If you have been on NY Magazine's "Science of Us" blog during the last two days, you might have seen this graph along with the conclusion that mass shootings in the United States are not on the rise:

The author looks at the graph, shrugs their shoulders and concludes "It's clear that there is no major upward trend." It has been quite a few years since I was a math student but this struck me as odd. Most strange is that, for a science blog, they should know that one does not analyze data simply by looking at it but with statistics. Posting this graph without analysis is both lazy and potentially dishonest. I went ahead did a linear regression on the linked raw data:

Despite the apparent scatter, there is actually an average trend upward. The increase is statistically significant for victims at a 95% confidence but for the number of incidents the relationship is not as certain (85%). So, although the data does not provide a super strong case for incidents being on the rise, it does suggest that the number of victims are. So, let's look at the number of victims per incident:

Indeed, there is an upward average trend for victims per incident (90% confidence). Moreover, one can see that the two most deadly years were in the last decade. So, far from simply being a product of an "availability heuristic" (perception of increase just based on more media coverage), it seems plausible that mass shooting victimization is on the rise. Of course, if one subtracted the outlier years, the relationship would likely weaken somewhat.

However, the above analysis is based on the absolute numbers of incidents and victims. What about the per capita figures? 

The best fit lines from the linear regression when the number of victims and incidents are divided by the United State's population figures seem to suggest a flat or downward trend in the per capita rates of mass shooting incidents and victims. However, for both cases the relationship is NOT statistically significant. 

This analysis took all of about twenty minutes to do in Excel and resulted in conclusions very different from those offered by the Science for Us blog. The data suggests that there are more victims and incidents today than in previous decades, though that fact is probably more related to the growth in the US population than an increased propensity toward mass shootings. However, it does seem that mass shootings have been (on average) getting more and more deadly. 

Nevertheless, one should not lose sight of the fact that these events have not declined along with the firearm homicide rates and could still be considered to occur far more frequently than is desirable. Motivation for sensible gun control measures and other changes to public policy does not rely on discovering a growing epidemic in the data but simply the belief that such needless violence could be prevented. Research continues to suggest that lax regulations combined with a strong "gun culture" contributes significantly to America's incredibly high rate of firearm related crime (compared to other countries). Gun violence, of course, is a complex issue that I think is not simply solved by restricting access but also providing better economic and political opportunities to those Americans more likely to be both the perpetrators and victims of gun violence (e.g., poor minorities). Furthermore, it is not just a civilian issue, police in this country shoot and kill far more citizens than other countries and are increasingly militarized to boot. Regardless, I will save a more in-depth analysis of gun violence for another post. 

Share