University of Alabama criminologist Adam Lankford has done a lot of interesting and provocative work on terrorists and mass killers -- arguing for instance that suicide bombers should be viewed not as rational political actors but as disturbed suicidal individuals motivated by similar factors to school shooters like those and Columbine or Sandy Hook.
A recent study by Lankford looks at data on mass shootings in the United States between 1966 and 2010 in order to look at the differences between shooters who survive their attacks and those that die with their victims. Using an NYPD database of "active shootings" -- those engaged in killing people in a populated area, excluding gang shootings, domestic violence, robberies, drive-bys and other more conventional crimes -- the study isolates a sample of 185 attacks, with the number of casualties ranging from two to the 32 killed by Seung Hui-cho at Virginia Tech in 2007. The shooters were 96 percent male with a mean age of 34.03.
38 percent of perpetrators of mass shootings commit suicide by their own hand and 48 percent die in the attacks. (Some of those extra 10 percent could probably also be categorized as "suicide by cop".)
Lankford also found that for each additional victim killed in the attack, the shooters' likelihood of dying was 1.2 times higher. Shooters are also far more likely to die when they bring multiple weapons to the scene and when attacking factories or warehouses rather than schools and office buildings.
The data in the paper is just from U.S. shootings and it would be interesting to know if the same patterns hold in other countries. Europe's most notable mass killer in recent years, for instance, is still very much alive.