Friday, November 11, 2005

Armistice Day thoughts on Iraq

Fatalities to date

The graph is made of data from the Iraq Casualities website. To control for differences in month lengths, the data are averaged to daily rates per month.

The loess regression is a locally weighted method, and shows relatively little change in the rate at which fatalities are increasing over time.

A while back, Dean Esmay and lies.com got into a thing over how to read this casualty data. I agree with lies.com that the monthly data are more appropriate, but I wanted to include the last month in the graph above, and I needed to weight by day length to make that work. As lies.com notes, there are individual days, and short runs of days that had higher average fatality rates than the initial invasion period, and this way of graphing confuses the issue.

Dean and lies.com argued about whether there was an upward trend over time (l.c: yes, Dean: no). Dean's funniest argument was that the appropriate regression was a four point model using aggregated data for each of four "phases" of the operation. The problem is, there's no empirical justification for those divisions. Look at the graph and find inflection points. They aren't there. And there's an unambiguous upward trend. The regression is tricky because of autocorrelation (the following month is likely to have similar casualty rates to this month, more similar than you'd expect of a randomly chosen month). Since a linear regression is close to statistical significance, that matters a lot.

As a consequence, we cannot substantiate the claim that events like the transfer of sovereignty (late June, 2004) or the elections (late January, 2005) made any persistent, measurable change in the insurgency (at least in terms of effectiveness at delivering fatalities).

All of which is very technical and of limited interest.

What's interesting is that these are a lot of (mostly) young American men and women who died in Iraq.

They died because the Bush administration lied about aluminum tubes. They died because the Bush administration lied about uranium from Africa. They died cleaning up a mess that resulted from poor planning by Don Rumsfeld.

It's telling that a lot of veterans are coming home from Iraq and running for office. They want a different direction, and are ready to serve and help.

If my sources are to be believed, we'll have another vet to add to that list soon, taking on Jim Ryun. A veteran vs. a runner. I like that fight.

Update: Apparently,

These Dems are dangerous, Treasonous and getting away with it. You don't say a president lied when it was proven he didn't lie, and when there is no evidence of it during a time of war. They are aiding and abetting the enemy.
Also, we "should question their patriotism. Because they're acting unpatriotically."

No. Asking these sorts of questions are exactly what Congressmen are supposed to do. It's called oversight. While the concept has languished in recent years, it's why Congress exists.