Nou Pa Bèt: Civic Substitution and Expressive Freedoms in Post-State Governance

Replication Instructions: Civic Engagement and Governance Analysis Overview This paper includes fully reproducible quantitative analysis using Python in Google Colab. To replicate our regression models and statistical tests, you will need to download and execute the Jupyter notebook (civic.ipynb) that accompanies this publication. This study utilizes the Varieties of Democracy (V-Dem) Country-Year Dataset v15 (2025) to analyze cross-national patterns of civic engagement. The V-Dem dataset is one of the most comprehensive expert-coded political science databases available, widely adopted for research on democratic governance and institutional quality. For this analysis, we focus on a targeted sample of low-income fragile states and high-income democracies, using the most recent ten-year period (2015–2024) to capture contemporary governance dynamics. The dataset includes six core indicators that serve as proxies for civic engagement and participatory governance: the Participatory Component Index (v2x_partip), Civil Society Participation Index (v2x_cspart), Freedom of Peaceful Assembly and Demonstration (v2cademmob), Freedom of Expression (v2cafres), Anti-System Movements (v2csantimv), and the Direct Popular Vote Index (v2xdd_dd). These variables collectively represent institutionalized opportunities for citizen participation, associational life, expressive freedoms, and mechanisms of direct democracy. Data processing was conducted in Python via Google Colab to ensure reproducibility, systematic filtering, and efficient management of country-year panel structures. Observations with low coder coverage (fewer than three expert raters per variable-year) were flagged and excluded to mitigate the risk of measurement error and ensure validity. This multidimensional and methodologically rigorous dataset enables granular comparisons across countries and over time. By employing empirically validated indicators, the V-Dem dataset provides a robust analytical foundation for exploring the role of civic engagement in shaping governance outcomes—particularly in fragile states like Haiti, where institutional voids hinder formal governance mechanisms and civic empowerment plays a critical role in reform pathways. Download civic.ipynb here to run regressions. What You'll Reproduce By running the notebook, you will generate: Two OLS Regression Models examining the relationship between civic engagement indicators and governance outcomes: Model 1: Control of Corruption (dependent variable) Model 2: Government Effectiveness (dependent variable) ANOVA Analysis comparing civic engagement levels between fragile states and durable democracies Descriptive Statistics for all variables in the analysis Data Visualizations including correlation matrices and coefficient plots Expected Results Regression Results (Table 2) When you run the analysis, you should see: Control of Corruption Model: Participatory Component Index: β = 2.89, p < 0.001 ✓ (Highly significant positive effect) Freedom of Expression: β = 0.17, p = 0.028 ✓ (Significant positive effect) Freedom of Peaceful Assembly: β = -0.26, p < 0.001 ✓ (Significant negative effect) Anti-System Movements: β = -0.42, p < 0.001 ✓ (Significant negative effect) Civil Society Participation: β = 0.64, p = 0.201 (Not significant) Direct Popular Vote Index: β = -0.67, p = 0.146 (Not significant) Government Effectiveness Model: Participatory Component Index: β = 3.04, p < 0.001 ✓ (Highly significant positive effect) Freedom of Peaceful Assembly: β = -0.27, p < 0.001 ✓ (Significant negative effect) Anti-System Movements: β = -0.45, p < 0.001 ✓ (Significant negative effect) Freedom of Expression: β = 0.10, p = 0.233 (Not significant) Civil Society Participation: β = 0.65, p = 0.220 (Not significant) Direct Popular Vote Index: β = -0.64, p = 0.187 (Not significant) ANOVA Results (Table 3) You should see highly significant differences between fragile and durable states across all four indicators: Participatory Component Index: F = 53.9, p < 0.001 Civil Society Participation: F = 39.8, p < 0.001 Freedom of Expression: F = 36.3, p < 0.001 Freedom of Peaceful Assembly: F = 34.7, p < 0.001 Step-by-Step Replication Guide Prerequisites Google account (for Google Colab access) Internet connection Web browser (Chrome, Firefox, Safari, or Edge) No software installation required — everything runs in the cloud via Google Colab. Step 1: Access the Dataset and Notebook Option A: Download from Zenodo (Recommended) Visit the dataset repository: https://doi.org/10.5281/zenodo.15058161 Download three files: civic.ipynb (Jupyter notebook with analysis code) CivicEngagement_SelectedCountries_Last10Years.xlsx (V-Dem data) wgidataset.xlsx (World Bank WGI data) Option B: Access via Paper Supplementary Materials If provided by the journal, download the replication package containing all three files. Step 2: Upload to Google Colab Go to https://colab.research.google.com/ Sign in with your Google account Click File → Upload notebook Select civic.ipynb from your downloads folder The notebook will open in a new browser tab Step 3: Upload Data Files to Colab Once the notebook is open: Look for the Files panel on the left sidebar (folder icon) Click the Upload button (page with up arrow icon) Upload both Excel files: CivicEngagement_SelectedCountries_Last10Years.xlsx wgidataset.xlsx Note: Files uploaded to Colab are temporary and will be deleted when your session ends. You'll need to re-upload them each time you open the notebook. Alternative (Persistent Storage): Mount your Google Drive: Run the cell with from google.colab import drive and drive.mount('/content/drive') Upload data files to a folder in your Google Drive Update file paths in the notebook to point to your Drive folder Step 4: Install Required Python Packages The notebook includes