Tuesday, February 27, 2007


Analytic Challenge

It seems like there’s an awful lot of analysis out there – note I didn’t say a lot of awful analysis – that warrants a lot of serious, statistical or logical challenge.

I have been tracking an ongoing conversation that springs from new reporting from Mother Jones, of all places, journalism-wise. As part of a series “Iraq 101,” Mother Jones hosts a piece written by Peter Bergen and Paul Cruickshank, entitled The Iraq Effect - The War in Iraq and Its Impact on the War on Terrorism.

Bergen is as reliable an anti-war critic as can be found, but he strenuously attempts to do so with evidence and data he marshals to achieve that end. He serves his purpose well, to the adoring and enduring gratitude of audiences for media such as The New York Times and Mother Jones. Preaches to the choir, he does, and thus his analysis predictably follows the liberal hymnal.

But don’t let me prejudice my readers’ opinions, here’s the raw data Bergen uses as cud for his chew, bottom line up front:


The Data:The Iraq War and Jihadist Terrorism
Period 1:September 12, 2001, to March 20, 2003 (invasion of Iraq): 18.25 months
Period 2: March 21, 2003, to September 30, 2006: 41.33 months

Let me don a statistical cap for a moment. Bergen qualifies his data selection, because indeed, the data shown above is an extraction of a fuller set of data that may or may not bear out the same arguments that Bergen makes on the basis of his custom set of data:

In order to zero in on The Iraq Effect, we focused on the rate of terrorist attacks in two time periods: September 12, 2001, to March 20, 2003 (the day of the Iraq invasion), and March 21, 2003, to September 30, 2006. Extending the data set before 9/11 would risk distorting the results, because the rate of attacks by jihadist groups jumped considerably after 9/11 as jihadist terrorists took inspiration from the events of that terrible day.

We first determined which terrorist organizations should be classified as jihadist. We included in this group Sunni extremist groups affiliated with or sympathetic to the ideology of Al Qaeda. We decided to exclude terrorist attacks by Palestinian groups, as they depend largely on factors particular to the Israeli-Palestinian conflict.

Our study draws its data from the MIPT-RAND Terrorism database (available at terrorismknowledgebase.org), which is widely considered to be the best publicly available database on terrorism incidents. RAND defines a terrorist attack as an attack on a civilian entity designed to promote fear or alarm and further a particular political agenda. In our study we only included attacks that caused at least one fatality and were attributed by RAND to a known jihadist group. In some terrorist attacks, and this is especially the case in Iraq, RAND has not been able to attribute a particular attack to a known jihadist group. Therefore our study likely understates the extent of jihadist terrorism in Iraq and around the world.
First statistical objection. Would Bergen be willing to further detail the factual basis for his claim that the “rate of attacks by jihadist groups jumped considerably after 9/11 as jihadist terrorists took inspiration from the events of that terrible day”? Only between September 12, 2001 and March 20, 2003? Couldn’t it be quite possible that would be Jihadists continued to take inspiration from 9/11 after March 2003, indeed, still do so to this day? The Jihadi triumph of 9/11 still played a central theme within the ideological (and pathological) rantings by some of the same Jihadis Bergen quotes elsewhere.

Forgive my cynicism, but I can’t help but feel there may be another factor at work in cropping the data the way Bergen does. It’s all about data selection and the specific objects for comparison, and Bergen, being an advocate for a particular position should be held to close scrutiny for his analytic assumptions. I would in fact suggest another possible reason for his decision, related to a second methodological objection, related to Bergen’s odd exclusion of Palestinian terrorist activity. This likewise explains Bergen’s careful choice of Jihadist as the definitional core of the terrorist activity he seeks to analyze, versus a more generic definition of terrorism.

This isn’t just some sop to Palestinian sympathizers, but an a priori (pre-existing) assumption that coincides with a particular geopolitical world view. Terrorist attacks by the PLO and related groups, aided and abetted by the same terror sponsors who sponsor Al Qaeda, are data inconvenient for Bergen’s argument and might cloud the conclusion. I don’t see mention of Hezbollah, but I would wonder if their terrorist activities couldn’t (and perhaps haven’t) been excluded by the same rationale.

By excluding considerations of prior eras and examples of modern terrorism, Bergen gets the result he wants, our efforts in Iraq make things worse. (Although, oddly, worse overwhelmingly limited in effect to the ongoing fighting in Iraq and Afghanistan. Funny how the Germans and Japanese actually killed more allies after we started fighting them.)

Just one objection more. If the data presented here were treated as a statistical sample, analytic results would be highly suspect because of the very small numbers involved. That, added to the known and documented strategy of media manipulation by Al Qaeda, Hezbollah and other like minded groups, in which attacks are carried out more for appearance than for effect, then forgive me if I draw one big, “So What?” conclusion.

Just for contrast, I’ve seen an analysis that suggests that when attacks against the West in the years 1998-2002 are compared to attacks in 2003-2007, you can actually calculate a dramatic decrease in Western deaths at all attributable to Jihadi terrorism (excluding Iraq and Afghanistan, as Bergen does). And that I guess should be the paramount objection to statistical cherry picking of this kind. You can come up with all kinds of justifications for why you include some and exclude other data occurrences, but when doing so yields the conclusion you seek, the analysis is dishonest, and the results unhelpful.

Granted, 9/11 is the great big anomaly in the data. It’s the great big motivator of Jihadis as well. The War in Iraq, much like the justification some of our enemies use from time to time about the plight of the Palestinians – a plight they decidedly ignore when given opportunity to offer tangible assistance – is just another brickbat to throw at the Americans.

I’ve heard a much more plausible explanation for why Jihadis are so energized against us. Not that our efforts in Iraq particularly enrage radical Muslims, but rather that apocalyptic calls for war against the US, the “Christian” West, and the “Zionist Enemies” get such incredible amounts of airtime and media play.

Linked by Mudville's Dawn Patrol.

Labels: ,

Links to this post:

Create a Link

<< Home

This page is powered by Blogger. Isn't yours?

Subscribe to Posts [Atom]