The United States of America has the highest rate of school shootings in the world. A number of high profile cases such as Columbine and Sandy Hook are terrible reminders of how shooters can terrorize and destroy communities. Gun control and school safety have somewhat faded from news headlines as this analysis is being done but it still remains a public safety issue that the US Government has not completely taken on to find a solution that both Republicans and Democrats can agree on.
This analysis will look into the trends of school shootings around the country to see what the risks are to schools and what local, state, and federal government bodies can do to curb these shootings. Taking away guns from all Americans is always touted as a solution, but it is not a viable one. The Constitution allows the American people the right to bear arms and as of 2017, there are more guns than people in the United States (393 million guns). The logistical nightmare and resistance the US Government would face make it impossible to take away all guns. However, through this analysis, it may be possible to identify measured responses that could garner bi-partisan support.
#run to make everything work from datascience import * import numpy as np %matplotlib inline import matplotlib.pyplot as plots plots.style.use('fivethirtyeight') #name the table shooters.csv shooters = Table.read_table('shooter.csv')
First, taking data from the Center for Homeland Defense and Security’s (CHDS) K-12 School Shooting Database, we can generate a table to start manipulating.
With our table in place, let’s start by looking at the yearly data. The most deadly year and the year with the most incidents.
shootersdead = shooters.select('Year', 'Month', 'Killed (includes shooter)') a = shootersdead.drop('Month').group('Year', sum) #Yearly incidents and Total Dead Yearly_itd = shootersdead.drop('Month'). group('Year').with_column( 'Total Dead', a.column(1)).relabel( 'count', 'Incidents' ) Yearly_itd.sort(‘Total Dead’, descending = True) Yearly_itd.sort(‘Incidents’, descending = True
The year with the most deaths from 1970-2018 was 2018 with 56 total dead. The years with the least deaths were 1976, 1977, and 2011 all with only 4 deaths. Most observations will exclude the year 2019 because it has not concluded.
The year with the most incidents was 2018 with 97. The year with the least incidents was 1976 with 11.
If we plot these two distributions, the graphs show that there are more deaths when there are more incidents. This is generally true but not always. For example, the yellow line spikes above the blue line twice between 1990-2000 and once between 2010 and 2020. It also tells us that the number of incidents far outweighs deaths from school shootings. Between 1970 and 2019 there were 1356 incidents and 724 deaths. This means there were an average of 0.53 deaths for every incident. A little less than every other school shooting will result in the death of one student.
We can look at how this breaks down by year and find the deadliest year by deaths per incident.
updated_yearly = Yearly_itd.with_column( 'D / Inc', np.round( Yearly_itd.column(2) / Yearly_itd.column(1), 2) ) updated_yearly.drop('Incidents', 'Total Dead').plot('Year')
Now we can see which was the deadliest year or time period. The spike at the end represents 2019 which we can ignore. It appears that 2012 was the deadliest year with 2.11 deaths per incident, followed by much of the 90’s. The safest year was (excluding 2019) 2009, followed by 1977, 2015, and 2007 all hovering around 0.2 - 0.25 deaths per incident. Where does 2018, the year with the most deaths rank? 2018 ranks the 13th deadliest year from 1970 to 2019 but still has the largest number of both deaths and incidents recorded. The 90’s saw a dramatically sustained increase in violence that abruptly ended in the new millenium and the 2000/2010’s have a relatively low death per incident rate with a spike in 2012. We will examine the spike and individual incidents later on.
We can also deduce, from the data, that a large number of school shootings do not result in anyone dying which is confirmed in this graph. Roughly about 800 incidents resulted in 0 deaths, 450 resulted in 1 death, 50 in 2 deaths, and 11 in 3 deaths. This fact may be touted by those who oppose stricter gun control laws and want to maintain the status quo. If a large portion of incidents do not result in any deaths, then why spend time and money into solving this problem? Aside from the fact that there could be traumatic consequences for students involved in a school shooting, even with no deaths, or that no student should ever have to experience fearing for their life while in school, the database also contains those who were wounded. Let’s extract and visualize the total number of wounded victims and the number of wounded per incident.
According to the data, it looks like 2001 and 2009 were also very dangerous years for students involved in school shootings with total wounded being the highest at 109 in 2009, followed by 86 in 2001. Proportionally, the worst year for most wounded per incident was 1976 with 49 total wounded in 11 incidents (4.45 W / Inc). The gap in the graph is for 1991 where there were no recorded wounded students. Let’s overlay total wounded with total deaths and then do the same with deaths per incident and wounded per incident. I created a new table with the code below that contained the Year, Incidents, Total Dead, Death / Incident, Total Wounded, Wounded / Incident, and the Rank out of 40 for Death / Incident.
formatted_yearly.select( 'Total Wounded', 'Total Dead', 'Year').plot( 'Year' ) formatted_yearly.select( 'W / Inc', 'D / Inc', 'Year').plot( 'Year' )
There appears to be a loose inverse relationship between Total Wounded and Total Dead. In 1990, 2001, 2003, 2009, and 2015 when Total Wounded was high, Total Dead was lower. Similarly, when Total Dead was high in 1994, 2005, 2013, and 2018 where Total Dead was higher, Total Wounded dropped. We can see similar trends in the per incident graph as well.
Let’s see if this relationship is still illustrated through a scatter plot.
x = year, y = # of Individuals
x = year, y = Wounded/Dead per Incident
Where Total Wounded is high, Total Dead is low and where Total Dead is high, Total Wounded is low, generally. This can be explained by observing shooter behavior. If a shooter is trying to attack as many people as possible, then the Total Wounded will be higher. However, if a shooter is being more methodical and targeting specific people, groups, or has an agenda in mind, they will not waste their time or energy shooting the targets in sight. This is also why the sum of Total Dead is much lower than Total Wounded. Inflicting mass harm will not result in mass deaths, generally.
