In the wake of the massacre in Las Vegas a study was published from the University of Illinois stating that mass shootings in America are not more common now than in the past. Researchers looked at shootings in which four or more people were killed between Jan. 1, 2006, to Oct. 4, 2016, and found the method and timing of the shootings are random and hard to predict.
The study’s authors, Douglas King and Sheldon Jacobson, argue that “numbers don’t lie,” which is true but some statistics don’t always tell the whole story.
The 10-year period covers shootings like those that occurred at Virginia Tech, Newtown, Conn., and Orlando, but does not include others such as Columbine and Las Vegas. Mother Jones has an online database of shootings of three or more fatalities from 1982–2017 that includes Las Vegas.
A graph of the distribution of mass shootings can be seen below.
The data set listing the information about the shootings was downloaded from the Mother Jones site to test their claim about the number of mass shootings. The number of shootings and fatalities are presented below, as well as the number of fatalities per shooting for each of the 35 years considered.
The fatalities per shooting are presented to see if the shootings are deadlier now than in the 1980s. The raw numbers for each year are presented in the blue bar graphs and the two-year moving average is shown in the black line.
The graph above shows the number of incidents have increased from one incident per year in the 1980s to 5.1 per year in the 2010s. The graph below shows the total number of fatalities per year mirrors the number of shootings in the same year.
At the bottom, the graph shows that the number of fatalities per shooting in the 2010s. Fatalities per shooting haven’t increased appreciably above the peak years 1986 and 1990, even with Orlando and Las Vegas included.
This is surprising given that more sophisticated assault weapons are available now.
Numbers don’t lie but numbers that summarize other numbers don’t tell the whole story. One must consider all of the available data.
By restricting the sample of incidents to those with at least four fatalities excludes five incidents with three deaths for 2017 and two for 2016. This would lead to a similar conclusion as reached by the University of Illinois’ researchers.
One’s definitions and assumptions make all the difference.