I saw the thread where this number was derived. It included only 19 incidents since 1983. It included only successful attacks and ignored unsuccessful ones; (ie:the shoe-bomber). As I remember, it included only attacks where Americans were killed and ignored attacks against other Western free-world targets.
It is an accepted statistical principle that any result derived from a small sample spread over a wide base (years in this case) is liable to be bogus - no meaningful conclusion can be drawn from the sample.
An extreme example would be to conclude that the Islamists attack only once every ten years if there are no more attacks until September 11, 2011.