🔍 The Problem Identified

Mixed-methods designs that nest small-N analysis (SNA) within large-N analysis (LNA) are increasingly popular. However, the LNA typically assumes independently distributed units and therefore cannot account for spatial dependence. When spatial dependence is present, it becomes a threat to inference rather than a subject of empirical or theoretical investigation—an important shortcoming given recent political science attention to diffusion and broader interconnectedness.

🧭 A Practical Framework: Geo-Nested Analysis

A framework labeled "geo-nested analysis" is developed to integrate spatial dependence into mixed-methods research. Key features include:

• Treating spatial dependence as an object of study rather than an inferential nuisance.
• Letting insights from each research step set the agenda for the next phase of analysis.
• Basing case selection for SNA on diagnostics from spatial-econometric analysis performed in the LNA.

📌 How the Framework Operates

• Conduct a large-N spatial-econometric analysis that explicitly tests for and models spatial dependence.
• Use diagnostic results from that analysis to guide purposeful selection of cases for small-N, qualitative, or process-tracing work.
• Iterate between quantitative diagnostics and qualitative investigation so each phase refines questions and case choices for the next phase.

🧪 Illustration Using Homicide Data

The framework is illustrated using data from a seminal study of homicides in the United States, demonstrating how spatial diagnostics can meaningfully shape case selection and interpretation.

❗ Why It Matters

Geo-nested analysis preserves the strengths of nested mixed-methods designs while addressing the inferential risks posed by spatial dependence. This approach helps align methodological practice with substantive interests in diffusion and interdependence across political units.