Structural Equation Modeling Overview

Introduction

• Structural Equation Modeling (SEM) is an analytical approach for testing theories.
• Combines confirmatory factor analyses and multiple regressions.
• Aimed at providing a high-level conceptual understanding of SEM.

Presenter's Background

• Ashley, a second-year PhD student, creates videos about goals and academics.
• Learning by teaching approach.

Structure of the Video

• Focus on Anderson and Gerbing's 1988 article as a foundation for the two-step SEM approach.
• Incorporation of contemporary best practices.

Two-Step SEM Approach

Step 1: Analyze the Measurement Model

• Measurement Model: Tests if we measure what we intend.
• Involves using Confirmatory Factor Analyses (CFA).
• Types of Measurement Invariance Tests:
  • Configural Invariance: Tests the factor structure (e.g., two-factor vs. three-factor).
  • Metric Invariance: Evaluates the expected factor loadings.
    • Aim for factor loadings above 0.5 for strength.
• Analyze measurement models independently for exogenous (X) and endogenous (Y) variables.

Theoretical Framework

• Utilize theory to re-specify the measurement model (e.g., allowing residuals to co-vary).
• Importance of guiding analyses with strong theoretical backing.
• Avoid "harking" (hypothesizing after results) to maintain scientific integrity.

Reporting and Replication

• Report degrees of freedom for all models tested to facilitate replication.
• Replication is essential for durable knowledge creation.

Step 2: Analyze the Path Model

• Path Model: Represents relationships between variables.
• Hypothesize the direction of relationships (e.g., X leads to Y).
  • Analyze the statistical significance and coefficient direction.
• Ensure all connections in the model are tested and theoretically justified.

Conclusion

• Importance of measuring and justifying all aspects of both models.
• Encourages a learning mindset; failures are opportunities to build knowledge.
• Invitation to explore further resources, such as Karma for deeper understanding of methods.

Call to Action

• Encourage feedback and engagement from the community.
• Promote further discussions on structural equation modeling.