This repository contains my final replication and extension project for 40800. The project replicates and extends Moyer et al. (2023), "Identifying Patterns in the Structural Drivers of Intrastate Conflict", using the authors' replication package and a cleaned Quarto workflow.
The report replicates the authors' base PITF clustering specification and several supplied sensitivity specifications. The base replication reconstructs the following workflow: complete-case filtering of PITF onset cases, min-max scaling of the six structural variables, Euclidean dissimilarity construction, AGNES hierarchical clustering with Ward linkage, and a three-cluster solution.
The extension has two parts. First, it adds logged population and a transformed peace-years measure to test whether the three-profile typology remains stable after adding conflict-history and population-scale information. Second, it evaluates methodological robustness using PAM, k-means, Gower dissimilarities, silhouette diagnostics, unstable-case checks, PCA visualization, and nearby cluster-number diagnostics.
| Path | Description |
|---|---|
40800-replication-final.pdf |
Final combined submission PDF. |
report.qmd |
Main academic report. |
appendix.qmd |
Cleaned reproducible Quarto appendix with code and diagnostics. |
references.bib |
Bibliography used by the Quarto report. |
tables/ |
Saved replication, membership, sensitivity, and diagnostic output tables. |
data/raw/ |
Original replication-package materials used for the analysis, including cleaned input files, supplied scripts, and saved Excel outputs. |
The project was rendered with Quarto. To reproduce the main report, appendix, and final combined PDF, run:
quarto render report.qmd --to pdf
quarto render appendix.qmd --to pdf
qpdf --empty --pages report.pdf appendix.pdf -- 40800-replication-final.pdf
The appendix includes the main reproducibility diagnostics and session information.
The main report is designed to read as a standalone academic article. The appendix contains the cleaned reproducible workflow, including data loading, variable selection, min-max scaling, Euclidean and Gower dissimilarity construction, AGNES/Ward clustering, reimplemented additional-IV, V-Dem, and PRIO sensitivity specifications, PAM and k-means extensions, membership-level diagnostics, silhouette diagnostics, unstable-case diagnostics, PCA diagnostics, saved output tables, and session information.
Rendered intermediate files such as report.pdf, appendix.pdf, report.html, and appendix.html are ignored by default. The final combined PDF is included in the repository for submission convenience.