O'Reilly, Kilmeade, Williams, and the Muslimness of Terrorism

The recent brouhaha, and Brian Kilmeade's statement that "All Muslims aren't terrorists but all terrorists are Muslim," prompted me to do a little number crunching:

I went to the START Global Terrorism Database and pulled all attacks between 2000 and 2008 with 51 or more casualties. (You can't export enough records if you include the small attacks.) I then tagged any attack with a known perpetrator that was clearly Islamic (i.e. I knew off the top of my head that the group was Islamic or it had "Islam", "Mujahadeen", "Mohammed", etc.) or, for unknown perps, that occurred in a country that was nearly uniformly Muslim.

Results:
Number of attacks with greater than 50 casualties: 424
Number of Islamic attacks: 342
Percentage Islamic attacks: 81%

Total casualities in attacks with >50 casualties: 49,842
Total casualties in Islamic attacks: 42,029
Percentage casualties from Islamic attacks: 84%

So Kilmeade was (obviously) incorrect. But you have to admit that Muslims being responsible for 84% of the casualties (or more--I didn't include unknown Indian attacks, many of which will be Muslim-originated) is kind of a large percentage.