cells to install necessary packages. Run these cells in order: !pip install pandas openpyxl statsmodels matplotlib seaborn scipy You should see: Installation progress bars "Successfully installed..." messages No error messages If you encounter errors: Try running the cell again Check your internet connection Verify you're using a standard Colab runtime (not a restricted environment) Step 5: Execute the Analysis Cells Run each code cell in sequence by: Clicking the Play button (▶) to the left of each cell, OR Using keyboard shortcut: Shift + Enter Important: Run cells in order from top to bottom. Later cells depend on variables and functions defined in earlier cells. You should see: Data Loading Section: Confirmation messages: "Data loaded successfully" Dataframe previews showing first few rows Shape information (number of rows and columns) Data Cleaning Section: Summary of missing values Confirmation of listwise deletion Final sample size (should be ~270 country-year observations) Descriptive Statistics Section: Tables showing mean, standard deviation, min, max for all variables Correlation matrix heatmap Regression Analysis Section: Full regression output tables matching Table 2 in the paper Coefficient estimates, standard errors, t-statistics, p-values R-squared values and model diagnostics ANOVA Section: F-statistics and p-values matching Table 3 Group means for fragile vs. durable states Visualizations Section: Coefficient plots showing effect sizes with confidence intervals Scatter plots of key relationships Distribution plots comparing fragile and durable states Step 6: Verify Your Results Match the Paper Check these key values: ✓ Participatory Component Index (Control of Corruption): β ≈ 2.89, p < 0.001✓ Freedom of Expression (Control of Corruption): β ≈ 0.17, p ≈ 0.028✓ R-squared for Control of Corruption model: R² ≈ 0.79✓ ANOVA F-statistic for Participatory Index: F ≈ 53.9, p < 0.001 If your results differ slightly: Minor numerical differences (±0.01) are normal due to floating-point precision Ensure you're using the correct data files Verify you ran all cells in order Check that no cells produced error messages Step 7: Modify and Extend (Optional) The notebook is designed for transparency and extensibility. You can: Explore Alternative Specifications: Add or remove control variables Test interaction effects Run robustness checks with different subsamples Example modifications included in notebook: Robust standard errors (default) Bootstrap confidence intervals Sensitivity analysis excluding outliers Generate Additional Visualizations: Change plot styles and colors Create custom graphics for presentations Export high-resolution figures Export Results: Download regression tables as CSV Save figures as PNG or PDF Generate formatted LaTeX tables Troubleshooting Common Issues "File not found" error Solution: Verify data files are uploaded to Colab. Check the Files panel (left sidebar) to confirm both Excel files are present. "ModuleNotFoundError" or "ImportError" Solution: Run the package installation cell: !pip install pandas openpyxl statsmodels ... Results differ from paper Solution: Confirm you're using the exact data files from Zenodo Verify all cells were executed in order Check for error messages in earlier cells "Memory error" or session crash Solution: Restart the runtime: Runtime → Restart runtime Re-upload data files Execute cells again from the beginning Colab disconnects or times out Solution: Colab sessions timeout after ~90 minutes of inactivity Save any modified code to your Google Drive Reconnect and re-run cells Technical Specifications Software Environment: Python 3.10+ pandas 2.0+ statsmodels 0.14+ matplotlib 3.7+ seaborn 0.12+ scipy 1.11+ Computational Requirements: Runtime: ~2-5 minutes on standard Colab instance Memory: ~2GB RAM No GPU required Operating System: Platform-independent (runs in browser) Tested on: Windows, macOS, Linux, ChromeOS Data Sources and Licensing V-Dem Dataset (v15, 2025): Source: Varieties of Democracy Institute License: CC BY-SA 4.0 Citation: Coppedge et al. (2025) World Governance Indicators (2024): Source: World Bank License: CC BY 4.0 Citation: Kaufmann, Kraay & Mastruzzi (2024) Analysis Code: License: CC BY 4.0 Authors: Brown, Fils-Aime, LaTortue Citation If you use this replication package in your research, please cite: Brown, S.M., Fils-Aime, J., & LaTortue, P. (2026). Nou Pa Bèt: Civic Substitution and Expressive Freedoms in Post-State Governance. Third World Quarterly. DOI: [to be assigned] And cite the data repository: Brown, S.M., Fils-Aime, J., & LaTortue, P. (2025). Corporate Governance under Institutional Collapse: Civic Engagement, Ethical Embeddedness, and Reform in Haiti [Data set]. Zenodo. https://doi.org/10.5281/zenodo.15058161 Support and Contact For technical issues with replication: Check the Zenodo repository for updates: https://doi.org/10.5281/zenodo.15058161 Review the notebook's built-in documentation and comments For questions about methodology or interpretation: Contact corresponding author: [e-mail] For dataset questions: V-Dem data: [e-mail] WGI data: World Bank Open Data Help Desk FAIR Principles Compliance This replication package adheres to FAIR data principles: Findable: Persistent DOI assigned via ZenodoAccessible: Open access, no authentication requiredInteroperable: Standard formats (XLSX, IPYNB), documented codeReusable: Clear licensing (CC BY 4.0), comprehensive documentation Acknowledgments We thank the V-Dem Institute and World Bank for making high-quality governance data publicly available. This analysis was conducted using open-source software to ensure maximum accessibility and reproducibility. Ready to Begin? Download civic.ipynb from Zenodo Upload to Google Colab Upload the two Excel data files Run all cells in order Verify results match Tables 2 and 3 Explore, modify, and extend as desired Happy replicating!