(https://rpubs.com/Riley_Ruble/Sabrina_Chen)

Created by Riley Ruble and Sabrina Chen. Updated 12/8/24

Increasing property taxes leads to higher 11th-grade school proficiency rates, and long-term economic strength for West Virginia.

Investing in education through higher property taxes significantly enhances academic performance for 11th-grade students, as demonstrated by improved math and reading proficiency rates. This analysis reveals that counties with higher property tax revenues correlate with better educational outcomes and higher future prospects for students, including better job opportunities and economic mobility. By prioritizing educational funding through increased property taxes, West Virginia can strengthen its workforce and build a more prosperous future.

Data Description

This analysis integrates datasets from multiple sources to examine the relationship between property taxes and 11th-grade proficiency rates in math and reading. The primary datasets include educational assessment results, financial records of school expenditures, and county-level demographic data. Math and reading proficiency rates were chosen because 11th grade marks the typical timing for SAT exams, making it a strong indicator of peak educational performance.

Key Variables:
  • average_proficiency_rate: Mean of math and reading proficiency rates (percentage).
  • local_property_tax_revenue: Numeric field of local property taxes collected by county.
  • edu_percent: Percentage of residents with bachelor’s degrees.
  • employed: Percentage of employed individuals per county.

Methods

Linear Model

Model Formula:
  • lm(formula = average_proficiency_rate ~ local_property_tax + edu_percent + employed, data = master_data)
Residuals:
  • Minimum: -11.87
  • Median: -0.85
  • Maximum: 15.79
Coefficients:
  • Local Property Tax: 0.00007206 (Marginally significant; p = 0.0728 )
  • Percentage with a Degree: 0.2996 (Significant; p = 0.0339 )
  • Employment Rate: 0.6272 (Significant; p = 0.0323 )
  • Adjusted R-squared: 0.418
Key Takeaways:
  • Local property taxes show a small but positive influence on proficiency rates.
  • Counties with higher education levels and employment rates see significantly better academic outcomes.
  • The model explains approximately 45% of the variation in proficiency rates.

West Virginia Proficiency by County

K-Means(unsupervised)

Scatter Plot

The clustering analysis groups West Virginia counties by proficiency rates, property tax revenue, and education levels. Cluster 2 (green) includes counties with high proficiency, high property taxes, and more residents with bachelor’s degrees. Cluster 3 (blue) represents low proficiency, low property taxes, and fewer college-educated residents. Cluster 1 (red) falls in between, with moderate proficiency, taxes, and education levels. This highlights the link between local investment, education levels, and academic success.

Clusters by County on West Virginia Map

Areas with higher proficiency rates like Jefferson, Monongalia, Putnam, and Tyler have better economies and are closer to metro areas like Pittsburgh and DC. Some people tend to commute from West Virginia to jobs in cities that have a higher cost of living which explains why income/property tax in those areas tend to be higher. The city of Kanawha is an exception because it is the Capital of West Virginia, which might explain why Putnam have a higher rate than surrounding counties.

Decision Tree (Supervised)

Decision Tree

This decision tree illustrates the relationship between local property taxes, educational attainment, and employment rates in predicting 11th-grade proficiency rates. The root decision splits based on whether local property taxes are below 21,000. Counties with lower property taxes are further divided by the percentage of residents with bachelor’s degrees. If this percentage is below 12%, proficiency rates are lower, particularly when employment rates are also low. On the other hand, counties with higher property taxes and a greater percentage of college-educated residents exhibit significantly higher proficiency rates, with the highest proficiency outcomes occurring in areas where property taxes exceed 21,000. This highlights the strong influence of local investments and education levels on academic success.

RMSE Across 40 Iterations

Each point represents the RMSE for a single iteration, indicating the model’s error in predicting outcomes during testing. The red dashed line represents the average RMSE across all iterations, providing a benchmark for the model’s overall performance. The RMSE fluctuates between approximately 5 and 9, showing some variability in prediction accuracy, but it remains relatively stable around the average of 6. This suggests the model’s performance is consistent, with minor variations due to the randomness of training and testing splits across iterations.

Limitations

Scope of Analysis:
  • This analysis is specific to West Virginia and focuses solely on 11th-grade proficiency rates. The results may not be applicable to other states with different economic, geographic, or educational contexts.
Data Constraints:
  • Some counties may have incomplete or missing data for key variables such as property taxes, employment rates, or proficiency scores. This could skew the results or limit the generalizability of the findings.
Suggestions for Improvement:
  • Include more nuanced variables, such as household income, school funding allocation, or community involvement metrics.
  • PCA data could have been used instead of kmean to produce a better/different outcome.
  • Explore the impact of specific educational policies or initiatives to better understand how increased funding is utilized.

References

Source included:

Thank you!