So how does one determine which year from 1970 to 2018 was the worst year for school shootings? Media companies in 2018 seemed to have agreed, from CNN to the BBC, that 2018 was the worst year for school shootings in the USA. In absolute terms, I would have to agree. Although 2018 ranked 13th for Deaths per Incident, it ranked 1st for both total dead at 56 and incidents at 97. However, if we look at which year had the deadliest single attack, it would be 2012 at Sandy Hook Elementary school with 28 killed. The second is the Parkland shooting in Florida in 2018 with 17 dead, and third is Columbine with 15 dead.
There is a heat map below with markers growing in size for each state that had victims in shootings.
Looking at this through a graph form may be easier to see the exact breakdown.
Rank 1-22
Rank 23-52
In the left graph, the distribution between rank 1 to about rank 22 is 86 total killed whereas between rank 23 and rank 52 (DC and US Virgin Islands included) the distribution is 10 total killed. California has roughly the same amount of fatalities as rank 23-36 combined.
At first glance, this makes it seem very scary to live in California or Texas, or Florida or anywhere in the US for that matter, aside from South Dakota, Rhode Island, Idaho, or Hawaii.
But, what do California, Texas, and Florida have in common as the top 3 deadliest states when it comes to school shootings? They are also the top three most populous states in the US. As of 2019, California is roughly 12.13% of the US population, Texas is 8.75% and Florida is 6.50%. Having larger populations will mean a larger probability of a school shooting happening. Let’s take a look at proportionality, though between the three states and see if things line up.
Judging by the calculations to the left, it looks like each state is about a percentage point lower than their population percentage. So it seems that population percent as compared to school shooting deaths percent is related. Until we look at New York which is roughly as populated as Florida but has had only 20 school shooting-related deaths. The number of New York deaths due to school shootings is 5.60% lower than Florida’s despite only being 0.46% lower in population. What is it about New York that makes it have less school shootings? New York was the 3rd safest state in the United States when it comes to gun violence per capita, coming behind only Massachusetts and Rhode Island.
US = 327,921,104 Deaths = 724
CA = 39,776,830 CADeaths = 96
TX = 28,704,330 TXDeaths = 67
FL = 21,312,211 FLDeaths = 55
NY = 19,862,512 NYDeaths = 20
CA / US = 12.13%
CADeaths / Deaths = 13.26%
TX / US = 8.75%
TXDeaths / Deaths = 9.25%
FL / US = 6.50%
FLDeaths / Deaths = 7.60%
NY / US = 6.06%
NYDeaths / Deaths = 2.00%
Many would think that a state like New York, with the country’s largest metropolitan center New York City, would rank lower in terms of safety. New York City had a reputation in the 70s and 80s, which still persists today, as a crime-ridden metro where streets were unsafe to walk and certain neighborhoods were treated like warzones. New York City’s crime rate has been steadily falling since the 1990s and many contribute the recent low gun-related crime rate to a piece of legislation from 2013 called SAFE. SAFE is the strictest gun-control law in the country and requires universal background checks.
For New York State Only
But this doesn’t completely explain the low school shooting-related death rate. From 2013 to 2018, there was only one death, but there was also only one from 2008 to 2013. How has New York managed to retain such a low school shooting death rate since around the end of the 90’s? Even lower than much less populous states like Pennsylvania, Washington, Illinois, Connecticut, and Colorado.
Mental health has been a factor in many school shootings and it may have a correlation on why New York state has a lower rate of school shootings. New York State mental health programs are robust and serve more than 700,000 individuals each year. The New York Office of Mental Health (OMH) operates psychiatric centers, has more than 4,500 programs operated by local government and nonprofit agencies which include inpatient and outpatient programs, emergency and community support, residential and family care programs. The total budget for 2016 to 2017 was close to 4 billion USD and was 40 million higher than in 2015 to 2016. This demonstrates mental health as a clear priority for the State.
Mental health is an underlying facet of people’s lives that effects everything from interpersonal relationships, perceptions of self, and overall behavior. A strong emphasis on mental health, especially on children and youth should be a top priority for federal, state, and local government. Societal biases surrounding seeking therapy and counseling still exist today and hinder people from getting the help they need. However, certain state initiatives like in New York City advocating for people to reach out and seek professional help for their loved ones can slowly change the narrative around mental health.
Following the Florida Parkland school shooting in 2018, Florida has allocated an additional 400 million USD for school safety and mental health. But, in the 2018-2019 budget, only about 10 million was given to the state’s mental health programs. With a similar population to New York, Florida is only funding its own programs by 10% of how much New York funds theirs. This does not prove causation but greater emphasis on mental health has some correlation to lower school shooting deaths.
Recommendations*
The United States federal government should enact universal and stricter gun control laws. This is something that should be embraced by both sides of the aisle to better protect their communities. It does not restrict the freedom to purchase guns but makes the process more rigorous and standardized. Mental health checks should also be included universally, however, the Trump administration has not been supportive of this.
State and Local Governments should put more emphasis on mental health programs. New York is a leader in this regard and other states should try to establish their own separate offices of mental health. Redistributing budgets to support mental health initiatives will help identify root causes and can prove to be more effective than further enhancing law enforcement and addressing the symptoms.
Non-profits, local governments, state governments, congressmen/women and all relevant stakeholders should send a strong message that mental health is not only important but that seeking professional help is encouraged. Even if more money or funding is poured into mental health services, if people feel that they will be judged for using them there will be no change.
*Mental health alone is by no means the only cause of school shootings. But, it is a salient factor and should be considered alongside other causes.