data Structure: Cross-sectional, panel, or time series

3. Treatment Assignment: Random, rule-based, or endogenous

4. Available Variables: Instruments, covariates, time dimensions

5. Assumptions: Which identifying assumptions are plausible

Causal Agent automatically evaluates these factors and recommends the most appropriate method, but understanding the trade-offs helps you make informed decisions about your analysis strategy.