UPDATE 10/28/18: In the original publishing of this report we rounded several numbers poorly or in a way that didn't portray the data as accurately as possible. For example, 75.8% was rounded to 75% and other similar issues. Further, the pie charts displayed in the graphics were not proportionally accurate and were designed more with visibility in mind as opposed to an accurate representation. These issues have all been corrected below. If you would like to download the original data you can still do so via the link below.
========================================================
After seeing John Lott from the Crime Prevention Research Center publish a rebuttal to a recent FBI report on Active Shooter events I decided to dive deeper into the data and do some analysis of my own. WARNING, this is a long article/report with a lot of images but I am confident it is worth your time!
The Original FBI Data Source
The FBI has published 3 reports that collectively detail active shooter events from 2000-2017. The first report covered events from 2000 to 2013, the second covered 2014-2015, and the third and most recent covered 2016-2017.
It is important to note that the FBI has no specific system in place for finding and cataloging active shooter events. They manually search for and include them in their reports the same way anyone else might Google it which of course means there is room for error particularly in missing events that should have been included.
The FBI definition of an Active Shooter event is: “One or more individuals actively engaged in killing or attempting to kill people in a populated area.”
A few important distinctions about the FBI definition of Active Shooter include:
- A firearm must be used by the attacker. This then means they have not included incidents like the armed citizen who saved a woman outside the GM building in Detroit from a stabber or the man who was stopped by a CCWer in a Smiths Grocery store in Salt Lake City when he was stabbing shoppers at random.
- Domestic incidents are not included. The FBI feels that an Active Shooter event has to be one in which the attacker is endangering strangers not only their own family members.
- Gang-related violence is excluded also.
- For the FBI to define an incident as an Active Shooter incident both law enforcement personnel and citizens have to have the potential to affect the outcome of the event based upon their responses to the situation.
So Is The FBI Data Complete?
Within the Active Shooter definition used by the FBI, it is broad enough that there are likely a large number of incidents that are being missed by the FBI. The Crime Prevention Research Center has taken the lead after each report has been published to identify events that should have been included that were missed. In some of those cases, the FBI has acknowledged their error but still never updated the list of events.
Lott found that there was a greater tendency to miss events from the first decade (2000 to 2010) than in more recent years. This is at least in part to the changes in technology and news reporting. In 2014 when the FBI did their first report it would have been difficult to search for and find Active Shooter events from the early 2000s. Lott suggests there may also have been some intentional bias in not reporting on some earlier events in order to show a greater increase in incidents over time.
So, for our own report that follows, we have included all of the FBI data but have also added a number of incidents that the FBI missed which were identified by the CPRC. Of the 283 Active Shooter events in our data pool; 248 of them come from the FBI's original reports while an additional 35 identified by the CPRC have been added. I carefully reviewed each of those 35 incidents to make sure they meet all the FBI Active Shooter criteria.
So What Does the Data Show?
This first chart simply shows the number of active shooter events over time. We believe the first 10 years or so reported are likely under-reported by the nature of how the data was compiled, but regardless one could arguably suggest that Active Shooter events are increasing.
Part of the gun-control political debate is the argument that semi-automatic rifles and high capacity magazines somehow increase the potential death rate in Active Shooter Events. In order to really draw any conclusions, one would have to know, for each incident, if a semi-automatic rifle and/or high capacity magazine was used. That information isn't available in the FBI's report or publicly available for our own research team to find.
So we encourage you to not draw any strong conclusion but as you can see from the below chart there doesn't appear to be any particular pattern or trend over this 18 year period despite the Assault Rifle ban of 1994 expiring in 2004 or the other various bans on AR-15s and high capacity magazines in a number of states.
We thought it would be interesting to see if the overall percentage of events at which an armed citizen was present was increasing given the rise in concealed carry permit holders and gun owners nationwide. The data set is low enough that it is hard to draw any strong conclusions when you are looking at 33 incidents out of 283 over an 18 year period.
It is worth noting that there may be more than 33 incidents in which an armed citizen was present given that we have no way of knowing if an armed citizen chose NOT to engage and run the other way. That may not even be known to law enforcement. Our objective here is to look at incidents in which an armed citizen was clearly present and to some degree engaged the active shooter.
The white line below shows the average by year which looks very up and down. The yellow line is a three-year moving average which does paint a bit of a picture that the likelihood of an Armed Citizen being present is increasing; especially in the last 3-4 years.
Looking at the 283 total Active Shooter events in our data pool, an Armed Citizen was Present and Engaged the Active Shooter in 33 total incidents (11.7%). This is all inclusive regardless of who the armed citizen was or their direct potential for stopping the shooter.
