There's been plenty of debunking aimed at the Weekly Standard's claim that more US troops in Iraq always means less violence, but I haven't seen this point. Stuntz argues that 18,000 more troops were deployed between November, 2004 and February, 2005, and the butcher's bill fell.
Well, it would. November, 2004 was the peak of the Fallujah campaign, the second Shia rising, and the insurgent seizure of Mosul. Violence continued fearsome into January, then dropped off some as the offensive wound down. It's simply an exercise in choosing a stretch of the trendline that agrees with you, then placing the goalposts at each end. Or to put it another way, an egregiously dishonest abuse of statistics.