ECONOMICS UNDERGRADUATE THESIS
This research project addresses whether or not firearm ownership rates affect violent crime rates. The study uses violent crime, gun ownership, household income, Gini coefficient, and unemployment rate data from all 50 states in the year 2013. The findings of this research indicate that firearm ownership rates are positively linked with violent crime rates at the state level, which confirms the sentiment of most existing literature on the subject. The graph above plots all 50 states according to their violent crime rate, as shown on the horizontal axis, and their gun ownership rate, as shown on the vertical axis. Some states of interest are Vermont, which has the lowest violent crime rate, Delaware, which has the lowest gun ownership rate, and Alaska, which has both the highest gun ownership rate and violent crime rate.
Click the images below to enlarge
 Correlation Coefficient MatrixThe correlation coefficient matrix shows that there are relatively low correlations between independent variables. Violent crime is more correlated to unemployment rate than any of the other independent variables, which is reflected in both regressions. As seen above, household income and poverty rate are very highly correlated, so they will not be used in the same regression. |  Linear RegressionFor the primary regression, all tests are one-tailed at a 95% confidence interval, yielding a critical t of 1.684. As seen above, both gun ownership and unemployment rate are statistically significant, as their t scores exceed 1.684. Household income has the opposite sign from what was expected, but was statistically insignificant. Gini coefficient produced the expected sign, but was not statistically significant. |  Summary StatsAs seen above, the skewness of every variable tested was between -1 and 1, which is very mild. The kurtosis of every variable was fairly close to 3, indicating that the tailedness of the data collected for each variable is close to that of that of a normal distribution. The Jarque-Bera test indicated that the data for all variables was normally distributed since the p-values were over 0.05 for every variable, meaning that we fail to reject the null hypothesis of a normal distribution. |
|---|