The Reindeer Rebellion: Institutional Forced Migration Cloaked in the Language of Reconciliation

Dataset and Analysis Code Authors: [Anon until Published] DOI: https://doi.org/10.5281/zenodo.17619998 Paper Citation: [Full citation to be added upon publication] OVERVIEW This dataset contains forestry permit data from Sweden (1995-2024) and analysis code used to evaluate the impact of the 2020 Girjas Supreme Court decision on clearcutting patterns in Sámi reindeer herding territories. The analysis employs Chow structural break tests to assess whether legal recognition of Indigenous land rights altered forestry governance behavior. Key Finding: The Girjas decision (January 23, 2020) produced statistically significant reductions in clearcutting permits in Sámi-dominant regions, but did not transfer jurisdictional authority to Sámi communities—demonstrating what we term "symbolic governance." FILES INCLUDED Data Files Girjas.csv - Clearcutting Pressure Index (CPI) data by region and year (1995-2024) Source: Swedish Forest Agency (Skogsstyrelsen) statistical database Variables: Region, Years (1995-2024), CPI per 1,000 hectares productive forest land 12 regions × 30 years = 360 observations Analysis Files sami.ipynb - Main analysis notebook (Chow structural break test for 2020) robustness_check.ipynb - Robustness check notebook (tests alternative break years 2018-2022) Documentation README.md - This file METHODOLOGY.md - Detailed explanation of Chow test methodology CODEBOOK.md - Variable definitions and data dictionary HOW TO USE THIS DATASET Option 1: Run Analysis in Google Colab (Recommended - No Installation Required) Google Colab is a free cloud-based Jupyter notebook environment that requires no software installation. All code runs in your browser. Step 1: Access Google Colab Go to https://colab.research.google.com/ Sign in with your Google account (free accounts work fine) Step 2: Upload the Notebook For Main Analysis (sami.ipynb): In Google Colab, click File → Upload notebook Click Choose File and select sami.ipynb from this dataset The notebook will open in a new tab For Robustness Check (robustness_check.ipynb): Repeat the same process with robustness_check.ipynb Step 3: Upload the Data File In the left sidebar, click the folder icon (📁) to open the file browser Click the upload icon (📄 with up arrow) Select Girjas.csv from this dataset Wait for upload to complete (file will appear in the file browser) IMPORTANT: You must upload Girjas.csv for BOTH notebooks separately. Google Colab does not persist files between sessions. Step 4: Run the Analysis For sami.ipynb (Main Analysis): Click Runtime → Run all to execute all cells The analysis will: Load and clean the data Perform Chow structural break test for year 2020 Generate results table with F-statistics and p-values Create visualization plots Export results to chow_test_results_complete.csv Expected runtime: 10-30 seconds Outputs: Console output with formatted results tables Two PNG files: Figure1_CPI_AllRegions.png and Figure1_KeyRegions.png CSV file: chow_test_results_complete.csv For robustness_check.ipynb (Robustness Analysis): Click Runtime → Run all The analysis will: Load the same data Test structural breaks for years 2018, 2019, 2020, 2021, 2022 Compare results across years to confirm 2020 temporal specificity Export results to robustness_alternative_years.csv Expected runtime: 15-45 seconds Outputs: Console output with year-by-year comparison Interpretation guidance for publication CSV file: robustness_alternative_years.csv Step 5: Download Results After running, generated files appear in the left sidebar file browser Right-click any file → Download to save to your computer CSV files can be opened in Excel, Google Sheets, or statistical software Option 2: Run Locally (Advanced Users) Requirements: Python 3.7+ Jupyter Notebook or JupyterLab Required packages: pandas, numpy, scipy, statsmodels, matplotlib Installation: pip install pandas numpy scipy statsmodels matplotlib jupyter Running: # Navigate to dataset directory cd path/to/dataset # Launch Jupyter jupyter notebook # Open sami.ipynb or robustness_check.ipynb in browser # Click "Run All" to execute INTERPRETING THE RESULTS Main Analysis (sami.ipynb) Output Table Structure: Region Chow F p-value Sig. Pre-2020 Post-2020 Change ------------------------------------------------------------------------------------------ 00 Hela landet Samtliga 10.98 <0.001 Yes 81.5 29.2 -52.3 25 Norrbottens Samtliga 3.72 0.038 Yes 73.2 28.1 -45.1 Norra Norrland Samtliga 8.60 0.001 Yes 76.6 21.8 -54.8 How to Read: Chow F: F-statistic for structural break test (higher = stronger break) p-value: Statistical significance (p < 0.05 = significant) Sig.: "Yes" if p < 0.05, "No" otherwise Pre-2020: Average CPI before Girjas decision Post-2020: Average CPI after Girjas decision Change: Absolute change (negative = decline in clearcutting) Key Findings: National level: Significant break (F=10.98, p<0.001), 64% decline in CPI Norrbotten (Girjas region): Significant break (F=3.72, p=0.038), 62% decline Norra Norrland (broader Sámi territory): Significant break (F=8.60, p=0.001), 72% decline State forestry lands: Significant breaks Private forestry lands: No significant breaks (suggests behavioral mechanism operates through litigation risk for state agencies) Robustness Check (robustness_check.ipynb) Purpose: Confirms that 2020 is the relevant break year rather than adjacent years (2018, 2019, 2021, 2022). Output Structure: Year F-Stat p-value Significant ------------------------------------------ 2018 6.85 0.0041 Yes 2019 6.36 0.0057 Yes 2020 4.83 0.0165 Yes ← GIRJAS 2021 3.72 0.0381 