The FBI says "active shooter" incidents are on the rise. What does that mean?

Fox isn’t convinced. “Unlike mass shooting data,” he says, “which come from routinely collected police reports, there is no official data source for active shooter events. Necessarily, these data derived from newspaper searches for the term ‘active shooter’ and similar words. Not only is the term ‘active shooter’ of relatively recent vintage (although created after the 1999 Columbine shooting, it wasn’t used much in news reports until the past two years), but the availability of digitized and searchable news services has grown tremendously over the time span covered by the data. Thus, it is not clear whether the increase is completely related to the actual case count or to the availability and accessibility of news reports surrounding such events.” Fox acknowledges that the new study draws on police records as well as press accounts, but he says the cases still have to be initially identified by searching news reports, “with police department records helpful for the details.” (Sure enough, when the FBI paper identifies its sources, it cites several studies that rely on such searches.)

Grant Duwe is skeptical about the numbers too. Duwe is the director of research and evaluation at the Minnesota Department of Corrections; each year he compiles a list of mass shootings that take place in public and are not a byproduct of some other felony. (He would not, for example, include a stick-up man who shoots several people while fleeing the police.) His figures show a steady decline from 1999 to 2011 and then a spike in 2012, which is not the pattern you might expect from those active-shooter numbers. While he appreciates some aspects of the FBI’s report—he singles out its “detailed breakdown of the specific locations where these incidents occurred”—he has issues with it as well, noting that 10 of the mass shootings on his list are missing from the new study. (These omissions appear throughout the period covered, without being clustered at either the beginning or the end.) “Given that incidents involving fewer victims generally attract less attention,” he adds, “I imagine the amount of underreporting for these cases is greater.” Since the FBI’s list “is neither a random nor an exhaustive sample,” he concludes, “it’s inappropriate to make any claims—which this report does—about trends in the prevalence of cases meeting their definition.” He thinks it possible that such a rise has happened, but he doesn’t think the report proves it.