new pdf

analysis/analysis4.ipynb

@@ -355,6 +355,58 @@
     "print(recent_model.summary())"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## **Notes on Regression Results: Historical vs. Recent Spills**\n",
+    "\n",
+    "### **Historical Spills Regression Results**\n",
+    "- **R² = 0.021** → The model explains **2.1%** of the variance in reporting delay. \n",
+    "- **Key Findings:**\n",
+    "  - **Income is negatively correlated with reporting delay**:  \n",
+    "    - **(-0.0009, p < 0.001)** → Lower **median household income** is associated with **longer reporting delays**.\n",
+    "  - **Percent White has a strong negative effect**:  \n",
+    "    - **(-5.55, p < 0.001)** → Spills in **whiter areas** are reported **faster**.\n",
+    "  - **Percent Hispanic also has a strong negative effect**:  \n",
+    "    - **(-3.16, p < 0.001)** → Spills in **Hispanic areas** have longer delays.\n",
+    "  - **Percent Poverty and Unemployment Rate are not significant**:  \n",
+    "    - Poverty: **(-0.83, p = 0.21)**  \n",
+    "    - Unemployment: **(3.31, p = 0.12)**\n",
+    "- **Implications:**  \n",
+    "  - Spills discovered late (historical) tend to be in **lower-income and more Hispanic areas**.\n",
+    "  - There may be **under-monitoring or regulatory neglect** in these communities.\n",
+    "\n",
+    "---\n",
+    "\n",
+    "### **Recent Spills Regression Results**\n",
+    "- **R² = 0.001** → The model explains almost **none** of the variance in reporting delay.\n",
+    "- **Key Findings:**\n",
+    "  - **Most variables are NOT statistically significant**:  \n",
+    "    - Income: **(6.59e-06, p = 0.79)**  \n",
+    "    - Poverty: **(-0.095, p = 0.29)**  \n",
+    "    - Unemployment: **(0.082, p = 0.82)**\n",
+    "  - **Percent White has a weak positive effect**:  \n",
+    "    - **(0.134, p = 0.046)** → In recent spills, whiter areas have **slightly longer delays**.\n",
+    "  - **Percent Hispanic is borderline significant**:  \n",
+    "    - **(0.086, p = 0.072)** → Suggests **slightly longer delays** in Hispanic areas, but not as strong as in historical spills.\n",
+    "- **Implications:**  \n",
+    "  - **Socioeconomic factors do not strongly influence recent spill reporting.**\n",
+    "  - This suggests that **recent spills are reported more uniformly** across different demographic areas.\n",
+    "  - **Improvements in real-time monitoring may have reduced disparities** in reporting speed.\n",
+    "\n",
+    "---\n",
+    "\n",
+    "### **Comparative Takeaways**\n",
+    "| Variable                | Historical Spills (p-value) | Recent Spills (p-value) | Interpretation |\n",
+    "|-------------------------|---------------------------|-------------------------|---------------|\n",
+    "| **Median Income**       | -0.0009 (**p < 0.001**)   | 6.59e-06 (p = 0.79)    | Income disparities matter **only for historical spills**. |\n",
+    "| **Percent Poverty**     | -0.835 (p = 0.21)         | -0.095 (p = 0.29)      | Poverty does not significantly impact reporting delay. |\n",
+    "| **Unemployment Rate**   | 3.307 (p = 0.12)         | 0.082 (p = 0.82)      | No meaningful relationship with reporting delay. |\n",
+    "| **Percent White**       | -5.548 (**p < 0.001**)    | 0.134 (*p = 0.046*)    | Whiter areas had **faster historical reporting**, but slightly **slower** recent reporting. |\n",
+    "| **Percent Hispanic**    | -3.158 (**p < 0.001**)    | 0.086 (p = 0.072)      | Hispanic areas experienced **longer delays historically**, but recent disparities are weaker. |\n"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": 18,