Yes 2022 3.41 0.0483 Yes Important Note on Robustness Results: The robustness check reveals that multiple years show significant breaks, suggesting a continuous declining trend rather than a discrete 2020 shock. This pattern is consistent with: Anticipatory governance effects (forestry agencies became cautious before the ruling) Girjas as catalytic event (formalized and legitimized existing caution) Peak effect in 2020 (highest F-statistics in Norra Norrland: F=29.93) This complexity strengthens rather than weakens the "symbolic governance" argument: legal recognition performs constraint without redistributing substantive authority. METHODOLOGY SUMMARY Chow Structural Break Test What it tests: Whether the statistical relationship between time and clearcutting intensity changed significantly at a specific year (the "break point"). How it works: Fits three linear regression models: Pooled model: Assumes no change across entire time period (1995-2024) Pre-break model: Fits data before break year Post-break model: Fits data after break year Compares fit quality using Residual Sum of Squares (RSS) Calculates F-statistic: measures whether separate models fit significantly better than pooled model Derives p-value from F-distribution Formula: F = [(RSS_pooled - (RSS_pre + RSS_post)) / k] / [(RSS_pre + RSS_post) / (n1 + n2 - 2k)] Where: k = number of parameters (2: intercept + slope) n1 = observations before break n2 = observations after break Interpretation: p < 0.05: Statistically significant structural break (trend changed at that year) p ≥ 0.05: No significant break (trend remained consistent) Significance threshold: α = 0.05 (95% confidence level) DATA SOURCE AND PROVENANCE Original Data Source: Swedish Forest Agency (Skogsstyrelsen) public statistical database: Table: "Ansökan om tillstånd till föryngringsavverkning inom fjällnära skog efter region och ägarkategori. År 1995–" Translation: "Applications for clearcut felling permits in mountain-near forests by region and owner category. Years 1995–" URL: https://www.skogsstyrelsen.se/statistik/ Data Collection: Accessed: January 2025 Coverage: 1995-2024 (30 years) Regions: 12 Swedish forestry regions Standardization: CPI calculated as permits per 1,000 hectares productive forest land Data Processing: Downloaded raw permit counts by region and year Obtained productive forest land area by region from Swedish Forest Agency Calculated Clearcutting Pressure Index (CPI) = (permits / forest area) × 1000 Cleaned and validated data for missing values and outliers Data Quality: Government-reported statistics (high reliability) Complete coverage for all regions and years No missing values for key variables REPRODUCIBILITY All analyses are fully reproducible: Data: Publicly available from Swedish Forest Agency (source URLs provided) Code: Complete analysis code provided in notebooks Environment: Google Colab ensures consistent computing environment Documentation: Detailed methodology and interpretation guidance included To reproduce: Download this dataset from Zenodo Follow "How to Use" instructions above Run notebooks in Google Colab Results should match those reported in the paper Version Control: Python: 3.10+ (Google Colab default) pandas: 1.5+ statsmodels: 0.13+ scipy: 1.9+ CITATION If you use this dataset or code, please cite: Dataset: [Full dataset citation to be added upon publication in Organization Studies] Paper: [Full paper citation to be added upon publication in Organization Studies] LICENSE Data: CC BY 4.0 (Creative Commons Attribution 4.0 International) You are free to: share, adapt, and build upon this data You must: give appropriate credit and indicate if changes were made Code: MIT License Free to use, modify, and distribute Attribution required SUPPORT AND CONTACT Questions about the data or analysis: Open an issue on [GitHub repository link if applicable] Contact: [corresponding author email] Questions about the paper: See published paper for detailed methodology and interpretation Contact corresponding author for clarifications Technical issues with Google Colab: See Google Colab documentation: https://colab.research.google.com/notebooks/intro.ipynb Most issues resolve by: clearing runtime (Runtime → Restart runtime) and re-uploading data TROUBLESHOOTING Problem: "NameError: name 'df_long' is not defined" Solution: You forgot to upload Girjas.csv. Upload it using the file browser (📁 icon in left sidebar). Problem: "FileNotFoundError: [Errno 2] No such file or directory: 'Girjas.csv'" Solution: Make sure Girjas.csv is uploaded to the Colab session. Files don't persist between sessions. Problem: Notebook runs but produces no output Solution: Click inside a cell and press Shift+Enter to run individual cells, or use Runtime → Run all Problem: Plots don't display Solution: Ensure matplotlib is imported. This is handled automatically in provided notebooks. Problem: Results differ slightly from paper Solution: Minor numerical differences (<0.01) may occur due to floating-point precision. Substantive findings should match. Problem: Can't download generated files Solution: Right-click the file in the left sidebar file browser → Download. If files don't appear, check that the cell finished running (no spinning icon). ACKNOWLEDGMENTS This research was supported by [funding sources]. Data provided by the Swedish Forest Agency (Skogsstyrelsen). We thank [acknowledgments] for feedback and support. VERSION HISTORY v1.0 (January 2025): Initial release with main analysis and robustness check Future versions will be documented here if dataset is updated Last Updated: January 2025 Maintained by: Scott Brown (corresponding author)