Gun Deaths in the United States (2012-2014)

Introduction

When compared to other nations the United States has a substantially larger rate of violent gun deaths. In comparison to the countries with the lowest rates of gun violence the United States has a rate almost 200 times higher (1). Compared to countries with the highest rates, the United States is not as high, however there are other issues in these countries that contribute to the higher rates such as gang presence, drug trafficking, political unrest, and economic meltdown that are not present in the United States (1). With the exception of the Philippines, the United States also has a higher violent gun death rate than the Asian countries with the highest rates (1). The rate is even higher than North African and Middle Eastern countries (after armed conflict deaths are factored out) (1).

The United States owns 46% of the world’s civilian firearms, while making up only 4% of the world’s population (2). The United States has the most guns per population in the World (3). The United States has an ownership rate three times higher than the next-highest country, Canada (2).

In 2017, suicide accounted for 60% of deaths related to guns and 37% of gun related deaths were homicide (4). The homicide rate for Blacks was 267% higher than the national average in 2015, and 54% and 14% lower for Whites and Hispanics respectively (5).

In the last 11 years, suicides involving guns has increased yearly (4). In 2017, gun accounted for 50.57% of all suicide deaths (6). Suicide rates in the United States, are the highest among Whites and Native Americans/Native Alaskans. Asians/Pacific Islanders have the lowest suicide rates in the country (6).

There are many gun related deaths in the United States each year. I want to determine if there are any significant relationships between these reported deaths.

Research Questions

Are police shootings more common with a specific race or gender?
Is there a connection between gender and suicide?
Is there a relationship between race and suicide?

Data

The data used for this analysis is the “Gun Deaths in the U.S. 2012-2014” dataset from kaggle.com. The dataset includes information about the deceased’s sex, race, age, and education level. As well as information about where they died, the intent (suicide, homicide, accidental, or unknown) and whether a police officer was involved in the shooting or not.

The dataset contains 100,798 observations. The dependent variables used were police (0 = “No”, 1 = “Yes”) indicating if a police officer was involved in the shooting and intent (0 = “Suicide”, 1 = “Homicide”, 2 = “Other”). The independent variables used were sex, race, and age.

Results

Descriptive Analysis

According to the observations in this dataset, Whites have the highest suicide percentage with 83.6% of the gun related deaths for Whites being suicide. Native Americans/Native Alaskans have the second highest percentage with 60.5% of their gun related deaths also being suicide. These are followed by Asians/Pacific Islanders, Hispanics, and Blacks with suicide percentages of 56.2%, 35.1%, and 14.3% respectively.

The homicide percentages according to this dataset are highest among Blacks, with 83.7% of the gun related deaths for Blacks being homicide. Hispanics have the second highest homicide percentage with 62.4% of their gun related deaths being homicide. These are followed by Asians/Pacific Islanders, Native Americans/Native Alaskans, and Whites with homicide percentages of 42.2%, 35.6%, and 13.8% respectively.

Small percentages for each race, less than 0.05%, of gun related deaths were accidental or unknown.

Regression Analysis

Several regression models were created in order to determine the what factors influence police involvement in the gun related deaths. Three logit models were created in order to determine which model was the best fit. The first model analyzed police involvement and sex, the second model had race added, and the third model was adjusted to include age as well. The best fit model was the third, that included all of the independent variables.

From this model, I was able to determine that the log odds of a police officer being involved in the shooting increases by 1.26 units for males. The log odds also increases if you are Hispanic or Native American/Native Alaskan by 0.18 and 0.12 unites respectively. The model also shows that the log odds decreases by 0.6 for Blacks, and 0.56 for Whites. It also shows that as age (of the deceased) increases by one year the log odds of a police officer being involved in the shooting decreases by 0.02.

Statistical models
	Model 1
(Intercept)	-4.14^***
	(0.23)
sexM	1.26^***
	(0.13)
raceBlack	-0.60^**
	(0.19)
raceHispanic	0.18
	(0.19)
raceNative American/Native Alaskan	0.12
	(0.27)
raceWhite	-0.56^**
	(0.19)
age	-0.02^***
	(0.00)
AIC	14279.27
BIC	14345.91
Log Likelihood	-7132.63
Deviance	14265.27
Num. obs.	100780
p < 0.001, p < 0.01, p < 0.05

Another series of logit models were created to determine the influence of other factors on suicide. In order to determine any patterns between the gun related deaths for suicide, I created three models in order to determine which is the best fit. The models were created similarly to the first set of models, using the same independent variables, with intent (filtered for suicide) now being the dependent variable. Once again, the third model was the best fit, that included all the independent variables.

From the best fit model, I was able to determine that the log odds of committing suicide increases by 2.05^-13 units for males. It then shows that the log odds decreases by 1.43^-11 for all the races (this makes it difficult to say there is any correlation). The model also shows that as age increases one year the log odds decreases by 3.16^-15 units.

However, these numbers are so small they are statistically insignificant. From this dataset it is difficult to determine the causation between suicide and race, gender, and age.

Statistical models
	Model 1
(Intercept)	-26.57
	(13821.77)
sexM	0.00
	(4123.85)
raceBlack	-0.00
	(14433.32)
raceHispanic	-0.00
	(14499.76)
raceNative American/Native Alaskan	-0.00
	(19979.66)
raceWhite	-0.00
	(13162.65)
age	-0.00
	(75.55)
AIC	14.00
BIC	77.37
Log Likelihood	-0.00
Deviance	0.00
Num. obs.	63168
p < 0.001, p < 0.01, p < 0.05

Conclusion

The analysis conducted was able to determine that it is more common for police to be involved in deadly shootings of males. It also determined that Hispanics and Native Americans/Native Alaskans have a higher log odds of being shot by the police. However, this is not supported by other sources.

The analysis conducted supports previous findings that suicide rates are higher among men. The analysis was unable to determine a significant relationship between race and suicide, however other sources have been able to determine that suicide rates are highest among Whites and Native Americans/Native Alaskans.

There are limitations to the analysis, including more than half of the observations in the dataset are suicide victims and that there are not enough observations for police involved deaths for a more accurate analysis.