In discussing a book on the 1999 Columbine shootings, Gamepolitics mentioned a data point from a 2002 Secret Service study cited in the book. Only one in eight people who engaged in school-targeted violence expressed an interest in violent videogames.
Only one in eight?
The study, available here as a PDF, considered 41 people who deliberately selected their own schools as the location for a violent attack. About a quarter of them were interested in violent movies or books. Nearly a third expressed violence in their own writing. But only an eighth expressed an interest in violent videogames.
It's worth noting, however, that the sample group of 41 people goes back to 1974. Basically, violent videogames didn't even exist for half of the time that the incidents took place.
It's an odd data point, and there is, of course, no attempt at a causal link. But given how children are growing up these days, I expect that number in a more current study would be much higher than one in eight.