Is the Occurrence of Mass Shootings Increasing?
This question seems ridiculous given that last Saturday the United States experienced two mass shootings within a 14-hour period. Certainly this was a shocking turn of events that has left us all reeling. But it turns out that this question is more difficult to answer than our emotions would suggest.
The header figure for this post was chosen to make this point. It shows data on mass shootings in the United States from three information sources, those being: (1) Mass Shooting Tracker, (2) Mother Jones and (3) the FBI. Note that for the year 2013 (the only year for which data is available from all three sources) the number of mass shooting events are 363, 5 and 17! The number of deaths and injuries all varies widely across these three sources. Also, no trend in number of events, deaths or injuries is apparent in any of these sources.
The primary reasons for this huge variation across sources are determined by the methodology and definitions used. For example, the Mass Shooting Tracker uses a crowd source (use of information contributed by the general public, often via the Internet and without compensation) methodology and a broad definition for mass shootings. Mother Jones and the FBI use much narrower definitions (but not identical) for mass shooting and the data is collected and validated by that organization. While there are cases for which crowd sourced data is credible, the uncontrolled nature of this methodology isn’t a good match for mass shootings.
Since Mother Jones* has the longest time span (1982 to 2019), a consistent definition and centrally managed methodology I will focus on their data. There are many ways to display data of this type. The following figure shows two key aspects of the mass shooting data, those being number of events and number of deaths per year.
It’s easier to identify trends if we plot the number of annual events and number of deaths separately, as shown in the following two figures. For each figure I have included the fifth-order polynomial best (least squares) fit curve to highlight the underlying data trend (dashed line).
The trend line clearly shows the number of mass shooting events began to rise significantly starting approximately in 2009. Note that this rising trend continues through the last full year of 2018.
The trend line clearly shows the number of mass shooting deaths began to rise significantly starting approximately in 2007. Note that this rising trend continues through the last full year of 2018.
These results indicate that our perception of significantly increased events and deaths due to mass shootings is backed up by the available data. Of course the Mother Jones data set isn’t perfect (nor is any other). However I’m satisfied that it is sufficient to accurately identify the trends associated with this tragic situation.
* Comment on credibility criteria
Some readers may wonder why I’m using mass shooting data from a very Progressive organization. After all, a primary theme of this blog is severe criticism of Progressivism. There’s no doubt that I strongly disagree with Mother Jones on interpretation of events and policy positions. However, all this doesn’t necessarily mean that Mother Jones’ mass shooting data itself is tainted. In fact, based on their methodology, definitions and transparency I’m comfortable using their data.
Mother Jones has clearly communicated their definitions and methodology for identifying and cataloguing mass shooting events. Of absolute critical importance, they are also completely transparent with respect to the results. I or anyone else can access their entire data set and evaluate it for consistency, accuracy and completeness. This implies they are making a good faith effort to generate useful information in this critical area.
Finally, I note that if anything the Mother Jones data is conservative even when compared to the FBI data (see the above figure). Were they attempting to bias their data to accentuate the number of mass shootings I very much doubt that this result would have occurred.
Note that the credibility of any data source will be severely compromised if these conditions are not met.