Dependent Variable

The outcome variable is Gun death in the United States in 2020. The sample size is 51. The rate is measure per 100,000 people. Many numbers of states are having median rates. The range is between 3.3 and 22.9. It seems more than half of states having higher than average death rate. The mean was 13.5, which is higher than the median. The standard deviation is approximately 5, and it is relatively low. Most data are laid on between 8.3 and 18.3. Which means, the data gathered mostly around the average. There are few outliers, and the states have death rate mostly between 10 and 20. My dependent variable is interval variable.

Independent Variable

My main independent variable is gun ownership rate in the United States. The sample size is 51. The data is showing %of population on gun ownership rate. The range is between 5.2% and 61.7%. Most states have gun ownership rate between 20% and 40%. There are not many outliers and the standard deviation is approximately 14. It is relatively high. My independent variable is interval variable.

Hypothesis

• My hypothesis is if there is increase in gun ownership rate, the gun related deaths will also increase.
• From figure3, there are strong relationship between gun ownership rate and gun related death rate. (cor = 0.69)
• I can expect based on my plot that as more people own guns in the United States will increase the number of people die related to guns.
• The t-statistic is t = 6.5838

Linear Regression Model

• Every one unit increase in firearm death rate, the gun ownership rate will increase by 0.136%.” It is for every one additional gun owner per 100,000 people in a state, we expect the firearm death rate to increase by 0.136 percentage points. It is statistically significant with p = 0.001.
• Every one unit increase in firearm death rate, the gun registration permit rate will decrease by 0.30%.” It is for every one additional gun registration permit in a state, we expect the firearm death rate to decrease by 0.3 percentage points. It is not statistically significant with p = 0.67.
• Every one unit increase in firearm death rate, the gun carry permit rate will decrease by 0.02%.” It is for every one additional gun carry permit in a state, we expect the firearm death rate to decrease by 0.02 percentage points. It is not statistically significant with p = 0.84.
• Every one unit increase in firearm death rate, the gun purchase permit rate will decrease by 4.36%.” It is for every one additional gun purchase permit in a state, we expect the firearm death rate to decrease by 4.36 percentage points. It is statistically significant with p = 0.0009
• Every one unit increase in firearm death rate, the gun open carry legalize rate will decrease by 1.30%.” It is for every one additional gun legalize in a state, we expect the firearm death rate to decrease by 1.3 percentage points. It is not statistically significant with p = 0.33.
• Every one unit increase in firearm death rate, the gun law score rate will decrease by 1.12%.” It is for every one additional gun law score in a state, we expect the firearm death rate to decrease by 1.12 percentage points. It is statistically significant with p = 0.047.

Model Prediction

Three significant variables for firearm related deaths

From figure 5, I can easily tell without numbers that between gun ownership and firearm related deaths has strong positive relationship. The number of gun ownership is statistically significant.

From figure 6, I can see that between firearm related deaths and gun law score has negative relationship, which means if there are more strict gun laws, less people will die related to firearms.

From figure 7, There are more people die without permit required for firearm purchase. As more permit required for gun purchasing will most likely reduce the firearm related deaths.

Model Fit

The adjusted R-squared value is 70.3%. There are 70.3% of the variation in gun ownership rate, gun counts, gun registration, permit for gun carry, permit for gun purchase, open carry legal, and gun law score that related to firearm deaths. Beside the adjusted R-squared, the residual std. error is 2.84%. Averagely, the error in predicting firearm related death is 2.84% which is relatively low. This model is well-fit model.