Thursday, October 12

Manipulating The Numbers


The media seems to have embraced 655,000 as the number of deaths since Lancet published a Johns Hopkins University study of mortality in Iraq. They've settled on this number despite the fact that the researchers themselves, reflecting the inherent uncertainties in such extrapolations, said they were 95 percent certain that the real number lay somewhere between 392,979 and 942,636 deaths.

Usually insightful Daniel Davis, Guardian Unlimited, and Tim Lambert, an Australian science blogger, have also weighed in on the matter, calling any attempts to refute the report devious hack-work, especially because the administration seems content with a number far lower, about 50,000. Davis and Lambert analyze the data using their own brand of statistical posturing based on survey samplings before Davis goes on to say that “there has to be some accountability here.”

I agree. There does have to be some accountability here. And while I'll stop short of saying the researchers lied or are frauds, I will point out that their excessively broad range (1/2 million +/- 5%) speaks volumes: they have no idea. In fact, I am equally or perhaps even more accurate in saying that there were between one and 1 million deaths.

In covering the original study, the media seems to have settled on the middle ground, coming up with the 600,000 to 655,000 range. While I have no idea whether or not the number is accurate (the method, considering it's from Johns Hopkins, seems less credible than usual), I do know that some members of the media have become more sloppy at accepting statical reports as newsworthy because they seem credible (no matter what the method) and always create a buzz of controversy.

That is how this topic ties into communication. All communicators, or editors, will be tempted from time to time to publish a statistical report that will generate a buzz (they always do), but they should consider that 'buzz' publishing is getting away from the intent of reporting, which is, simply put, to get at the truth. It seems to me that publishing this one, given the method and given that people lie when taking such surveys, did little to do that.

What do I mean? Well, if you asked the same number of citizens if they had a loved one, or if someone they knew had a loved one, who died in 9/11, and applied the same statiscial theory that Davis applied in his post to defend the study, then I'd wager the death toll would exceed 1 million. Thank goodness it did not.

3 comments:

Rich on 10/12/06, 5:32 PM said...

Famous Last Words:

"It may not be extremely precise, but at least it gets us in the right ballpark." - Les Roberts, co-author of the report

John H on 10/13/06, 1:36 AM said...

Well, as I said in my own post on this topic, the 650,000 figure is a mistake precisely because it presents an easy target that can be attacked while ignoring the real qualitative details in the report.

The real picture can be seen when one looks at the difference in death-rates experienced within the sample (of 12,000 people in 1,849 households in 47 locations), where the number of violent deaths increased from two in the 18 months prior to the invasion, to 300 in the 39 months since the invasion - a 70-fold increase in the rate of violent deaths.

Even setting aside the question of whether this can be safely extrapolated to the entire population, Daniel Davies is surely correct to say that this figure is "not even nearly consistent" with a 60,000 figure for total deaths nationwide, and that it is clear that, since the invasion, "things have got worse, and they have got a lot worse, not a little bit worse".

Also, your point about people "lying" in surveys ignores the fact that 92% of the reported deaths were confirmed by death certificates.

Rich on 10/13/06, 8:22 AM said...

Thank you for the comment, John. Always appreciated.

From a communication aspect, my point is that the report did not present as well as an argument as it was intended to or even could have. And many media outlets did a poor job reporting on it as they sometimes do when it comes to statistical information, which is a growing trend that communication professionals should be aware of.

In this instance, the quantitate results are used to attract attention, but then the readers are asked to ignore them in favor of the qualitative assessments. When the qualitative assessments are questioned, the supporters of the report go back to the quantitate results to defend their case, which hardly allows for an open, honest discussion of the subject.

There are many problems with the survey, as even Les Roberts concedes. I tend to agree.

One of several examples: there has been even more weight placed on the concept that 12,000 people were surveyed from 1,849 households in 47 locations. The weight is usually placed here because 12,000 seems to be a large sampling size.

Yet, if 12,000 people are in 1,849 households that means each household consists of 6.5 people. So if one household reports one death, are we counting that death 6.5 times? I'm not saying that is what happened, but these are the types of questions one needs to ask before embracing a survey.

Likewise, if 92 percent of the reported deaths were confirmed by death certificates, then how could those deaths not be counted, assuming the death certificates were issued by the same people in charge of the counting?

But, of course, this is a quantitate argument, not qualitative, which is really what people seem to want to discuss.

While I'm not convinced things are neither better nor dramatically worse, I do believe that the death rate prior to invasion/liberation (depending on what side of the fence you're on) seems deplorable in that it the governing body was often directly responsible for those deaths, execution style. The death rate today, on the other hand, may well be the result of living in a country with an active insurgency as opposed to government run executions.

In conclusion, this is precisely why I look forward to the United States implementing its exit strategy. There is no question that it is increasingly damaging for our country and its allies to be there. And that is the best qualitative conclusion of all.

Post a Comment

 

Blog Archive

Google+ Followers

by Rich Becker Copyright © 2010 Designed by Bie Blogger Template