I printed out the study - I was thinking it might be something I could give my Stats students to look over - but it doesn't really detail any of the input data, or methodology for computing his final numbers. It just describes the numbers he presents.
I suspect either the period he were looking at ('76-'84) was much more violent than average, or they were using some unsound methods. But, again, it's hard to say without looking at his raw data or methodology.
As Samuel Clemens (I think) said: "there are lies, damned lies, and statistics". Statistics are far more subject to manipulation than most any other form of mathematics, and it would be in the DOJ's interest to paint a pretty dark picture for funding's sake - much like they release studies that show how gun control reduces crime.
si vis pacem para bellum