The daily fatality rate for coalition troops since the codpiece moment three years ago has steadily risen. (Note, the linear regression above does not incorporate the first month's data, so it's a regression of the fatalities during the occupation alone.)
The lowess regression shows that the fatality rate has been leveling off, but it's also noteworthy that 2004 was highly variable, two months have rates which are almost 2.5 standard deviations from the line, and two others which are nearly 2 deviations from the expectation.
If we reweight the loess regression by the inverse of the residuals, the line rises slowly over the last two years.
The point being: the rate at which soldiers are dying in Iraq has been rising no matter how many times the President exhorts the press to tell more pleasant tales. That means more mothers, fathers, wives, husbands and children are being taken away from friends and family every day we stay in Iraq.