Network analysis has become increasingly important in political science over recent years. This paper reviews three key approaches—Quadratic Assignment Procedure, Exponential Random Graph Models (ERGM), and Latent Space Network Models—for statistically valid network inference.

New Tools for Complex Questions:

These techniques go beyond basic descriptive measures of network structure to help researchers understand the intricate relationships within political networks.

Methodological Comparison:

• QAP: Focuses on testing relations between node attributes and whole-network statistics, treating ties as dependent.
• ERGM: Models network structures directly using probability distributions while accounting for dependencies.
• Latent Space Model: Represents actors with latent positions in an underlying space to explain observed connections.

Superior Performance:

Analysis of climate change policy networks demonstrates that all three inferential techniques significantly outperform standard logit models across multiple criteria.

Practical Guidance: The paper helps political scientists navigate these complex methods and choose the most appropriate technique for their research questions.