In a few examples, the armed citizen was at their home near the event when they heard shots fired and rushed to the scene to intervene and thus despite not being present when the incident began those Active Shooter events are included in the 11.7% below.
In one other example, the victims of the attack were hunters that were effectively ambushed by their killer. We are assuming the hunters possessed firearms and thus that incident is included in the 11.7% below even though the armed citizen wasn't attempting to intervene to save others but was, in fact, the targeted victim.
We decided to take a look at one specific metric that John Lott and the FBI didn't really consider. The success rate of armed citizens. John Lott's team did talk about the overall success rate of armed citizens against all Active Shooter incidents in the US but they failed to consider the most important variable. OPPORTUNITY.
If we were to look at 100 active shooter events and an armed citizen was present at 1 of them and succeeded at stopping the active shooter then certainly we can say that armed citizens stop active shooters 1% of the time but in doing so we imply that armed citizens failed to stop 99% of active shooters.
This is an inaccurate implication since no armed citizen was available to stop the active shooter in the other 99 incidents. More helpful when considering the effectiveness of armed citizens in stopping active shooters would be to accurately state that Armed Citizens stopped 100% of active shooters at incidents at which an armed citizen was available to do so.
So the below graphic does just that. Of all the active shooter events there were 33 at which an armed citizen was present. Of those, Armed Citizens were successful at stopping the Active shooter 75.8% of the time (25 incidents) and were successful in reducing the loss of life in an additional 18.2% (6) of incidents. In only 2 of the 33 incidents (6.1%) was the Armed Citizen(s) not helpful in any way in stopping the active shooter or reducing the loss of life.
Thus the headline of our report that
Armed Citizens Are Successful 94% Of The Time At Active Shooter Events.
In the 2 incidents at which the armed citizen “failed” to stop or slow the active shooter, one is the previously mentioned incident with hunters. The other is an incident in which the CCWer was shot in the back in a Las Vegas Walmart when he failed to identify that there were 2 Active Shooters involved in the attack. He neglected to identify the one that shot him in the back while he was trying to ambush the other perpetrator.
We also decided to look at the breakdown of events that took place in gun free zones and the relative death toll from events in gun free zones vs non-gun-free zones.
Of the 283 incidents in our data pool, we were unable to identify if the event took place in a gun-free zone in a large number (41%) of the events. Most of the events took place at a business, church, home, or other places at which as a rule of law it is not a gun free zone but potentially could have been declared one by the property owner. Without any information in the FBI study or any indication one way or the other from the news reports, we have indicated that event with a question mark.
If you look at all of the Active Shooter events (pie chart on the top) you see that for those which we have the information, almost twice as many took place in gun free zones than not; but realistically the vast majority of those for which we have no information (indicated as ?) are probably NOT gun free zones.
If you isolate just the events at which 8 or more people were killed the data paints a different picture (pie chart on the bottom). In these incidents, 77.8% took place in a gun-free zone suggesting that gun free zones lead to a higher death rate vs active shooter events in general.
I was curious if there would be any significant trend in looking at where these events took place. We defined a total of 12 location categories and categorized each event based on where the incident BEGAN. In many Active Shooter events, there are multiple locations where the attacker acts but for the purpose of this report we looked only at where the incident first started.
Since gun laws vary from state to state a common part of the political discussion is the effectiveness of state laws in stopping or slowing the instances of Active Shooter events. I don't think it particularly wise to draw conclusions from the below chart for a few reasons.
First, the data pool isn't big enough. 283 total events equate to about 5 1/2 per state on average … over an 18 year period. For example, North Dakota had 1 incident with a population of fewer than 1 million people. Its possible another 10 years could go by without any more incidents so calling it a high probability Active Shooter State is logically unsound.
One of the final metrics we thought was important to consider is the potential tendency for armed citizens to injure or kill innocent people in their attempt to “save the day.” A common point in political discussions is to point out the lack of training of most armed citizens and the decrease in safety inherent in their presence during violent encounters.
As you can see below, however, at the 33 incidents at which Armed Citizens were present, there were zero situations at which the Armed Citizen injured or killed an innocent person. It never happened.
That concludes our reporting. I have a few requests of anyone who reviews this.
First, share this on all your social media profiles so that accurate information can debunk the myths and propaganda shared by both sides of the political debate.
Third, you can use the comments below to let us know what thoughts or questions you have about the report.
Fourth if you would like to listen to our analysis of this data, use the below audio player to tune into Episode 266 of
The Concealed Carry Podcast.
Lastly, below is a complete infographic. For those of you who have an interest in embedding this on your site, you can find the embed code below as well.
Sources:
Embed This Image On Your Site (copy code below):