Hurricanes and SAT Scores
Finding the Link between Weather and Future Economic Well-Being
The senior thesis for the Quantitative Economics B.A. at UC Irvine was on the effects of hurricanes on SAT scores. The project was inspired by then-recent Hurricane Katrina, where the long-term environmental, social, and economic effects were not yet known. The hypothesis was that public high school students living in extreme weather regions produced significantly lower SAT than those who lived in mild weather regions.
Thinking Process
My reasoning was, if you as a high school student cannot attend school due to inclement weather, or due to home, family loss, and/or displacement cannot function normally as a student and perform well on the SAT, the extreme weather may then affect what university or college that student may be accepted into, if any.
Since numerous studies have shown that income level is significantly higher for someone who has graduated from college compared to someone with only a high school diploma, then whether or not someone gets into college, and the clout of the college that person gets into, would have a longer lasting effect on lifetime total income. Therefore, the underlying question was, "are people who live in intense weather regions (affected by hurricanes, tornadoes, flooding, etc.) at a long-term economic disadvantage compared to those who live in mild weather areas?"
Quantitative Analysis
Statistical Test Design
Due to data availability and biases, the public high school systems of Florida (test group, known for high instances of extreme weather) and Arizona (control group, known for few instances of extreme weather) were compared. The hypothesis and data available determined that the Ordinary Least Squares (OLS) regression could be used.
Data Collection and Conversion
SAT test score, income, location, land use type, population data, and environmental damage (cost) was easily available online. Since the amount of damage done by a hurricane is related to the type (h1, h2, h3, h4, h5), spin, hurricane path, and duration, estimates were drawn from historical data provided by a storm map on the NOAA website. Information was converted into binary, categorical, and numerical data, and a 5-year time-series format was chosen.
Econometric Analysis
The Ordinary Least Squares (OLS) regression was run using EViews software. It was found that at the 0.05 significance level, Type 1 Hurricanes are statistically significant. Some possible explanations could be the fact that hurricanes in Florida are a common occurrence and therefore students may be used to subsequent interruptions in their studies, or perhaps students may simply opt to take the test another year in light of an especially bad hurricane.