Hullabaloo

Saturday, September 08, 2007

 
Calling Statistics Geeks

by tristero

[UPDATED}

I think I'm pretty good at comprehending statistical information for a layperson. A long time ago I took a grad level course in stats and ruined the curve because my test grades were 20 points higher than the next student's.* But for the life of me, I can't square the data displayed on the charts accompanying this article with Michael Gordon's interpretation:

The most comprehensive and up-to-date military statistics show that American forces have made some headway toward a crucial goal of protecting the Iraqi population. Data on car bombs, suicide attacks, civilian casualties and other measures of the bloodshed in Iraq indicate that violence has been on the decline, though the levels generally remain higher than in 2004 and 2005.

However, what I see, if one looks at trends longer than a few months, is no significant decline but rather that, at best, the "surge" has prevented things from getting worse. The only improvement appears to be a decline in sectarian attacks (the last graph), but it is unclear from the data in that chart whether the drop off represents a longer trend. I talking in general here, obviously. There just doesn't seem to be significant changes pre/post "surge."

Anyone knowledgeable about stats care to comment and educate me/us? BTW, ad hominem remarks about Michael Gordon won't be terribly helpful. The issue is not his biases, which we all know. What I'd like to know, instead, is whether the conclusions he draws in the paragraph quoted are reasonable ones based upon the evidence he presents. Likewise, I'm aware of the "correlation does not necessarily equal causation" fallacy which Gordon flirts with. In this case, tho, I think it is very reasonable to assume that additional military power might "cause" decreased numbers of attacks. I just don't see them; the improvements, except as noted above, seem to be mostly temporary fluctuations.

As far as I can tell, he is stretching to be optimistic. If so, his follow up question is utterly disingenuous:

Can the drop be sustained over the coming months, especially with the approach of Ramadan, the Islamic holy month in which violence has often increased?

because it implies an improvement that is illusory but which, nevertheless, might be useful to try to sustain.

I'm serious. I'm puzzled and hope that someone who really, really knows how to study statistical data might want to take a moment and weigh in.

--

[UPDATE: Thanks, folks, for a very interesting discussion. From what I gather, the data are displayed in ways that make it difficult to compare apples to apples, and that many of you have doubts about the integrity of the data. Even assuming the data are accurate and honestly presented, I gather that none of you think "the surge" has accomplished very much.

As it happens, The Times agrees with you; they don't share Gordon's optimism and they sent additional reporters to Baghdad to get second opinions:

To study the full effects of the troop increase at ground level, reporters for The New York Times repeatedly visited at least 20 neighborhoods in Baghdad and its surrounding belts , interviewing more than 150 Iraq residents, in addition to sectarian militia members, Americans patrolling the city and Iraqi officials. They found that the additional troops had slowed, but far from stopped, Iraq’s still-burning civil war. Baghdad remains a city where sectarian violence can flare at any moment, and where the central government is becoming less reliable and relevant as Shiite or Sunni vigilantes demand submission to their own brand of law.

“These improvements in the face of the general devastation look small and insignificant because the devastation is so much bigger,” said Haidar Minathar, an Iraqi author, actor and director. He added that the security gains “have no great influence.”

The troop increase was meant to create conditions that could lead from improved security in Baghdad to national reconciliation to a strong central government to American military withdrawal. In recent weeks, President Bush and his commanders have shifted their emphasis to new alliances with tribal leaders that have improved security in Diyala Province, the Sunni Triangle and other Sunni areas, most notably Anbar Province.

This, not Baghdad, was the area Mr. Bush conspicuously chose to visit this week.

But when he announced Jan. 10 his plan to add 20,000 to 30,000 troops to Iraq, Mr. Bush emphasized that Baghdad was the linchpin for creating a stable Iraq.]

[UPDATE: Josh finds Gordon overly credulous in accepting the numbers from the military and recc'ds this WaPo article for another point of view. ]

*[UPDATE: Perhaps I should explain that it was a class for clinical psychology grad students who, if this class was any indication, are pretty ...challenged... when it came to grasping quantitative data. But they had other virtues. Many of the students, I'm sure, have gone on to become superb psychologists.]

tristero 9/08/2007 05:54:00 AM |